diff --git a/CHANGELOG.md b/CHANGELOG.md index 072d2c53..b4e6d9a3 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -11,6 +11,9 @@ * `simpleSeparateCompilation` demonstrating NVCC compilation to a static library (reason: trivial) * `simpleTemplates_nvrtc` demonstrating NVRTC usage for `simpleTemplates` sample (reason: redundant) * `simpleVoteIntrinsics_nvrtc` demonstrating NVRTC usage for `simpleVoteIntrinsics` sample (reason: redundant) + * `2_Concepts_and_Techniques` + * `cuHook` demonstrating dlsym hooks. (reason: incompatible with modern `glibc`) + ### CUDA 12.5 diff --git a/Samples/0_Introduction/simpleTexture3D/findgllib.mk b/Samples/0_Introduction/simpleTexture3D/findgllib.mk deleted file mode 100644 index 998fcf0f..00000000 --- a/Samples/0_Introduction/simpleTexture3D/findgllib.mk +++ /dev/null @@ -1,136 +0,0 @@ -################################################################################ -# Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved. -# -# Redistribution and use in source and binary forms, with or without -# modification, are permitted provided that the following conditions -# are met: -# * Redistributions of source code must retain the above copyright -# notice, this list of conditions and the following disclaimer. -# * Redistributions in binary form must reproduce the above copyright -# notice, this list of conditions and the following disclaimer in the -# documentation and/or other materials provided with the distribution. -# * Neither the name of NVIDIA CORPORATION nor the names of its -# contributors may be used to endorse or promote products derived -# from this software without specific prior written permission. -# -# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY -# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE -# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR -# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR -# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, -# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, -# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR -# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY -# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -# -################################################################################ -# -# findgllib.mk is used to find the necessary GL Libraries for specific distributions -# this is supported on Mac OSX and Linux Platforms -# -################################################################################ - -# Determine OS platform and unix distribution -ifeq ("$(TARGET_OS)","linux") - # first search lsb_release - DISTRO := $(shell lsb_release -i -s 2>/dev/null | tr "[:upper:]" "[:lower:]") - ifeq ("$(DISTRO)","") - # second search and parse /etc/issue - DISTRO := $(shell awk '{print $$1}' /etc/issue | tr -d "[:space:]" | sed -e "/^$$/d" | tr "[:upper:]" "[:lower:]") - # ensure data from /etc/issue is valid - ifeq (,$(filter $(DISTRO),ubuntu fedora red rhel centos suse)) - DISTRO := - endif - ifeq ("$(DISTRO)","") - # third, we can search in /etc/os-release or /etc/{distro}-release - DISTRO := $(shell awk '/ID/' /etc/*-release | sed 's/ID=//' | grep -v "VERSION" | grep -v "ID" | grep -v "DISTRIB") - endif - endif -endif - -ifeq ("$(TARGET_OS)","linux") - # $(info) >> findgllib.mk -> LINUX path <<<) - # Each set of Linux Distros have different paths for where to find their OpenGL libraries reside - UBUNTU = $(shell echo $(DISTRO) | grep -i ubuntu >/dev/null 2>&1; echo $$?) - FEDORA = $(shell echo $(DISTRO) | grep -i fedora >/dev/null 2>&1; echo $$?) - RHEL = $(shell echo $(DISTRO) | grep -i 'red\|rhel' >/dev/null 2>&1; echo $$?) - CENTOS = $(shell echo $(DISTRO) | grep -i centos >/dev/null 2>&1; echo $$?) - SUSE = $(shell echo $(DISTRO) | grep -i 'suse\|sles' >/dev/null 2>&1; echo $$?) - KYLIN = $(shell echo $(DISTRO) | grep -i kylin >/dev/null 2>&1; echo $$?) - ifeq ("$(UBUNTU)","0") - ifeq ($(HOST_ARCH)-$(TARGET_ARCH),x86_64-armv7l) - GLPATH := /usr/arm-linux-gnueabihf/lib - GLLINK := -L/usr/arm-linux-gnueabihf/lib - ifneq ($(TARGET_FS),) - GLPATH += $(TARGET_FS)/usr/lib/arm-linux-gnueabihf - GLLINK += -L$(TARGET_FS)/usr/lib/arm-linux-gnueabihf - endif - else ifeq ($(HOST_ARCH)-$(TARGET_ARCH),x86_64-aarch64) - GLPATH := /usr/aarch64-linux-gnu/lib - GLLINK := -L/usr/aarch64-linux-gnu/lib - ifneq ($(TARGET_FS),) - GLPATH += $(TARGET_FS)/usr/lib - GLPATH += $(TARGET_FS)/usr/lib/aarch64-linux-gnu - GLLINK += -L$(TARGET_FS)/usr/lib/aarch64-linux-gnu - endif - else ifeq ($(HOST_ARCH)-$(TARGET_ARCH),x86_64-ppc64le) - GLPATH := /usr/powerpc64le-linux-gnu/lib - GLLINK := -L/usr/powerpc64le-linux-gnu/lib - else - UBUNTU_PKG_NAME = $(shell which dpkg >/dev/null 2>&1 && dpkg -l 'nvidia-*' | grep '^ii' | awk '{print $$2}' | head -1) - ifneq ("$(UBUNTU_PKG_NAME)","") - GLPATH ?= /usr/lib/$(UBUNTU_PKG_NAME) - GLLINK ?= -L/usr/lib/$(UBUNTU_PKG_NAME) - endif - - DFLT_PATH ?= /usr/lib - endif - endif - - ifeq ("$(SUSE)","0") - GLPATH ?= /usr/X11R6/lib64 - GLLINK ?= -L/usr/X11R6/lib64 - DFLT_PATH ?= /usr/lib64 - else - GLPATH ?= /usr/lib64/nvidia - GLLINK ?= -L/usr/lib64/nvidia - DFLT_PATH ?= /usr/lib64 - endif - - # find libGL, libGLU - GLLIB := $(shell find -L $(GLPATH) $(DFLT_PATH) -name libGL.so -print 2>/dev/null) - GLULIB := $(shell find -L $(GLPATH) $(DFLT_PATH) -name libGLU.so -print 2>/dev/null) - - ifeq ("$(GLLIB)","") - $(info >>> WARNING - libGL.so not found, refer to CUDA Getting Started Guide for how to find and install them. <<<) - SAMPLE_ENABLED := 0 - endif - ifeq ("$(GLULIB)","") - $(info >>> WARNING - libGLU.so not found, refer to CUDA Getting Started Guide for how to find and install them. <<<) - SAMPLE_ENABLED := 0 - endif - - HEADER_SEARCH_PATH ?= $(TARGET_FS)/usr/include - ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_OS),x86_64-armv7l-linux) - HEADER_SEARCH_PATH += /usr/arm-linux-gnueabihf/include - else ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_OS),x86_64-aarch64-linux) - HEADER_SEARCH_PATH += /usr/aarch64-linux-gnu/include - endif - - GLHEADER := $(shell find -L $(HEADER_SEARCH_PATH) -name gl.h -print 2>/dev/null) - GLUHEADER := $(shell find -L $(HEADER_SEARCH_PATH) -name glu.h -print 2>/dev/null) - - ifeq ("$(GLHEADER)","") - $(info >>> WARNING - gl.h not found, refer to CUDA Getting Started Guide for how to find and install them. <<<) - SAMPLE_ENABLED := 0 - endif - ifeq ("$(GLUHEADER)","") - $(info >>> WARNING - glu.h not found, refer to CUDA Getting Started Guide for how to find and install them. <<<) - SAMPLE_ENABLED := 0 - endif -else - # This would be the Mac OS X path if we had to do anything special -endif - diff --git a/Samples/2_Concepts_and_Techniques/CMakeLists.txt b/Samples/2_Concepts_and_Techniques/CMakeLists.txt new file mode 100644 index 00000000..289a341c --- /dev/null +++ b/Samples/2_Concepts_and_Techniques/CMakeLists.txt @@ -0,0 +1,33 @@ +#add_subdirectory(EGLStream_CUDA_CrossGPU) +#add_subdirectory(EGLStream_CUDA_Interop) +#add_subdirectory(EGLSync_CUDAEvent_Interop) +#add_subdirectory(FunctionPointers) +#add_subdirectory(MC_EstimatePiInlineP) +#add_subdirectory(MC_EstimatePiInlineQ) +#add_subdirectory(MC_EstimatePiP) +#add_subdirectory(MC_EstimatePiQ) +#add_subdirectory(MC_SingleAsianOptionP) +add_subdirectory(boxFilter) +add_subdirectory(convolutionSeparable) +add_subdirectory(convolutionTexture) +add_subdirectory(dct8x8) +#add_subdirectory(eigenvalues) +#add_subdirectory(histogram) +#add_subdirectory(imageDenoising) +#add_subdirectory(inlinePTX) +#add_subdirectory(inlinePTX_nvrtc) +#add_subdirectory(interval) +#add_subdirectory(particles) +#add_subdirectory(radixSortThrust) +#add_subdirectory(reduction) +#add_subdirectory(reductionMultiBlockCG) +#add_subdirectory(scalarProd) +#add_subdirectory(scan) +#add_subdirectory(segmentationTreeThrust) +#add_subdirectory(shfl_scan) +#add_subdirectory(sortingNetworks) +#add_subdirectory(streamOrderedAllocation) +#add_subdirectory(streamOrderedAllocationIPC) +#add_subdirectory(streamOrderedAllocationP2P) +#add_subdirectory(threadFenceReduction) +#add_subdirectory(threadMigration) diff --git a/Samples/2_Concepts_and_Techniques/boxFilter/CMakeLists.txt b/Samples/2_Concepts_and_Techniques/boxFilter/CMakeLists.txt new file mode 100644 index 00000000..f30f36df --- /dev/null +++ b/Samples/2_Concepts_and_Techniques/boxFilter/CMakeLists.txt @@ -0,0 +1,35 @@ +# Include directories and libraries +include_directories(../../../Common) + +find_package(OpenGL) +find_package(GLUT) + +# Source file +set(SRC_FILES + boxFilter.cpp + boxFilter_cpu.cpp + boxFilter_kernel.cu +) + +if(${OpenGL_FOUND}) + if (${GLUT_FOUND}) + # Add target for boxFilter + add_executable(boxFilter ${SRC_FILES}) + set_target_properties(boxFilter PROPERTIES CUDA_SEPARABLE_COMPILATION ON) + + target_include_directories(boxFilter PUBLIC + ${OPENGL_INCLUDE_DIR} + ${CUDAToolkit_INCLUDE_DIRS} + ${GLUT_INCLUDE_DIRS} + ) + + target_link_libraries(boxFilter + ${OPENGL_LIBRARIES} + ${GLUT_LIBRARIES} + ) + else() + message(STATUS "GLUT not found - will not build sample 'boxFilter'") + endif() +else() + message(STATUS "OpenGL not found - will not build sample 'boxFilter'") +endif() diff --git a/Samples/2_Concepts_and_Techniques/boxFilter/Makefile b/Samples/2_Concepts_and_Techniques/boxFilter/Makefile deleted file mode 100644 index 1f4275f8..00000000 --- a/Samples/2_Concepts_and_Techniques/boxFilter/Makefile +++ /dev/null @@ -1,385 +0,0 @@ -################################################################################ -# Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved. -# -# Redistribution and use in source and binary forms, with or without -# modification, are permitted provided that the following conditions -# are met: -# * Redistributions of source code must retain the above copyright -# notice, this list of conditions and the following disclaimer. -# * Redistributions in binary form must reproduce the above copyright -# notice, this list of conditions and the following disclaimer in the -# documentation and/or other materials provided with the distribution. -# * Neither the name of NVIDIA CORPORATION nor the names of its -# contributors may be used to endorse or promote products derived -# from this software without specific prior written permission. -# -# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY -# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE -# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR -# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR -# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, -# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, -# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR -# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY -# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -# -################################################################################ -# -# Makefile project only supported on Mac OS X and Linux Platforms) -# -################################################################################ - -# Location of the CUDA Toolkit -CUDA_PATH ?= /usr/local/cuda - -############################## -# start deprecated interface # -############################## -ifeq ($(x86_64),1) - $(info WARNING - x86_64 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=x86_64 instead) - TARGET_ARCH ?= x86_64 -endif -ifeq ($(ARMv7),1) - $(info WARNING - ARMv7 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=armv7l instead) - TARGET_ARCH ?= armv7l -endif -ifeq ($(aarch64),1) - $(info WARNING - aarch64 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=aarch64 instead) - TARGET_ARCH ?= aarch64 -endif -ifeq ($(ppc64le),1) - $(info WARNING - ppc64le variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=ppc64le instead) - TARGET_ARCH ?= ppc64le -endif -ifneq ($(GCC),) - $(info WARNING - GCC variable has been deprecated) - $(info WARNING - please use HOST_COMPILER=$(GCC) instead) - HOST_COMPILER ?= $(GCC) -endif -ifneq ($(abi),) - $(error ERROR - abi variable has been removed) -endif -############################ -# end deprecated interface # -############################ - -# architecture -HOST_ARCH := $(shell uname -m) -TARGET_ARCH ?= $(HOST_ARCH) -ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 sbsa ppc64le armv7l)) - ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 sbsa ppc64le)) - TARGET_SIZE := 64 - else ifneq (,$(filter $(TARGET_ARCH),armv7l)) - TARGET_SIZE := 32 - endif - else - TARGET_SIZE := $(shell getconf LONG_BIT) - endif -else - $(error ERROR - unsupported value $(TARGET_ARCH) for TARGET_ARCH!) -endif - -# sbsa and aarch64 systems look similar. Need to differentiate them at host level for now. -ifeq ($(HOST_ARCH),aarch64) - ifeq ($(CUDA_PATH)/targets/sbsa-linux,$(shell ls -1d $(CUDA_PATH)/targets/sbsa-linux 2>/dev/null)) - HOST_ARCH := sbsa - TARGET_ARCH := sbsa - endif -endif - -ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq (,$(filter $(HOST_ARCH)-$(TARGET_ARCH),aarch64-armv7l x86_64-armv7l x86_64-aarch64 x86_64-sbsa x86_64-ppc64le)) - $(error ERROR - cross compiling from $(HOST_ARCH) to $(TARGET_ARCH) is not supported!) - endif -endif - -# When on native aarch64 system with userspace of 32-bit, change TARGET_ARCH to armv7l -ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_SIZE),aarch64-aarch64-32) - TARGET_ARCH = armv7l -endif - -# operating system -HOST_OS := $(shell uname -s 2>/dev/null | tr "[:upper:]" "[:lower:]") -TARGET_OS ?= $(HOST_OS) -ifeq (,$(filter $(TARGET_OS),linux darwin qnx android)) - $(error ERROR - unsupported value $(TARGET_OS) for TARGET_OS!) -endif - -# host compiler -ifdef HOST_COMPILER - CUSTOM_HOST_COMPILER = 1 -endif - -ifeq ($(TARGET_OS),darwin) - ifeq ($(shell expr `xcodebuild -version | grep -i xcode | awk '{print $$2}' | cut -d'.' -f1` \>= 5),1) - HOST_COMPILER ?= clang++ - endif -else ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq ($(HOST_ARCH)-$(TARGET_ARCH),x86_64-armv7l) - ifeq ($(TARGET_OS),linux) - HOST_COMPILER ?= arm-linux-gnueabihf-g++ - else ifeq ($(TARGET_OS),qnx) - ifeq ($(QNX_HOST),) - $(error ERROR - QNX_HOST must be passed to the QNX host toolchain) - endif - ifeq ($(QNX_TARGET),) - $(error ERROR - QNX_TARGET must be passed to the QNX target toolchain) - endif - export QNX_HOST - export QNX_TARGET - HOST_COMPILER ?= $(QNX_HOST)/usr/bin/arm-unknown-nto-qnx6.6.0eabi-g++ - else ifeq ($(TARGET_OS),android) - HOST_COMPILER ?= arm-linux-androideabi-g++ - endif - else ifeq ($(TARGET_ARCH),aarch64) - ifeq ($(TARGET_OS), linux) - HOST_COMPILER ?= aarch64-linux-gnu-g++ - else ifeq ($(TARGET_OS),qnx) - ifeq ($(QNX_HOST),) - $(error ERROR - QNX_HOST must be passed to the QNX host toolchain) - endif - ifeq ($(QNX_TARGET),) - $(error ERROR - QNX_TARGET must be passed to the QNX target toolchain) - endif - export QNX_HOST - export QNX_TARGET - HOST_COMPILER ?= $(QNX_HOST)/usr/bin/q++ - else ifeq ($(TARGET_OS), android) - HOST_COMPILER ?= aarch64-linux-android-clang++ - endif - else ifeq ($(TARGET_ARCH),sbsa) - HOST_COMPILER ?= aarch64-linux-gnu-g++ - else ifeq ($(TARGET_ARCH),ppc64le) - HOST_COMPILER ?= powerpc64le-linux-gnu-g++ - endif -endif -HOST_COMPILER ?= g++ -NVCC := $(CUDA_PATH)/bin/nvcc -ccbin $(HOST_COMPILER) - -# internal flags -NVCCFLAGS := -m${TARGET_SIZE} -CCFLAGS := -LDFLAGS := - -# build flags - -# Link flag for customized HOST_COMPILER with gcc realpath -GCC_PATH := $(shell which gcc) -ifeq ($(CUSTOM_HOST_COMPILER),1) - ifneq ($(filter /%,$(HOST_COMPILER)),) - ifneq ($(findstring gcc,$(HOST_COMPILER)),) - ifneq ($(GCC_PATH),$(HOST_COMPILER)) - LDFLAGS += -lstdc++ - endif - endif - endif -endif - -ifeq ($(TARGET_OS),darwin) - LDFLAGS += -rpath $(CUDA_PATH)/lib - CCFLAGS += -arch $(HOST_ARCH) -else ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_OS),x86_64-armv7l-linux) - LDFLAGS += --dynamic-linker=/lib/ld-linux-armhf.so.3 - CCFLAGS += -mfloat-abi=hard -else ifeq ($(TARGET_OS),android) - LDFLAGS += -pie - CCFLAGS += -fpie -fpic -fexceptions -endif - -ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-linux) - ifneq ($(TARGET_FS),) - GCCVERSIONLTEQ46 := $(shell expr `$(HOST_COMPILER) -dumpversion` \<= 4.6) - ifeq ($(GCCVERSIONLTEQ46),1) - CCFLAGS += --sysroot=$(TARGET_FS) - endif - LDFLAGS += --sysroot=$(TARGET_FS) - LDFLAGS += -rpath-link=$(TARGET_FS)/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib/arm-linux-gnueabihf - endif - endif - ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-linux) - ifneq ($(TARGET_FS),) - GCCVERSIONLTEQ46 := $(shell expr `$(HOST_COMPILER) -dumpversion` \<= 4.6) - ifeq ($(GCCVERSIONLTEQ46),1) - CCFLAGS += --sysroot=$(TARGET_FS) - endif - LDFLAGS += --sysroot=$(TARGET_FS) - LDFLAGS += -rpath-link=$(TARGET_FS)/lib -L$(TARGET_FS)/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/lib/aarch64-linux-gnu -L$(TARGET_FS)/lib/aarch64-linux-gnu - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib -L$(TARGET_FS)/usr/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib/aarch64-linux-gnu -L$(TARGET_FS)/usr/lib/aarch64-linux-gnu - LDFLAGS += --unresolved-symbols=ignore-in-shared-libs - CCFLAGS += -isystem=$(TARGET_FS)/usr/include -I$(TARGET_FS)/usr/include -I$(TARGET_FS)/usr/include/libdrm - CCFLAGS += -isystem=$(TARGET_FS)/usr/include/aarch64-linux-gnu -I$(TARGET_FS)/usr/include/aarch64-linux-gnu - endif - endif - ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx) - NVCCFLAGS += -D_QNX_SOURCE - NVCCFLAGS += --qpp-config 8.3.0,gcc_ntoaarch64le - CCFLAGS += -DWIN_INTERFACE_CUSTOM -I/usr/include/aarch64-qnx-gnu - LDFLAGS += -lsocket - LDFLAGS += -L/usr/lib/aarch64-qnx-gnu - CCFLAGS += "-Wl\,-rpath-link\,/usr/lib/aarch64-qnx-gnu" - ifdef TARGET_OVERRIDE - LDFLAGS += -lslog2 - endif - - ifneq ($(TARGET_FS),) - LDFLAGS += -L$(TARGET_FS)/usr/lib - CCFLAGS += "-Wl\,-rpath-link\,$(TARGET_FS)/usr/lib" - LDFLAGS += -L$(TARGET_FS)/usr/libnvidia - CCFLAGS += "-Wl\,-rpath-link\,$(TARGET_FS)/usr/libnvidia" - CCFLAGS += -I$(TARGET_FS)/../include - endif - endif -endif - -ifdef TARGET_OVERRIDE # cuda toolkit targets override - NVCCFLAGS += -target-dir $(TARGET_OVERRIDE) -endif - -# Install directory of different arch -CUDA_INSTALL_TARGET_DIR := -ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-linux) - CUDA_INSTALL_TARGET_DIR = targets/armv7-linux-gnueabihf/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-linux) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-linux/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),sbsa-linux) - CUDA_INSTALL_TARGET_DIR = targets/sbsa-linux/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-android) - CUDA_INSTALL_TARGET_DIR = targets/armv7-linux-androideabi/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-android) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-linux-androideabi/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-qnx) - CUDA_INSTALL_TARGET_DIR = targets/ARMv7-linux-QNX/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-qnx/ -else ifeq ($(TARGET_ARCH),ppc64le) - CUDA_INSTALL_TARGET_DIR = targets/ppc64le-linux/ -endif - -# Debug build flags -ifeq ($(dbg),1) - NVCCFLAGS += -g -G - BUILD_TYPE := debug -else - BUILD_TYPE := release -endif - -ALL_CCFLAGS := -ALL_CCFLAGS += $(NVCCFLAGS) -ALL_CCFLAGS += $(EXTRA_NVCCFLAGS) -ALL_CCFLAGS += $(addprefix -Xcompiler ,$(CCFLAGS)) -ALL_CCFLAGS += $(addprefix -Xcompiler ,$(EXTRA_CCFLAGS)) - -SAMPLE_ENABLED := 1 - -# This sample is not supported on QNX -ifeq ($(TARGET_OS),qnx) - $(info >>> WARNING - boxFilter is not supported on QNX - waiving sample <<<) - SAMPLE_ENABLED := 0 -endif - -ALL_LDFLAGS := -ALL_LDFLAGS += $(ALL_CCFLAGS) -ALL_LDFLAGS += $(addprefix -Xlinker ,$(LDFLAGS)) -ALL_LDFLAGS += $(addprefix -Xlinker ,$(EXTRA_LDFLAGS)) - -# Common includes and paths for CUDA -INCLUDES := -I../../../Common -LIBRARIES := - -################################################################################ - -# Makefile include to help find GL Libraries -include ./findgllib.mk - -# OpenGL specific libraries -ifeq ($(TARGET_OS),darwin) - # Mac OSX specific libraries and paths to include - LIBRARIES += -L/System/Library/Frameworks/OpenGL.framework/Libraries - LIBRARIES += -lGL -lGLU - ALL_LDFLAGS += -Xlinker -framework -Xlinker GLUT -else - LIBRARIES += $(GLLINK) - LIBRARIES += -lGL -lGLU -lglut -endif - -# Gencode arguments -ifeq ($(TARGET_ARCH),$(filter $(TARGET_ARCH),armv7l aarch64 sbsa)) -SMS ?= 53 61 70 72 75 80 86 87 90 -else -SMS ?= 50 52 60 61 70 75 80 86 89 90 -endif - -ifeq ($(SMS),) -$(info >>> WARNING - no SM architectures have been specified - waiving sample <<<) -SAMPLE_ENABLED := 0 -endif - -ifeq ($(GENCODE_FLAGS),) -# Generate SASS code for each SM architecture listed in $(SMS) -$(foreach sm,$(SMS),$(eval GENCODE_FLAGS += -gencode arch=compute_$(sm),code=sm_$(sm))) - -# Generate PTX code from the highest SM architecture in $(SMS) to guarantee forward-compatibility -HIGHEST_SM := $(lastword $(sort $(SMS))) -ifneq ($(HIGHEST_SM),) -GENCODE_FLAGS += -gencode arch=compute_$(HIGHEST_SM),code=compute_$(HIGHEST_SM) -endif -endif - -ALL_CCFLAGS += --threads 0 --std=c++11 - -ifeq ($(SAMPLE_ENABLED),0) -EXEC ?= @echo "[@]" -endif - -################################################################################ - -# Target rules -all: build - -build: boxFilter - -check.deps: -ifeq ($(SAMPLE_ENABLED),0) - @echo "Sample will be waived due to the above missing dependencies" -else - @echo "Sample is ready - all dependencies have been met" -endif - -boxFilter.o:boxFilter.cpp - $(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $< - -boxFilter_cpu.o:boxFilter_cpu.cpp - $(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $< - -boxFilter_kernel.o:boxFilter_kernel.cu - $(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $< - -boxFilter: boxFilter.o boxFilter_cpu.o boxFilter_kernel.o - $(EXEC) $(NVCC) $(ALL_LDFLAGS) $(GENCODE_FLAGS) -o $@ $+ $(LIBRARIES) - $(EXEC) mkdir -p ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE) - $(EXEC) cp $@ ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE) - -run: build - $(EXEC) ./boxFilter - -testrun: build - $(EXEC) ./boxFilter -radius=14 -file=ref_14.ppm - $(EXEC) ./boxFilter -radius=22 -file=ref_22.ppm - -clean: - rm -f boxFilter boxFilter.o boxFilter_cpu.o boxFilter_kernel.o - rm -rf ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE)/boxFilter - -clobber: clean diff --git a/Samples/2_Concepts_and_Techniques/boxFilter/NsightEclipse.xml b/Samples/2_Concepts_and_Techniques/boxFilter/NsightEclipse.xml deleted file mode 100644 index aa65918c..00000000 --- a/Samples/2_Concepts_and_Techniques/boxFilter/NsightEclipse.xml +++ /dev/null @@ -1,111 +0,0 @@ - - - - boxFilter - - cudaGraphicsUnmapResources - cudaCreateChannelDesc - cudaMallocArray - cudaFreeArray - cudaFree - cudaGetErrorString - cudaMemcpy - cudaGraphicsResourceGetMappedPointer - cudaGraphicsMapResources - cudaDestroyTextureObject - cudaDeviceSynchronize - cudaCreateTextureObject - cudaGraphicsUnregisterResource - cudaMalloc - cudaGraphicsGLRegisterBuffer - - - whole - - data\teapot1024.ppm - data\ref_14.ppm - data\ref_22.ppm - - - ./ - ../ - ../../../Common - - - Graphics Interop - Image Processing - - - GPGPU - CUDA - Image Processing - box filter - opengl - openGL - - - GLU - GL - GLUT - GLEW - glut - X11 - - - ../../../common/lib/linux/x86_64 - ../../../common/lib/linux/armv7l - ../../../common/lib/darwin - - true - boxFilter.cpp - - -radius=14 -file=ref_14.ppm - -radius=22 -file=ref_22.ppm - - - X11 - GL - - - 2:Graphics Interop - 2:Image Processing - 2:Computer Vision - - sm50 - sm52 - sm53 - sm60 - sm61 - sm70 - sm72 - sm75 - sm80 - sm86 - sm87 - sm89 - sm90 - - - x86_64 - linux - - - windows7 - - - x86_64 - macosx - - - arm - - - sbsa - - - - all - - Box Filter - exe - diff --git a/Samples/2_Concepts_and_Techniques/boxFilter/findgllib.mk b/Samples/2_Concepts_and_Techniques/boxFilter/findgllib.mk deleted file mode 100644 index 998fcf0f..00000000 --- a/Samples/2_Concepts_and_Techniques/boxFilter/findgllib.mk +++ /dev/null @@ -1,136 +0,0 @@ -################################################################################ -# Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved. -# -# Redistribution and use in source and binary forms, with or without -# modification, are permitted provided that the following conditions -# are met: -# * Redistributions of source code must retain the above copyright -# notice, this list of conditions and the following disclaimer. -# * Redistributions in binary form must reproduce the above copyright -# notice, this list of conditions and the following disclaimer in the -# documentation and/or other materials provided with the distribution. -# * Neither the name of NVIDIA CORPORATION nor the names of its -# contributors may be used to endorse or promote products derived -# from this software without specific prior written permission. -# -# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY -# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE -# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR -# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR -# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, -# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, -# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR -# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY -# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -# -################################################################################ -# -# findgllib.mk is used to find the necessary GL Libraries for specific distributions -# this is supported on Mac OSX and Linux Platforms -# -################################################################################ - -# Determine OS platform and unix distribution -ifeq ("$(TARGET_OS)","linux") - # first search lsb_release - DISTRO := $(shell lsb_release -i -s 2>/dev/null | tr "[:upper:]" "[:lower:]") - ifeq ("$(DISTRO)","") - # second search and parse /etc/issue - DISTRO := $(shell awk '{print $$1}' /etc/issue | tr -d "[:space:]" | sed -e "/^$$/d" | tr "[:upper:]" "[:lower:]") - # ensure data from /etc/issue is valid - ifeq (,$(filter $(DISTRO),ubuntu fedora red rhel centos suse)) - DISTRO := - endif - ifeq ("$(DISTRO)","") - # third, we can search in /etc/os-release or /etc/{distro}-release - DISTRO := $(shell awk '/ID/' /etc/*-release | sed 's/ID=//' | grep -v "VERSION" | grep -v "ID" | grep -v "DISTRIB") - endif - endif -endif - -ifeq ("$(TARGET_OS)","linux") - # $(info) >> findgllib.mk -> LINUX path <<<) - # Each set of Linux Distros have different paths for where to find their OpenGL libraries reside - UBUNTU = $(shell echo $(DISTRO) | grep -i ubuntu >/dev/null 2>&1; echo $$?) - FEDORA = $(shell echo $(DISTRO) | grep -i fedora >/dev/null 2>&1; echo $$?) - RHEL = $(shell echo $(DISTRO) | grep -i 'red\|rhel' >/dev/null 2>&1; echo $$?) - CENTOS = $(shell echo $(DISTRO) | grep -i centos >/dev/null 2>&1; echo $$?) - SUSE = $(shell echo $(DISTRO) | grep -i 'suse\|sles' >/dev/null 2>&1; echo $$?) - KYLIN = $(shell echo $(DISTRO) | grep -i kylin >/dev/null 2>&1; echo $$?) - ifeq ("$(UBUNTU)","0") - ifeq ($(HOST_ARCH)-$(TARGET_ARCH),x86_64-armv7l) - GLPATH := /usr/arm-linux-gnueabihf/lib - GLLINK := -L/usr/arm-linux-gnueabihf/lib - ifneq ($(TARGET_FS),) - GLPATH += $(TARGET_FS)/usr/lib/arm-linux-gnueabihf - GLLINK += -L$(TARGET_FS)/usr/lib/arm-linux-gnueabihf - endif - else ifeq ($(HOST_ARCH)-$(TARGET_ARCH),x86_64-aarch64) - GLPATH := /usr/aarch64-linux-gnu/lib - GLLINK := -L/usr/aarch64-linux-gnu/lib - ifneq ($(TARGET_FS),) - GLPATH += $(TARGET_FS)/usr/lib - GLPATH += $(TARGET_FS)/usr/lib/aarch64-linux-gnu - GLLINK += -L$(TARGET_FS)/usr/lib/aarch64-linux-gnu - endif - else ifeq ($(HOST_ARCH)-$(TARGET_ARCH),x86_64-ppc64le) - GLPATH := /usr/powerpc64le-linux-gnu/lib - GLLINK := -L/usr/powerpc64le-linux-gnu/lib - else - UBUNTU_PKG_NAME = $(shell which dpkg >/dev/null 2>&1 && dpkg -l 'nvidia-*' | grep '^ii' | awk '{print $$2}' | head -1) - ifneq ("$(UBUNTU_PKG_NAME)","") - GLPATH ?= /usr/lib/$(UBUNTU_PKG_NAME) - GLLINK ?= -L/usr/lib/$(UBUNTU_PKG_NAME) - endif - - DFLT_PATH ?= /usr/lib - endif - endif - - ifeq ("$(SUSE)","0") - GLPATH ?= /usr/X11R6/lib64 - GLLINK ?= -L/usr/X11R6/lib64 - DFLT_PATH ?= /usr/lib64 - else - GLPATH ?= /usr/lib64/nvidia - GLLINK ?= -L/usr/lib64/nvidia - DFLT_PATH ?= /usr/lib64 - endif - - # find libGL, libGLU - GLLIB := $(shell find -L $(GLPATH) $(DFLT_PATH) -name libGL.so -print 2>/dev/null) - GLULIB := $(shell find -L $(GLPATH) $(DFLT_PATH) -name libGLU.so -print 2>/dev/null) - - ifeq ("$(GLLIB)","") - $(info >>> WARNING - libGL.so not found, refer to CUDA Getting Started Guide for how to find and install them. <<<) - SAMPLE_ENABLED := 0 - endif - ifeq ("$(GLULIB)","") - $(info >>> WARNING - libGLU.so not found, refer to CUDA Getting Started Guide for how to find and install them. <<<) - SAMPLE_ENABLED := 0 - endif - - HEADER_SEARCH_PATH ?= $(TARGET_FS)/usr/include - ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_OS),x86_64-armv7l-linux) - HEADER_SEARCH_PATH += /usr/arm-linux-gnueabihf/include - else ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_OS),x86_64-aarch64-linux) - HEADER_SEARCH_PATH += /usr/aarch64-linux-gnu/include - endif - - GLHEADER := $(shell find -L $(HEADER_SEARCH_PATH) -name gl.h -print 2>/dev/null) - GLUHEADER := $(shell find -L $(HEADER_SEARCH_PATH) -name glu.h -print 2>/dev/null) - - ifeq ("$(GLHEADER)","") - $(info >>> WARNING - gl.h not found, refer to CUDA Getting Started Guide for how to find and install them. <<<) - SAMPLE_ENABLED := 0 - endif - ifeq ("$(GLUHEADER)","") - $(info >>> WARNING - glu.h not found, refer to CUDA Getting Started Guide for how to find and install them. <<<) - SAMPLE_ENABLED := 0 - endif -else - # This would be the Mac OS X path if we had to do anything special -endif - diff --git a/Samples/2_Concepts_and_Techniques/convolutionSeparable/CMakeLists.txt b/Samples/2_Concepts_and_Techniques/convolutionSeparable/CMakeLists.txt new file mode 100644 index 00000000..ca5b2cbf --- /dev/null +++ b/Samples/2_Concepts_and_Techniques/convolutionSeparable/CMakeLists.txt @@ -0,0 +1,17 @@ +# Include directories and libraries +include_directories(../../../Common) + +# Source file +set(SRC_FILES + convolutionSeparable.cu + convolutionSeparable_gold.cpp + main.cpp +) + +# Add target for convolutionSeparable +add_executable(convolutionSeparable ${SRC_FILES}) +set_target_properties(convolutionSeparable PROPERTIES CUDA_SEPARABLE_COMPILATION ON) + +target_include_directories(convolutionSeparable PUBLIC + ${CUDAToolkit_INCLUDE_DIRS} +) diff --git a/Samples/2_Concepts_and_Techniques/convolutionSeparable/Makefile b/Samples/2_Concepts_and_Techniques/convolutionSeparable/Makefile deleted file mode 100644 index baca6908..00000000 --- a/Samples/2_Concepts_and_Techniques/convolutionSeparable/Makefile +++ /dev/null @@ -1,363 +0,0 @@ -################################################################################ -# Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved. -# -# Redistribution and use in source and binary forms, with or without -# modification, are permitted provided that the following conditions -# are met: -# * Redistributions of source code must retain the above copyright -# notice, this list of conditions and the following disclaimer. -# * Redistributions in binary form must reproduce the above copyright -# notice, this list of conditions and the following disclaimer in the -# documentation and/or other materials provided with the distribution. -# * Neither the name of NVIDIA CORPORATION nor the names of its -# contributors may be used to endorse or promote products derived -# from this software without specific prior written permission. -# -# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY -# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE -# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR -# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR -# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, -# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, -# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR -# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY -# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -# -################################################################################ -# -# Makefile project only supported on Mac OS X and Linux Platforms) -# -################################################################################ - -# Location of the CUDA Toolkit -CUDA_PATH ?= /usr/local/cuda - -############################## -# start deprecated interface # -############################## -ifeq ($(x86_64),1) - $(info WARNING - x86_64 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=x86_64 instead) - TARGET_ARCH ?= x86_64 -endif -ifeq ($(ARMv7),1) - $(info WARNING - ARMv7 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=armv7l instead) - TARGET_ARCH ?= armv7l -endif -ifeq ($(aarch64),1) - $(info WARNING - aarch64 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=aarch64 instead) - TARGET_ARCH ?= aarch64 -endif -ifeq ($(ppc64le),1) - $(info WARNING - ppc64le variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=ppc64le instead) - TARGET_ARCH ?= ppc64le -endif -ifneq ($(GCC),) - $(info WARNING - GCC variable has been deprecated) - $(info WARNING - please use HOST_COMPILER=$(GCC) instead) - HOST_COMPILER ?= $(GCC) -endif -ifneq ($(abi),) - $(error ERROR - abi variable has been removed) -endif -############################ -# end deprecated interface # -############################ - -# architecture -HOST_ARCH := $(shell uname -m) -TARGET_ARCH ?= $(HOST_ARCH) -ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 sbsa ppc64le armv7l)) - ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 sbsa ppc64le)) - TARGET_SIZE := 64 - else ifneq (,$(filter $(TARGET_ARCH),armv7l)) - TARGET_SIZE := 32 - endif - else - TARGET_SIZE := $(shell getconf LONG_BIT) - endif -else - $(error ERROR - unsupported value $(TARGET_ARCH) for TARGET_ARCH!) -endif - -# sbsa and aarch64 systems look similar. Need to differentiate them at host level for now. -ifeq ($(HOST_ARCH),aarch64) - ifeq ($(CUDA_PATH)/targets/sbsa-linux,$(shell ls -1d $(CUDA_PATH)/targets/sbsa-linux 2>/dev/null)) - HOST_ARCH := sbsa - TARGET_ARCH := sbsa - endif -endif - -ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq (,$(filter $(HOST_ARCH)-$(TARGET_ARCH),aarch64-armv7l x86_64-armv7l x86_64-aarch64 x86_64-sbsa x86_64-ppc64le)) - $(error ERROR - cross compiling from $(HOST_ARCH) to $(TARGET_ARCH) is not supported!) - endif -endif - -# When on native aarch64 system with userspace of 32-bit, change TARGET_ARCH to armv7l -ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_SIZE),aarch64-aarch64-32) - TARGET_ARCH = armv7l -endif - -# operating system -HOST_OS := $(shell uname -s 2>/dev/null | tr "[:upper:]" "[:lower:]") -TARGET_OS ?= $(HOST_OS) -ifeq (,$(filter $(TARGET_OS),linux darwin qnx android)) - $(error ERROR - unsupported value $(TARGET_OS) for TARGET_OS!) -endif - -# host compiler -ifdef HOST_COMPILER - CUSTOM_HOST_COMPILER = 1 -endif - -ifeq ($(TARGET_OS),darwin) - ifeq ($(shell expr `xcodebuild -version | grep -i xcode | awk '{print $$2}' | cut -d'.' -f1` \>= 5),1) - HOST_COMPILER ?= clang++ - endif -else ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq ($(HOST_ARCH)-$(TARGET_ARCH),x86_64-armv7l) - ifeq ($(TARGET_OS),linux) - HOST_COMPILER ?= arm-linux-gnueabihf-g++ - else ifeq ($(TARGET_OS),qnx) - ifeq ($(QNX_HOST),) - $(error ERROR - QNX_HOST must be passed to the QNX host toolchain) - endif - ifeq ($(QNX_TARGET),) - $(error ERROR - QNX_TARGET must be passed to the QNX target toolchain) - endif - export QNX_HOST - export QNX_TARGET - HOST_COMPILER ?= $(QNX_HOST)/usr/bin/arm-unknown-nto-qnx6.6.0eabi-g++ - else ifeq ($(TARGET_OS),android) - HOST_COMPILER ?= arm-linux-androideabi-g++ - endif - else ifeq ($(TARGET_ARCH),aarch64) - ifeq ($(TARGET_OS), linux) - HOST_COMPILER ?= aarch64-linux-gnu-g++ - else ifeq ($(TARGET_OS),qnx) - ifeq ($(QNX_HOST),) - $(error ERROR - QNX_HOST must be passed to the QNX host toolchain) - endif - ifeq ($(QNX_TARGET),) - $(error ERROR - QNX_TARGET must be passed to the QNX target toolchain) - endif - export QNX_HOST - export QNX_TARGET - HOST_COMPILER ?= $(QNX_HOST)/usr/bin/q++ - else ifeq ($(TARGET_OS), android) - HOST_COMPILER ?= aarch64-linux-android-clang++ - endif - else ifeq ($(TARGET_ARCH),sbsa) - HOST_COMPILER ?= aarch64-linux-gnu-g++ - else ifeq ($(TARGET_ARCH),ppc64le) - HOST_COMPILER ?= powerpc64le-linux-gnu-g++ - endif -endif -HOST_COMPILER ?= g++ -NVCC := $(CUDA_PATH)/bin/nvcc -ccbin $(HOST_COMPILER) - -# internal flags -NVCCFLAGS := -m${TARGET_SIZE} -CCFLAGS := -LDFLAGS := - -# build flags - -# Link flag for customized HOST_COMPILER with gcc realpath -GCC_PATH := $(shell which gcc) -ifeq ($(CUSTOM_HOST_COMPILER),1) - ifneq ($(filter /%,$(HOST_COMPILER)),) - ifneq ($(findstring gcc,$(HOST_COMPILER)),) - ifneq ($(GCC_PATH),$(HOST_COMPILER)) - LDFLAGS += -lstdc++ - endif - endif - endif -endif - -ifeq ($(TARGET_OS),darwin) - LDFLAGS += -rpath $(CUDA_PATH)/lib - CCFLAGS += -arch $(HOST_ARCH) -else ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_OS),x86_64-armv7l-linux) - LDFLAGS += --dynamic-linker=/lib/ld-linux-armhf.so.3 - CCFLAGS += -mfloat-abi=hard -else ifeq ($(TARGET_OS),android) - LDFLAGS += -pie - CCFLAGS += -fpie -fpic -fexceptions -endif - -ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-linux) - ifneq ($(TARGET_FS),) - GCCVERSIONLTEQ46 := $(shell expr `$(HOST_COMPILER) -dumpversion` \<= 4.6) - ifeq ($(GCCVERSIONLTEQ46),1) - CCFLAGS += --sysroot=$(TARGET_FS) - endif - LDFLAGS += --sysroot=$(TARGET_FS) - LDFLAGS += -rpath-link=$(TARGET_FS)/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib/arm-linux-gnueabihf - endif - endif - ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-linux) - ifneq ($(TARGET_FS),) - GCCVERSIONLTEQ46 := $(shell expr `$(HOST_COMPILER) -dumpversion` \<= 4.6) - ifeq ($(GCCVERSIONLTEQ46),1) - CCFLAGS += --sysroot=$(TARGET_FS) - endif - LDFLAGS += --sysroot=$(TARGET_FS) - LDFLAGS += -rpath-link=$(TARGET_FS)/lib -L$(TARGET_FS)/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/lib/aarch64-linux-gnu -L$(TARGET_FS)/lib/aarch64-linux-gnu - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib -L$(TARGET_FS)/usr/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib/aarch64-linux-gnu -L$(TARGET_FS)/usr/lib/aarch64-linux-gnu - LDFLAGS += --unresolved-symbols=ignore-in-shared-libs - CCFLAGS += -isystem=$(TARGET_FS)/usr/include -I$(TARGET_FS)/usr/include -I$(TARGET_FS)/usr/include/libdrm - CCFLAGS += -isystem=$(TARGET_FS)/usr/include/aarch64-linux-gnu -I$(TARGET_FS)/usr/include/aarch64-linux-gnu - endif - endif - ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx) - NVCCFLAGS += -D_QNX_SOURCE - NVCCFLAGS += --qpp-config 8.3.0,gcc_ntoaarch64le - CCFLAGS += -DWIN_INTERFACE_CUSTOM -I/usr/include/aarch64-qnx-gnu - LDFLAGS += -lsocket - LDFLAGS += -L/usr/lib/aarch64-qnx-gnu - CCFLAGS += "-Wl\,-rpath-link\,/usr/lib/aarch64-qnx-gnu" - ifdef TARGET_OVERRIDE - LDFLAGS += -lslog2 - endif - - ifneq ($(TARGET_FS),) - LDFLAGS += -L$(TARGET_FS)/usr/lib - CCFLAGS += "-Wl\,-rpath-link\,$(TARGET_FS)/usr/lib" - LDFLAGS += -L$(TARGET_FS)/usr/libnvidia - CCFLAGS += "-Wl\,-rpath-link\,$(TARGET_FS)/usr/libnvidia" - CCFLAGS += -I$(TARGET_FS)/../include - endif - endif -endif - -ifdef TARGET_OVERRIDE # cuda toolkit targets override - NVCCFLAGS += -target-dir $(TARGET_OVERRIDE) -endif - -# Install directory of different arch -CUDA_INSTALL_TARGET_DIR := -ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-linux) - CUDA_INSTALL_TARGET_DIR = targets/armv7-linux-gnueabihf/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-linux) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-linux/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),sbsa-linux) - CUDA_INSTALL_TARGET_DIR = targets/sbsa-linux/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-android) - CUDA_INSTALL_TARGET_DIR = targets/armv7-linux-androideabi/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-android) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-linux-androideabi/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-qnx) - CUDA_INSTALL_TARGET_DIR = targets/ARMv7-linux-QNX/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-qnx/ -else ifeq ($(TARGET_ARCH),ppc64le) - CUDA_INSTALL_TARGET_DIR = targets/ppc64le-linux/ -endif - -# Debug build flags -ifeq ($(dbg),1) - NVCCFLAGS += -g -G - BUILD_TYPE := debug -else - BUILD_TYPE := release -endif - -ALL_CCFLAGS := -ALL_CCFLAGS += $(NVCCFLAGS) -ALL_CCFLAGS += $(EXTRA_NVCCFLAGS) -ALL_CCFLAGS += $(addprefix -Xcompiler ,$(CCFLAGS)) -ALL_CCFLAGS += $(addprefix -Xcompiler ,$(EXTRA_CCFLAGS)) - -SAMPLE_ENABLED := 1 - -ALL_LDFLAGS := -ALL_LDFLAGS += $(ALL_CCFLAGS) -ALL_LDFLAGS += $(addprefix -Xlinker ,$(LDFLAGS)) -ALL_LDFLAGS += $(addprefix -Xlinker ,$(EXTRA_LDFLAGS)) - -# Common includes and paths for CUDA -INCLUDES := -I../../../Common -LIBRARIES := - -################################################################################ - -# Gencode arguments -ifeq ($(TARGET_ARCH),$(filter $(TARGET_ARCH),armv7l aarch64 sbsa)) -SMS ?= 53 61 70 72 75 80 86 87 90 -else -SMS ?= 50 52 60 61 70 75 80 86 89 90 -endif - -ifeq ($(SMS),) -$(info >>> WARNING - no SM architectures have been specified - waiving sample <<<) -SAMPLE_ENABLED := 0 -endif - -ifeq ($(GENCODE_FLAGS),) -# Generate SASS code for each SM architecture listed in $(SMS) -$(foreach sm,$(SMS),$(eval GENCODE_FLAGS += -gencode arch=compute_$(sm),code=sm_$(sm))) - -# Generate PTX code from the highest SM architecture in $(SMS) to guarantee forward-compatibility -HIGHEST_SM := $(lastword $(sort $(SMS))) -ifneq ($(HIGHEST_SM),) -GENCODE_FLAGS += -gencode arch=compute_$(HIGHEST_SM),code=compute_$(HIGHEST_SM) -endif -endif - -ALL_CCFLAGS += --threads 0 --std=c++11 - -ifeq ($(SAMPLE_ENABLED),0) -EXEC ?= @echo "[@]" -endif - -################################################################################ - -# Target rules -all: build - -build: convolutionSeparable - -check.deps: -ifeq ($(SAMPLE_ENABLED),0) - @echo "Sample will be waived due to the above missing dependencies" -else - @echo "Sample is ready - all dependencies have been met" -endif - -convolutionSeparable.o:convolutionSeparable.cu - $(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $< - -convolutionSeparable_gold.o:convolutionSeparable_gold.cpp - $(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $< - -main.o:main.cpp - $(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $< - -convolutionSeparable: convolutionSeparable.o convolutionSeparable_gold.o main.o - $(EXEC) $(NVCC) $(ALL_LDFLAGS) $(GENCODE_FLAGS) -o $@ $+ $(LIBRARIES) - $(EXEC) mkdir -p ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE) - $(EXEC) cp $@ ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE) - -run: build - $(EXEC) ./convolutionSeparable - -testrun: build - -clean: - rm -f convolutionSeparable convolutionSeparable.o convolutionSeparable_gold.o main.o - rm -rf ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE)/convolutionSeparable - -clobber: clean diff --git a/Samples/2_Concepts_and_Techniques/convolutionSeparable/NsightEclipse.xml b/Samples/2_Concepts_and_Techniques/convolutionSeparable/NsightEclipse.xml deleted file mode 100644 index 2e706e5f..00000000 --- a/Samples/2_Concepts_and_Techniques/convolutionSeparable/NsightEclipse.xml +++ /dev/null @@ -1,85 +0,0 @@ - - - - convolutionSeparable - - cudaMemcpy - cudaFree - cudaDeviceSynchronize - cudaMemcpyToSymbol - cudaMalloc - - - whole - - ./ - ../ - ../../../Common - - - Image Processing - Data Parallel Algorithms - - - CUDA - DSP - separable convolution - finite impulse-response filter - Image Processing - - - - - - true - main.cpp - - 1:CUDA Basic Topics - 1:Data-Parallel Algorithms - 3:Linear Algebra - 2:Image Processing - 2:Computer Vision - - sm50 - sm52 - sm53 - sm60 - sm61 - sm70 - sm72 - sm75 - sm80 - sm86 - sm87 - sm89 - sm90 - - - x86_64 - linux - - - windows7 - - - x86_64 - macosx - - - arm - - - sbsa - - - ppc64le - linux - - - - all - - CUDA Separable Convolution - exe - doc\convolutionSeparable.pdf - diff --git a/Samples/2_Concepts_and_Techniques/convolutionTexture/CMakeLists.txt b/Samples/2_Concepts_and_Techniques/convolutionTexture/CMakeLists.txt new file mode 100644 index 00000000..67ec400a --- /dev/null +++ b/Samples/2_Concepts_and_Techniques/convolutionTexture/CMakeLists.txt @@ -0,0 +1,17 @@ +# Include directories and libraries +include_directories(../../../Common) + +# Source file +set(SRC_FILES + convolutionTexture.cu + convolutionTexture_gold.cpp + main.cpp +) + +# Add target for convolutionTexture +add_executable(convolutionTexture ${SRC_FILES}) +set_target_properties(convolutionTexture PROPERTIES CUDA_SEPARABLE_COMPILATION ON) + +target_include_directories(convolutionTexture PUBLIC + ${CUDAToolkit_INCLUDE_DIRS} +) diff --git a/Samples/2_Concepts_and_Techniques/convolutionTexture/Makefile b/Samples/2_Concepts_and_Techniques/convolutionTexture/Makefile deleted file mode 100644 index ad059e93..00000000 --- a/Samples/2_Concepts_and_Techniques/convolutionTexture/Makefile +++ /dev/null @@ -1,363 +0,0 @@ -################################################################################ -# Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved. -# -# Redistribution and use in source and binary forms, with or without -# modification, are permitted provided that the following conditions -# are met: -# * Redistributions of source code must retain the above copyright -# notice, this list of conditions and the following disclaimer. -# * Redistributions in binary form must reproduce the above copyright -# notice, this list of conditions and the following disclaimer in the -# documentation and/or other materials provided with the distribution. -# * Neither the name of NVIDIA CORPORATION nor the names of its -# contributors may be used to endorse or promote products derived -# from this software without specific prior written permission. -# -# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY -# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE -# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR -# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR -# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, -# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, -# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR -# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY -# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -# -################################################################################ -# -# Makefile project only supported on Mac OS X and Linux Platforms) -# -################################################################################ - -# Location of the CUDA Toolkit -CUDA_PATH ?= /usr/local/cuda - -############################## -# start deprecated interface # -############################## -ifeq ($(x86_64),1) - $(info WARNING - x86_64 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=x86_64 instead) - TARGET_ARCH ?= x86_64 -endif -ifeq ($(ARMv7),1) - $(info WARNING - ARMv7 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=armv7l instead) - TARGET_ARCH ?= armv7l -endif -ifeq ($(aarch64),1) - $(info WARNING - aarch64 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=aarch64 instead) - TARGET_ARCH ?= aarch64 -endif -ifeq ($(ppc64le),1) - $(info WARNING - ppc64le variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=ppc64le instead) - TARGET_ARCH ?= ppc64le -endif -ifneq ($(GCC),) - $(info WARNING - GCC variable has been deprecated) - $(info WARNING - please use HOST_COMPILER=$(GCC) instead) - HOST_COMPILER ?= $(GCC) -endif -ifneq ($(abi),) - $(error ERROR - abi variable has been removed) -endif -############################ -# end deprecated interface # -############################ - -# architecture -HOST_ARCH := $(shell uname -m) -TARGET_ARCH ?= $(HOST_ARCH) -ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 sbsa ppc64le armv7l)) - ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 sbsa ppc64le)) - TARGET_SIZE := 64 - else ifneq (,$(filter $(TARGET_ARCH),armv7l)) - TARGET_SIZE := 32 - endif - else - TARGET_SIZE := $(shell getconf LONG_BIT) - endif -else - $(error ERROR - unsupported value $(TARGET_ARCH) for TARGET_ARCH!) -endif - -# sbsa and aarch64 systems look similar. Need to differentiate them at host level for now. -ifeq ($(HOST_ARCH),aarch64) - ifeq ($(CUDA_PATH)/targets/sbsa-linux,$(shell ls -1d $(CUDA_PATH)/targets/sbsa-linux 2>/dev/null)) - HOST_ARCH := sbsa - TARGET_ARCH := sbsa - endif -endif - -ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq (,$(filter $(HOST_ARCH)-$(TARGET_ARCH),aarch64-armv7l x86_64-armv7l x86_64-aarch64 x86_64-sbsa x86_64-ppc64le)) - $(error ERROR - cross compiling from $(HOST_ARCH) to $(TARGET_ARCH) is not supported!) - endif -endif - -# When on native aarch64 system with userspace of 32-bit, change TARGET_ARCH to armv7l -ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_SIZE),aarch64-aarch64-32) - TARGET_ARCH = armv7l -endif - -# operating system -HOST_OS := $(shell uname -s 2>/dev/null | tr "[:upper:]" "[:lower:]") -TARGET_OS ?= $(HOST_OS) -ifeq (,$(filter $(TARGET_OS),linux darwin qnx android)) - $(error ERROR - unsupported value $(TARGET_OS) for TARGET_OS!) -endif - -# host compiler -ifdef HOST_COMPILER - CUSTOM_HOST_COMPILER = 1 -endif - -ifeq ($(TARGET_OS),darwin) - ifeq ($(shell expr `xcodebuild -version | grep -i xcode | awk '{print $$2}' | cut -d'.' -f1` \>= 5),1) - HOST_COMPILER ?= clang++ - endif -else ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq ($(HOST_ARCH)-$(TARGET_ARCH),x86_64-armv7l) - ifeq ($(TARGET_OS),linux) - HOST_COMPILER ?= arm-linux-gnueabihf-g++ - else ifeq ($(TARGET_OS),qnx) - ifeq ($(QNX_HOST),) - $(error ERROR - QNX_HOST must be passed to the QNX host toolchain) - endif - ifeq ($(QNX_TARGET),) - $(error ERROR - QNX_TARGET must be passed to the QNX target toolchain) - endif - export QNX_HOST - export QNX_TARGET - HOST_COMPILER ?= $(QNX_HOST)/usr/bin/arm-unknown-nto-qnx6.6.0eabi-g++ - else ifeq ($(TARGET_OS),android) - HOST_COMPILER ?= arm-linux-androideabi-g++ - endif - else ifeq ($(TARGET_ARCH),aarch64) - ifeq ($(TARGET_OS), linux) - HOST_COMPILER ?= aarch64-linux-gnu-g++ - else ifeq ($(TARGET_OS),qnx) - ifeq ($(QNX_HOST),) - $(error ERROR - QNX_HOST must be passed to the QNX host toolchain) - endif - ifeq ($(QNX_TARGET),) - $(error ERROR - QNX_TARGET must be passed to the QNX target toolchain) - endif - export QNX_HOST - export QNX_TARGET - HOST_COMPILER ?= $(QNX_HOST)/usr/bin/q++ - else ifeq ($(TARGET_OS), android) - HOST_COMPILER ?= aarch64-linux-android-clang++ - endif - else ifeq ($(TARGET_ARCH),sbsa) - HOST_COMPILER ?= aarch64-linux-gnu-g++ - else ifeq ($(TARGET_ARCH),ppc64le) - HOST_COMPILER ?= powerpc64le-linux-gnu-g++ - endif -endif -HOST_COMPILER ?= g++ -NVCC := $(CUDA_PATH)/bin/nvcc -ccbin $(HOST_COMPILER) - -# internal flags -NVCCFLAGS := -m${TARGET_SIZE} -CCFLAGS := -LDFLAGS := - -# build flags - -# Link flag for customized HOST_COMPILER with gcc realpath -GCC_PATH := $(shell which gcc) -ifeq ($(CUSTOM_HOST_COMPILER),1) - ifneq ($(filter /%,$(HOST_COMPILER)),) - ifneq ($(findstring gcc,$(HOST_COMPILER)),) - ifneq ($(GCC_PATH),$(HOST_COMPILER)) - LDFLAGS += -lstdc++ - endif - endif - endif -endif - -ifeq ($(TARGET_OS),darwin) - LDFLAGS += -rpath $(CUDA_PATH)/lib - CCFLAGS += -arch $(HOST_ARCH) -else ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_OS),x86_64-armv7l-linux) - LDFLAGS += --dynamic-linker=/lib/ld-linux-armhf.so.3 - CCFLAGS += -mfloat-abi=hard -else ifeq ($(TARGET_OS),android) - LDFLAGS += -pie - CCFLAGS += -fpie -fpic -fexceptions -endif - -ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-linux) - ifneq ($(TARGET_FS),) - GCCVERSIONLTEQ46 := $(shell expr `$(HOST_COMPILER) -dumpversion` \<= 4.6) - ifeq ($(GCCVERSIONLTEQ46),1) - CCFLAGS += --sysroot=$(TARGET_FS) - endif - LDFLAGS += --sysroot=$(TARGET_FS) - LDFLAGS += -rpath-link=$(TARGET_FS)/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib/arm-linux-gnueabihf - endif - endif - ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-linux) - ifneq ($(TARGET_FS),) - GCCVERSIONLTEQ46 := $(shell expr `$(HOST_COMPILER) -dumpversion` \<= 4.6) - ifeq ($(GCCVERSIONLTEQ46),1) - CCFLAGS += --sysroot=$(TARGET_FS) - endif - LDFLAGS += --sysroot=$(TARGET_FS) - LDFLAGS += -rpath-link=$(TARGET_FS)/lib -L$(TARGET_FS)/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/lib/aarch64-linux-gnu -L$(TARGET_FS)/lib/aarch64-linux-gnu - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib -L$(TARGET_FS)/usr/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib/aarch64-linux-gnu -L$(TARGET_FS)/usr/lib/aarch64-linux-gnu - LDFLAGS += --unresolved-symbols=ignore-in-shared-libs - CCFLAGS += -isystem=$(TARGET_FS)/usr/include -I$(TARGET_FS)/usr/include -I$(TARGET_FS)/usr/include/libdrm - CCFLAGS += -isystem=$(TARGET_FS)/usr/include/aarch64-linux-gnu -I$(TARGET_FS)/usr/include/aarch64-linux-gnu - endif - endif - ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx) - NVCCFLAGS += -D_QNX_SOURCE - NVCCFLAGS += --qpp-config 8.3.0,gcc_ntoaarch64le - CCFLAGS += -DWIN_INTERFACE_CUSTOM -I/usr/include/aarch64-qnx-gnu - LDFLAGS += -lsocket - LDFLAGS += -L/usr/lib/aarch64-qnx-gnu - CCFLAGS += "-Wl\,-rpath-link\,/usr/lib/aarch64-qnx-gnu" - ifdef TARGET_OVERRIDE - LDFLAGS += -lslog2 - endif - - ifneq ($(TARGET_FS),) - LDFLAGS += -L$(TARGET_FS)/usr/lib - CCFLAGS += "-Wl\,-rpath-link\,$(TARGET_FS)/usr/lib" - LDFLAGS += -L$(TARGET_FS)/usr/libnvidia - CCFLAGS += "-Wl\,-rpath-link\,$(TARGET_FS)/usr/libnvidia" - CCFLAGS += -I$(TARGET_FS)/../include - endif - endif -endif - -ifdef TARGET_OVERRIDE # cuda toolkit targets override - NVCCFLAGS += -target-dir $(TARGET_OVERRIDE) -endif - -# Install directory of different arch -CUDA_INSTALL_TARGET_DIR := -ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-linux) - CUDA_INSTALL_TARGET_DIR = targets/armv7-linux-gnueabihf/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-linux) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-linux/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),sbsa-linux) - CUDA_INSTALL_TARGET_DIR = targets/sbsa-linux/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-android) - CUDA_INSTALL_TARGET_DIR = targets/armv7-linux-androideabi/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-android) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-linux-androideabi/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-qnx) - CUDA_INSTALL_TARGET_DIR = targets/ARMv7-linux-QNX/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-qnx/ -else ifeq ($(TARGET_ARCH),ppc64le) - CUDA_INSTALL_TARGET_DIR = targets/ppc64le-linux/ -endif - -# Debug build flags -ifeq ($(dbg),1) - NVCCFLAGS += -g -G - BUILD_TYPE := debug -else - BUILD_TYPE := release -endif - -ALL_CCFLAGS := -ALL_CCFLAGS += $(NVCCFLAGS) -ALL_CCFLAGS += $(EXTRA_NVCCFLAGS) -ALL_CCFLAGS += $(addprefix -Xcompiler ,$(CCFLAGS)) -ALL_CCFLAGS += $(addprefix -Xcompiler ,$(EXTRA_CCFLAGS)) - -SAMPLE_ENABLED := 1 - -ALL_LDFLAGS := -ALL_LDFLAGS += $(ALL_CCFLAGS) -ALL_LDFLAGS += $(addprefix -Xlinker ,$(LDFLAGS)) -ALL_LDFLAGS += $(addprefix -Xlinker ,$(EXTRA_LDFLAGS)) - -# Common includes and paths for CUDA -INCLUDES := -I../../../Common -LIBRARIES := - -################################################################################ - -# Gencode arguments -ifeq ($(TARGET_ARCH),$(filter $(TARGET_ARCH),armv7l aarch64 sbsa)) -SMS ?= 53 61 70 72 75 80 86 87 90 -else -SMS ?= 50 52 60 61 70 75 80 86 89 90 -endif - -ifeq ($(SMS),) -$(info >>> WARNING - no SM architectures have been specified - waiving sample <<<) -SAMPLE_ENABLED := 0 -endif - -ifeq ($(GENCODE_FLAGS),) -# Generate SASS code for each SM architecture listed in $(SMS) -$(foreach sm,$(SMS),$(eval GENCODE_FLAGS += -gencode arch=compute_$(sm),code=sm_$(sm))) - -# Generate PTX code from the highest SM architecture in $(SMS) to guarantee forward-compatibility -HIGHEST_SM := $(lastword $(sort $(SMS))) -ifneq ($(HIGHEST_SM),) -GENCODE_FLAGS += -gencode arch=compute_$(HIGHEST_SM),code=compute_$(HIGHEST_SM) -endif -endif - -ALL_CCFLAGS += --threads 0 --std=c++11 - -ifeq ($(SAMPLE_ENABLED),0) -EXEC ?= @echo "[@]" -endif - -################################################################################ - -# Target rules -all: build - -build: convolutionTexture - -check.deps: -ifeq ($(SAMPLE_ENABLED),0) - @echo "Sample will be waived due to the above missing dependencies" -else - @echo "Sample is ready - all dependencies have been met" -endif - -convolutionTexture.o:convolutionTexture.cu - $(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $< - -convolutionTexture_gold.o:convolutionTexture_gold.cpp - $(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $< - -main.o:main.cpp - $(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $< - -convolutionTexture: convolutionTexture.o convolutionTexture_gold.o main.o - $(EXEC) $(NVCC) $(ALL_LDFLAGS) $(GENCODE_FLAGS) -o $@ $+ $(LIBRARIES) - $(EXEC) mkdir -p ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE) - $(EXEC) cp $@ ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE) - -run: build - $(EXEC) ./convolutionTexture - -testrun: build - -clean: - rm -f convolutionTexture convolutionTexture.o convolutionTexture_gold.o main.o - rm -rf ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE)/convolutionTexture - -clobber: clean diff --git a/Samples/2_Concepts_and_Techniques/convolutionTexture/NsightEclipse.xml b/Samples/2_Concepts_and_Techniques/convolutionTexture/NsightEclipse.xml deleted file mode 100644 index eb57a81e..00000000 --- a/Samples/2_Concepts_and_Techniques/convolutionTexture/NsightEclipse.xml +++ /dev/null @@ -1,90 +0,0 @@ - - - - convolutionTexture - - cudaMemcpy - cudaMallocArray - cudaFreeArray - cudaFree - cudaMemcpyToArray - cudaDeviceSynchronize - cudaCreateTextureObject - cudaMemcpyToSymbol - cudaMalloc - - - whole - - ./ - ../ - ../../../Common - - - Image Processing - Texture - Data Parallel Algorithms - - - CUDA - DSP - separable convolution - finite impulse-response filter - Image Processing - - - - - - true - main.cpp - - 1:CUDA Basic Topics - 2:Texture - 1:Data-Parallel Algorithms - 3:Linear Algebra - 2:Image Processing - 2:Computer Vision - - sm50 - sm52 - sm53 - sm60 - sm61 - sm70 - sm72 - sm75 - sm80 - sm86 - sm87 - sm89 - sm90 - - - x86_64 - linux - - - windows7 - - - x86_64 - macosx - - - arm - - - sbsa - - - ppc64le - linux - - - - all - - Texture-based Separable Convolution - exe - diff --git a/Samples/2_Concepts_and_Techniques/cuHook/.vscode/c_cpp_properties.json b/Samples/2_Concepts_and_Techniques/cuHook/.vscode/c_cpp_properties.json deleted file mode 100644 index f0066b0f..00000000 --- a/Samples/2_Concepts_and_Techniques/cuHook/.vscode/c_cpp_properties.json +++ /dev/null @@ -1,18 +0,0 @@ -{ - "configurations": [ - { - "name": "Linux", - "includePath": [ - "${workspaceFolder}/**", - "${workspaceFolder}/../../../Common" - ], - "defines": [], - "compilerPath": "/usr/local/cuda/bin/nvcc", - "cStandard": "gnu17", - "cppStandard": "gnu++14", - "intelliSenseMode": "linux-gcc-x64", - "configurationProvider": "ms-vscode.makefile-tools" - } - ], - "version": 4 -} diff --git a/Samples/2_Concepts_and_Techniques/cuHook/.vscode/extensions.json b/Samples/2_Concepts_and_Techniques/cuHook/.vscode/extensions.json deleted file mode 100644 index c7eb54dc..00000000 --- a/Samples/2_Concepts_and_Techniques/cuHook/.vscode/extensions.json +++ /dev/null @@ -1,7 +0,0 @@ -{ - "recommendations": [ - "nvidia.nsight-vscode-edition", - "ms-vscode.cpptools", - "ms-vscode.makefile-tools" - ] -} diff --git a/Samples/2_Concepts_and_Techniques/cuHook/.vscode/launch.json b/Samples/2_Concepts_and_Techniques/cuHook/.vscode/launch.json deleted file mode 100644 index 16951c9e..00000000 --- a/Samples/2_Concepts_and_Techniques/cuHook/.vscode/launch.json +++ /dev/null @@ -1,10 +0,0 @@ -{ - "configurations": [ - { - "name": "CUDA C++: Launch", - "type": "cuda-gdb", - "request": "launch", - "program": "${workspaceFolder}/cuHook" - } - ] -} diff --git a/Samples/2_Concepts_and_Techniques/cuHook/.vscode/tasks.json b/Samples/2_Concepts_and_Techniques/cuHook/.vscode/tasks.json deleted file mode 100644 index 4509aeb1..00000000 --- a/Samples/2_Concepts_and_Techniques/cuHook/.vscode/tasks.json +++ /dev/null @@ -1,15 +0,0 @@ -{ - "version": "2.0.0", - "tasks": [ - { - "label": "sample", - "type": "shell", - "command": "make dbg=1", - "problemMatcher": ["$nvcc"], - "group": { - "kind": "build", - "isDefault": true - } - } - ] -} diff --git a/Samples/2_Concepts_and_Techniques/cuHook/Makefile b/Samples/2_Concepts_and_Techniques/cuHook/Makefile deleted file mode 100644 index 9ccdd509..00000000 --- a/Samples/2_Concepts_and_Techniques/cuHook/Makefile +++ /dev/null @@ -1,482 +0,0 @@ -################################################################################ -# Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved. -# -# Redistribution and use in source and binary forms, with or without -# modification, are permitted provided that the following conditions -# are met: -# * Redistributions of source code must retain the above copyright -# notice, this list of conditions and the following disclaimer. -# * Redistributions in binary form must reproduce the above copyright -# notice, this list of conditions and the following disclaimer in the -# documentation and/or other materials provided with the distribution. -# * Neither the name of NVIDIA CORPORATION nor the names of its -# contributors may be used to endorse or promote products derived -# from this software without specific prior written permission. -# -# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY -# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE -# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR -# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR -# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, -# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, -# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR -# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY -# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -# -################################################################################ -# -# Makefile project only supported on Mac OS X and Linux Platforms) -# -################################################################################ - -# Location of the CUDA Toolkit -CUDA_PATH ?= /usr/local/cuda - -############################## -# start deprecated interface # -############################## -ifeq ($(x86_64),1) - $(info WARNING - x86_64 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=x86_64 instead) - TARGET_ARCH ?= x86_64 -endif -ifeq ($(ARMv7),1) - $(info WARNING - ARMv7 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=armv7l instead) - TARGET_ARCH ?= armv7l -endif -ifeq ($(aarch64),1) - $(info WARNING - aarch64 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=aarch64 instead) - TARGET_ARCH ?= aarch64 -endif -ifeq ($(ppc64le),1) - $(info WARNING - ppc64le variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=ppc64le instead) - TARGET_ARCH ?= ppc64le -endif -ifneq ($(GCC),) - $(info WARNING - GCC variable has been deprecated) - $(info WARNING - please use HOST_COMPILER=$(GCC) instead) - HOST_COMPILER ?= $(GCC) -endif -ifneq ($(abi),) - $(error ERROR - abi variable has been removed) -endif -############################ -# end deprecated interface # -############################ - -# architecture -HOST_ARCH := $(shell uname -m) -TARGET_ARCH ?= $(HOST_ARCH) -ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 sbsa ppc64le armv7l)) - ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 sbsa ppc64le)) - TARGET_SIZE := 64 - else ifneq (,$(filter $(TARGET_ARCH),armv7l)) - TARGET_SIZE := 32 - endif - else - TARGET_SIZE := $(shell getconf LONG_BIT) - endif -else - $(error ERROR - unsupported value $(TARGET_ARCH) for TARGET_ARCH!) -endif - -# sbsa and aarch64 systems look similar. Need to differentiate them at host level for now. -ifeq ($(HOST_ARCH),aarch64) - ifeq ($(CUDA_PATH)/targets/sbsa-linux,$(shell ls -1d $(CUDA_PATH)/targets/sbsa-linux 2>/dev/null)) - HOST_ARCH := sbsa - TARGET_ARCH := sbsa - endif -endif - -ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq (,$(filter $(HOST_ARCH)-$(TARGET_ARCH),aarch64-armv7l x86_64-armv7l x86_64-aarch64 x86_64-sbsa x86_64-ppc64le)) - $(error ERROR - cross compiling from $(HOST_ARCH) to $(TARGET_ARCH) is not supported!) - endif -endif - -# When on native aarch64 system with userspace of 32-bit, change TARGET_ARCH to armv7l -ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_SIZE),aarch64-aarch64-32) - TARGET_ARCH = armv7l -endif - -# operating system -HOST_OS := $(shell uname -s 2>/dev/null | tr "[:upper:]" "[:lower:]") -TARGET_OS ?= $(HOST_OS) -ifeq (,$(filter $(TARGET_OS),linux darwin qnx android)) - $(error ERROR - unsupported value $(TARGET_OS) for TARGET_OS!) -endif - -# host compiler -ifdef HOST_COMPILER - CUSTOM_HOST_COMPILER = 1 -endif - -ifeq ($(TARGET_OS),darwin) - ifeq ($(shell expr `xcodebuild -version | grep -i xcode | awk '{print $$2}' | cut -d'.' -f1` \>= 5),1) - HOST_COMPILER ?= clang++ - endif -else ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq ($(HOST_ARCH)-$(TARGET_ARCH),x86_64-armv7l) - ifeq ($(TARGET_OS),linux) - HOST_COMPILER ?= arm-linux-gnueabihf-g++ - else ifeq ($(TARGET_OS),qnx) - ifeq ($(QNX_HOST),) - $(error ERROR - QNX_HOST must be passed to the QNX host toolchain) - endif - ifeq ($(QNX_TARGET),) - $(error ERROR - QNX_TARGET must be passed to the QNX target toolchain) - endif - export QNX_HOST - export QNX_TARGET - HOST_COMPILER ?= $(QNX_HOST)/usr/bin/arm-unknown-nto-qnx6.6.0eabi-g++ - else ifeq ($(TARGET_OS),android) - HOST_COMPILER ?= arm-linux-androideabi-g++ - endif - else ifeq ($(TARGET_ARCH),aarch64) - ifeq ($(TARGET_OS), linux) - HOST_COMPILER ?= aarch64-linux-gnu-g++ - else ifeq ($(TARGET_OS),qnx) - ifeq ($(QNX_HOST),) - $(error ERROR - QNX_HOST must be passed to the QNX host toolchain) - endif - ifeq ($(QNX_TARGET),) - $(error ERROR - QNX_TARGET must be passed to the QNX target toolchain) - endif - export QNX_HOST - export QNX_TARGET - HOST_COMPILER ?= $(QNX_HOST)/usr/bin/q++ - else ifeq ($(TARGET_OS), android) - HOST_COMPILER ?= aarch64-linux-android-clang++ - endif - else ifeq ($(TARGET_ARCH),sbsa) - HOST_COMPILER ?= aarch64-linux-gnu-g++ - else ifeq ($(TARGET_ARCH),ppc64le) - HOST_COMPILER ?= powerpc64le-linux-gnu-g++ - endif -endif -HOST_COMPILER ?= g++ -NVCC := $(CUDA_PATH)/bin/nvcc -ccbin $(HOST_COMPILER) - -# internal flags -NVCCFLAGS := -m${TARGET_SIZE} -CCFLAGS := -LDFLAGS := - -# build flags - -# Link flag for customized HOST_COMPILER with gcc realpath -GCC_PATH := $(shell which gcc) -ifeq ($(CUSTOM_HOST_COMPILER),1) - ifneq ($(filter /%,$(HOST_COMPILER)),) - ifneq ($(findstring gcc,$(HOST_COMPILER)),) - ifneq ($(GCC_PATH),$(HOST_COMPILER)) - LDFLAGS += -lstdc++ - endif - endif - endif -endif - -ifeq ($(TARGET_OS),darwin) - LDFLAGS += -rpath $(CUDA_PATH)/lib - CCFLAGS += -arch $(HOST_ARCH) -else ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_OS),x86_64-armv7l-linux) - LDFLAGS += --dynamic-linker=/lib/ld-linux-armhf.so.3 - CCFLAGS += -mfloat-abi=hard -else ifeq ($(TARGET_OS),android) - LDFLAGS += -pie - CCFLAGS += -fpie -fpic -fexceptions -endif - -ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-linux) - ifneq ($(TARGET_FS),) - GCCVERSIONLTEQ46 := $(shell expr `$(HOST_COMPILER) -dumpversion` \<= 4.6) - ifeq ($(GCCVERSIONLTEQ46),1) - CCFLAGS += --sysroot=$(TARGET_FS) - endif - LDFLAGS += --sysroot=$(TARGET_FS) - LDFLAGS += -rpath-link=$(TARGET_FS)/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib/arm-linux-gnueabihf - endif - endif - ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-linux) - ifneq ($(TARGET_FS),) - GCCVERSIONLTEQ46 := $(shell expr `$(HOST_COMPILER) -dumpversion` \<= 4.6) - ifeq ($(GCCVERSIONLTEQ46),1) - CCFLAGS += --sysroot=$(TARGET_FS) - endif - LDFLAGS += --sysroot=$(TARGET_FS) - LDFLAGS += -rpath-link=$(TARGET_FS)/lib -L$(TARGET_FS)/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/lib/aarch64-linux-gnu -L$(TARGET_FS)/lib/aarch64-linux-gnu - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib -L$(TARGET_FS)/usr/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib/aarch64-linux-gnu -L$(TARGET_FS)/usr/lib/aarch64-linux-gnu - LDFLAGS += --unresolved-symbols=ignore-in-shared-libs - CCFLAGS += -isystem=$(TARGET_FS)/usr/include -I$(TARGET_FS)/usr/include -I$(TARGET_FS)/usr/include/libdrm - CCFLAGS += -isystem=$(TARGET_FS)/usr/include/aarch64-linux-gnu -I$(TARGET_FS)/usr/include/aarch64-linux-gnu - endif - endif - ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx) - NVCCFLAGS += -D_QNX_SOURCE - NVCCFLAGS += --qpp-config 8.3.0,gcc_ntoaarch64le - CCFLAGS += -DWIN_INTERFACE_CUSTOM -I/usr/include/aarch64-qnx-gnu - LDFLAGS += -lsocket - LDFLAGS += -L/usr/lib/aarch64-qnx-gnu - CCFLAGS += "-Wl\,-rpath-link\,/usr/lib/aarch64-qnx-gnu" - ifdef TARGET_OVERRIDE - LDFLAGS += -lslog2 - endif - - ifneq ($(TARGET_FS),) - LDFLAGS += -L$(TARGET_FS)/usr/lib - CCFLAGS += "-Wl\,-rpath-link\,$(TARGET_FS)/usr/lib" - LDFLAGS += -L$(TARGET_FS)/usr/libnvidia - CCFLAGS += "-Wl\,-rpath-link\,$(TARGET_FS)/usr/libnvidia" - CCFLAGS += -I$(TARGET_FS)/../include - endif - endif -endif - -ifdef TARGET_OVERRIDE # cuda toolkit targets override - NVCCFLAGS += -target-dir $(TARGET_OVERRIDE) -endif - -# Install directory of different arch -CUDA_INSTALL_TARGET_DIR := -ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-linux) - CUDA_INSTALL_TARGET_DIR = targets/armv7-linux-gnueabihf/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-linux) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-linux/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),sbsa-linux) - CUDA_INSTALL_TARGET_DIR = targets/sbsa-linux/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-android) - CUDA_INSTALL_TARGET_DIR = targets/armv7-linux-androideabi/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-android) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-linux-androideabi/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-qnx) - CUDA_INSTALL_TARGET_DIR = targets/ARMv7-linux-QNX/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-qnx/ -else ifeq ($(TARGET_ARCH),ppc64le) - CUDA_INSTALL_TARGET_DIR = targets/ppc64le-linux/ -endif - -# Debug build flags -ifeq ($(dbg),1) - NVCCFLAGS += -g -G - BUILD_TYPE := debug -else - BUILD_TYPE := release -endif - -ALL_CCFLAGS := -ALL_CCFLAGS += $(NVCCFLAGS) -ALL_CCFLAGS += $(EXTRA_NVCCFLAGS) -ALL_CCFLAGS += $(addprefix -Xcompiler ,$(CCFLAGS)) -ALL_CCFLAGS += $(addprefix -Xcompiler ,$(EXTRA_CCFLAGS)) - -UBUNTU = $(shell lsb_release -i -s 2>/dev/null | grep -i ubuntu) - -SAMPLE_ENABLED := 1 - -# This sample is not supported on Mac OSX -ifeq ($(TARGET_OS),darwin) - $(info >>> WARNING - cuHook is not supported on Mac OSX - waiving sample <<<) - SAMPLE_ENABLED := 0 -endif - -# This sample is not supported on ARMv7 -ifeq ($(TARGET_ARCH),armv7l) - $(info >>> WARNING - cuHook is not supported on ARMv7 - waiving sample <<<) - SAMPLE_ENABLED := 0 -endif - -# This sample is not supported on aarch64 -ifeq ($(TARGET_ARCH),aarch64) - $(info >>> WARNING - cuHook is not supported on aarch64 - waiving sample <<<) - SAMPLE_ENABLED := 0 -endif - -# This sample is not supported on sbsa -ifeq ($(TARGET_ARCH),sbsa) - $(info >>> WARNING - cuHook is not supported on sbsa - waiving sample <<<) - SAMPLE_ENABLED := 0 -endif - -ALL_LDFLAGS := -ALL_LDFLAGS += $(ALL_CCFLAGS) -ALL_LDFLAGS += $(addprefix -Xlinker ,$(LDFLAGS)) -ALL_LDFLAGS += $(addprefix -Xlinker ,$(EXTRA_LDFLAGS)) - -# Common includes and paths for CUDA -INCLUDES := -I../../../Common -LIBRARIES := - -################################################################################ - -#check glibc version is <= 2.33 -ifeq ($(TARGET_OS),linux) - empty := - space := $(empty) $(empty) - GLIBCVERSIONSTRING := $(shell ldd --version | head -1 | rev | cut -f1 -d' ' | rev) -#Create version number without "." - GLIBCVERSION := $(shell expr `echo $(GLIBCVERSIONSTRING)` | cut -f1 -d.) - GLIBCVERSION += $(shell expr `echo $(GLIBCVERSIONSTRING)` | cut -f2 -d.) -# Make sure the version number has at least 3 decimals - GLIBCVERSION += 00 -# Remove spaces from the version number - GLIBCVERSION := $(subst $(space),$(empty),$(GLIBCVERSION)) -#$(warning $(GLIBCVERSION)) - - IS_MIN_VERSION := $(shell expr `echo $(GLIBCVERSION)` \<= 23300) - - ifeq ($(IS_MIN_VERSION), 1) - $(info >>> GLIBC Version is less or equal to 2.33 <<<) - else - $(info >>> Waiving build. GLIBC > 2.33 is not supported<<<) - SAMPLE_ENABLED := 0 - endif -endif - -# Gencode arguments -ifeq ($(TARGET_ARCH),$(filter $(TARGET_ARCH),armv7l aarch64 sbsa)) -SMS ?= 53 61 70 72 75 80 86 87 90 -else -SMS ?= 50 52 60 61 70 75 80 86 89 90 -endif - -ifeq ($(SMS),) -$(info >>> WARNING - no SM architectures have been specified - waiving sample <<<) -SAMPLE_ENABLED := 0 -endif - -ifeq ($(GENCODE_FLAGS),) -# Generate SASS code for each SM architecture listed in $(SMS) -$(foreach sm,$(SMS),$(eval GENCODE_FLAGS += -gencode arch=compute_$(sm),code=sm_$(sm))) - -# Generate PTX code from the highest SM architecture in $(SMS) to guarantee forward-compatibility -HIGHEST_SM := $(lastword $(sort $(SMS))) -ifneq ($(HIGHEST_SM),) -GENCODE_FLAGS += -gencode arch=compute_$(HIGHEST_SM),code=compute_$(HIGHEST_SM) -endif -endif - -ifeq ($(TARGET_OS),darwin) - ALL_LDFLAGS += -Xcompiler -F/Library/Frameworks -Xlinker -framework -Xlinker CUDA -else - ifeq ($(TARGET_ARCH),x86_64) - CUDA_SEARCH_PATH ?= $(CUDA_PATH)/lib64/stubs - CUDA_SEARCH_PATH += $(CUDA_PATH)/lib/stubs - CUDA_SEARCH_PATH += $(CUDA_PATH)/targets/x86_64-linux/lib/stubs - endif - - ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-linux) - CUDA_SEARCH_PATH ?= $(CUDA_PATH)/targets/armv7-linux-gnueabihf/lib/stubs - endif - - ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-linux) - CUDA_SEARCH_PATH ?= $(CUDA_PATH)/targets/aarch64-linux/lib/stubs - endif - - ifeq ($(TARGET_ARCH)-$(TARGET_OS),sbsa-linux) - CUDA_SEARCH_PATH ?= $(CUDA_PATH)/targets/sbsa-linux/lib/stubs - endif - - ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-android) - CUDA_SEARCH_PATH ?= $(CUDA_PATH)/targets/armv7-linux-androideabi/lib/stubs - endif - - ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-android) - CUDA_SEARCH_PATH ?= $(CUDA_PATH)/targets/aarch64-linux-androideabi/lib/stubs - endif - - ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-qnx) - CUDA_SEARCH_PATH ?= $(CUDA_PATH)/targets/ARMv7-linux-QNX/lib/stubs - endif - - ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx) - CUDA_SEARCH_PATH ?= $(CUDA_PATH)/targets/aarch64-qnx/lib/stubs - ifdef TARGET_OVERRIDE - CUDA_SEARCH_PATH := $(CUDA_PATH)/targets/$(TARGET_OVERRIDE)/lib/stubs - endif - endif - - ifeq ($(TARGET_ARCH),ppc64le) - CUDA_SEARCH_PATH ?= $(CUDA_PATH)/targets/ppc64le-linux/lib/stubs - endif - - ifeq ($(HOST_ARCH),ppc64le) - CUDA_SEARCH_PATH += $(CUDA_PATH)/lib64/stubs - endif - - CUDALIB ?= $(shell find -L $(CUDA_SEARCH_PATH) -maxdepth 1 -name libcuda.so 2> /dev/null) - ifeq ("$(CUDALIB)","") - $(info >>> WARNING - libcuda.so not found, CUDA Driver is not installed. Please re-install the driver. <<<) - SAMPLE_ENABLED := 0 - else - CUDALIB := $(shell echo $(CUDALIB) | sed "s/ .*//" | sed "s/\/libcuda.so//" ) - LIBRARIES += -L$(CUDALIB) -lcuda - endif -endif - -ALL_CCFLAGS += --threads 0 --std=c++11 - -LIBSIZE := -ifneq ($(TARGET_OS),darwin) -ifeq ($(TARGET_SIZE),64) -LIBSIZE := 64 -endif -endif - -LIBRARIES += -L$(CUDA_PATH)/lib -L$(CUDA_PATH)/lib$(LIBSIZE) -ldl - -ifeq ($(SAMPLE_ENABLED),0) -EXEC ?= @echo "[@]" -endif - -################################################################################ - -# Target rules -all: build - -build: cuHook libcuhook.so.1 - -check.deps: -ifeq ($(SAMPLE_ENABLED),0) - @echo "Sample will be waived due to the above missing dependencies" -else - @echo "Sample is ready - all dependencies have been met" -endif - -cuHook.o:cuHook.cpp - $(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $< - -libcuhook.o:libcuhook.cpp - $(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) --compiler-options '-fPIC' $(GENCODE_FLAGS) -o $@ -c $< - -libcuhook.so.1:libcuhook.o - $(EXEC) $(NVCC) -shared $(ALL_LDFLAGS) $(GENCODE_FLAGS) -o $@ $+ $(LIBRARIES) - $(EXEC) mkdir -p ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE) - $(EXEC) cp $@ ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE) - -cuHook: cuHook.o - $(EXEC) $(NVCC) $(ALL_LDFLAGS) $(GENCODE_FLAGS) -o $@ $+ $(LIBRARIES) - $(EXEC) mkdir -p ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE) - $(EXEC) cp $@ ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE) - -run: build - $(EXEC) LD_PRELOAD=./libcuhook.so.1 ./cuHook - -testrun: build - -clean: - rm -f cuHook cuHook.o libcuhook.o libcuhook.so.1 - rm -rf ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE)/cuHook - -clobber: clean diff --git a/Samples/2_Concepts_and_Techniques/cuHook/README.md b/Samples/2_Concepts_and_Techniques/cuHook/README.md deleted file mode 100644 index 7427c465..00000000 --- a/Samples/2_Concepts_and_Techniques/cuHook/README.md +++ /dev/null @@ -1,65 +0,0 @@ -# cuHook - CUDA Interception Library - -## Description - -This sample demonstrates how to build and use an intercept library with CUDA. The library has to be loaded via LD_PRELOAD, e.g. LD_PRELOAD=/libcuhook.so.1 ./cuHook - -NOTE: Sample will be waived if the glibc version >= 2.34, as the sample was using these private glibc functions `__libc_dlsym()`, `__libc_dlopen_mode()` which are not exposed in 2.34 version. - -## Key Concepts - -Debugging - -## Supported SM Architectures - -[SM 5.0 ](https://developer.nvidia.com/cuda-gpus) [SM 5.2 ](https://developer.nvidia.com/cuda-gpus) [SM 5.3 ](https://developer.nvidia.com/cuda-gpus) [SM 6.0 ](https://developer.nvidia.com/cuda-gpus) [SM 6.1 ](https://developer.nvidia.com/cuda-gpus) [SM 7.0 ](https://developer.nvidia.com/cuda-gpus) [SM 7.2 ](https://developer.nvidia.com/cuda-gpus) [SM 7.5 ](https://developer.nvidia.com/cuda-gpus) [SM 8.0 ](https://developer.nvidia.com/cuda-gpus) [SM 8.6 ](https://developer.nvidia.com/cuda-gpus) [SM 8.7 ](https://developer.nvidia.com/cuda-gpus) [SM 8.9 ](https://developer.nvidia.com/cuda-gpus) [SM 9.0 ](https://developer.nvidia.com/cuda-gpus) - -## Supported OSes - -Linux - -## Supported CPU Architecture - -x86_64, ppc64le - -## CUDA APIs involved - -### [CUDA Driver API](http://docs.nvidia.com/cuda/cuda-driver-api/index.html) -cuHook, cuMemAlloc, cuHookInfo, cuHookRegisterCallback, cuCtxDestroy, cuMemFree, cuDeviceGetCount, cuCtxCreate, cuInit - -### [CUDA Runtime API](http://docs.nvidia.com/cuda/cuda-runtime-api/index.html) -cudaDeviceReset, cudaFree - -## Prerequisites - -Download and install the [CUDA Toolkit 12.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform. - -## Build and Run - -### Linux -The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make: -``` -$ cd -$ make -``` -The samples makefiles can take advantage of certain options: -* **TARGET_ARCH=** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le. - By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.
-`$ make TARGET_ARCH=x86_64`
`$ make TARGET_ARCH=ppc64le`
- See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details. -* **dbg=1** - build with debug symbols - ``` - $ make dbg=1 - ``` -* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`. - ``` - $ make SMS="50 60" - ``` - -* **HOST_COMPILER=** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers. -``` - $ make HOST_COMPILER=g++ -``` - -## References (for more details) - diff --git a/Samples/2_Concepts_and_Techniques/cuHook/cuHook.cpp b/Samples/2_Concepts_and_Techniques/cuHook/cuHook.cpp deleted file mode 100644 index c9ff3f9d..00000000 --- a/Samples/2_Concepts_and_Techniques/cuHook/cuHook.cpp +++ /dev/null @@ -1,123 +0,0 @@ -/* Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * * Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * * Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * * Neither the name of NVIDIA CORPORATION nor the names of its - * contributors may be used to endorse or promote products derived - * from this software without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY - * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR - * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR - * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, - * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, - * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR - * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY - * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT - * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE - * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - */ - -// This sample demonstrates how to use the CUDA hook library to receive -// callbacks - -#include -#include -#include - -#include "libcuhook.h" - -#define ASSERT_COND(x, msg) \ - do { \ - if (!(x)) { \ - fprintf(stderr, "Error: Condition (%s) failed at %s:%d\n", #x, __FILE__, \ - __LINE__); \ - fprintf(stderr, "cuHook sample failed (%s)\n", msg); \ - exit(1); \ - } \ - } while (0) - -/* -** Example of how to use the CUDA Interception Library, libcuhook.so -** The library has to be loaded via LD_PRELOAD, e.g. -*LD_PRELOAD=/libcuhook.so.1 ./cuHook -*/ - -static int allocation_cb = 0; -static int free_cb = 0; -static int destroy_ctx_cb = 0; - -CUresult device_allocation_callback(CUdeviceptr *dptr, size_t bytesize) { - fprintf(stdout, "Received memory allocation callback!\n"); - allocation_cb++; - return CUDA_SUCCESS; -} - -CUresult device_free_callback(CUdeviceptr dptr) { - fprintf(stdout, "Received memory de-allocation callback!\n"); - free_cb++; - return CUDA_SUCCESS; -} - -CUresult destroy_context_callback(CUcontext ctx) { - fprintf(stdout, "Received context destroy event!\n"); - destroy_ctx_cb++; - return CUDA_SUCCESS; -} - -int main() { - int count; - CUcontext ctx; - - count = 0; - - cuInit(0); - cuDeviceGetCount(&count); - ASSERT_COND(count > 0, "No suitable devices found"); - - // Load the cudaHookRegisterCallback symbol using the default library search - // order. If we found the symbol, then the hooking library has been loaded - fnCuHookRegisterCallback cuHook = - (fnCuHookRegisterCallback)dlsym(RTLD_DEFAULT, "cuHookRegisterCallback"); - // ASSERT_COND(cuHook, dlerror()); - if (cuHook) { - // CUDA Runtime symbols cannot be hooked but the underlying driver ones - // _can_. Example: - // - cudaFree() will trigger cuMemFree - // - cudaDeviceReset() will trigger a context change and you would need to - // intercept cuCtxGetCurrent/cuCtxSetCurrent - cuHook(CU_HOOK_MEM_ALLOC, POST_CALL_HOOK, - (void *)device_allocation_callback); - cuHook(CU_HOOK_MEM_FREE, PRE_CALL_HOOK, (void *)device_free_callback); - cuHook(CU_HOOK_CTX_DESTROY, POST_CALL_HOOK, - (void *)destroy_context_callback); - } - - cuCtxCreate(&ctx, 0, 0); - { - CUresult status; - CUdeviceptr dptr; - - status = cuMemAlloc(&dptr, 1024); - ASSERT_COND(status == CUDA_SUCCESS, "cuMemAlloc call failed"); - - status = cuMemFree(dptr); - ASSERT_COND(status == CUDA_SUCCESS, "cuMemFree call failed"); - } - cuCtxDestroy(ctx); - - ASSERT_COND(allocation_cb == 1, "Didn't receive the allocation callback"); - ASSERT_COND(free_cb == 1, "Didn't receive the free callback"); - ASSERT_COND(destroy_ctx_cb == 1, - "Didn't receive the destroy context callback"); - - fprintf(stdout, "Sample finished successfully.\n"); - return (0); -} diff --git a/Samples/2_Concepts_and_Techniques/cuHook/libcuhook.cpp b/Samples/2_Concepts_and_Techniques/cuHook/libcuhook.cpp deleted file mode 100644 index 84d022c5..00000000 --- a/Samples/2_Concepts_and_Techniques/cuHook/libcuhook.cpp +++ /dev/null @@ -1,188 +0,0 @@ -/* Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * * Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * * Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * * Neither the name of NVIDIA CORPORATION nor the names of its - * contributors may be used to endorse or promote products derived - * from this software without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY - * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR - * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR - * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, - * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, - * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR - * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY - * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT - * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE - * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - */ - -// This sample demonstrates a simple library to interpose CUDA symbols - -#define __USE_GNU -#include -#include -#include -#include -#include - -#include -#include "libcuhook.h" - -// For interposing dlsym(). See elf/dl-libc.c for the internal dlsym interface -// function For interposing dlopen(). Sell elf/dl-lib.c for the internal -// dlopen_mode interface function -extern "C" { -void* __libc_dlsym(void* map, const char* name); -} -extern "C" { -void* __libc_dlopen_mode(const char* name, int mode); -} - -// We need to give the pre-processor a chance to replace a function, such as: -// cuMemAlloc => cuMemAlloc_v2 -#define STRINGIFY(x) #x -#define CUDA_SYMBOL_STRING(x) STRINGIFY(x) - -// We need to interpose dlsym since anyone using dlopen+dlsym to get the CUDA -// driver symbols will bypass the hooking mechanism (this includes the CUDA -// runtime). Its tricky though, since if we replace the real dlsym with ours, we -// can't dlsym() the real dlsym. To get around that, call the 'private' libc -// interface called __libc_dlsym to get the real dlsym. -typedef void* (*fnDlsym)(void*, const char*); - -static void* real_dlsym(void* handle, const char* symbol) { - static fnDlsym internal_dlsym = (fnDlsym)__libc_dlsym( - __libc_dlopen_mode("libdl.so.2", RTLD_LAZY), "dlsym"); - return (*internal_dlsym)(handle, symbol); -} - -// Main structure that gets initialized at library load time -// Choose a unique name, or it can clash with other preloaded libraries. -struct cuHookInfo { - void* handle; - void* preHooks[CU_HOOK_SYMBOLS]; - void* postHooks[CU_HOOK_SYMBOLS]; - - // Debugging/Stats Info - int bDebugEnabled; - int hookedFunctionCalls[CU_HOOK_SYMBOLS]; - - cuHookInfo() { - const char* envHookDebug; - - // Check environment for CU_HOOK_DEBUG to facilitate debugging - envHookDebug = getenv("CU_HOOK_DEBUG"); - if (envHookDebug && envHookDebug[0] == '1') { - bDebugEnabled = 1; - fprintf(stderr, "* %6d >> CUDA HOOK Library loaded.\n", getpid()); - } - } - - ~cuHookInfo() { - if (bDebugEnabled) { - pid_t pid = getpid(); - // You can gather statistics, timings, etc. - fprintf(stderr, "* %6d >> CUDA HOOK Library Unloaded - Statistics:\n", - pid); - fprintf(stderr, "* %6d >> %20s ... %d\n", pid, - CUDA_SYMBOL_STRING(cuMemAlloc), - hookedFunctionCalls[CU_HOOK_MEM_ALLOC]); - fprintf(stderr, "* %6d >> %20s ... %d\n", pid, - CUDA_SYMBOL_STRING(cuMemFree), - hookedFunctionCalls[CU_HOOK_MEM_FREE]); - fprintf(stderr, "* %6d >> %20s ... %d\n", pid, - CUDA_SYMBOL_STRING(cuCtxGetCurrent), - hookedFunctionCalls[CU_HOOK_CTX_GET_CURRENT]); - fprintf(stderr, "* %6d >> %20s ... %d\n", pid, - CUDA_SYMBOL_STRING(cuCtxSetCurrent), - hookedFunctionCalls[CU_HOOK_CTX_SET_CURRENT]); - fprintf(stderr, "* %6d >> %20s ... %d\n", pid, - CUDA_SYMBOL_STRING(cuCtxDestroy), - hookedFunctionCalls[CU_HOOK_CTX_DESTROY]); - } - if (handle) { - dlclose(handle); - } - } -}; - -static struct cuHookInfo cuhl; - -// Exposed API -void cuHookRegisterCallback(HookSymbols symbol, HookTypes type, - void* callback) { - if (type == PRE_CALL_HOOK) { - cuhl.preHooks[symbol] = callback; - } else if (type == POST_CALL_HOOK) { - cuhl.postHooks[symbol] = callback; - } -} - -/* - ** Interposed Functions - */ -void* dlsym(void* handle, const char* symbol) { - // Early out if not a CUDA driver symbol - if (strncmp(symbol, "cu", 2) != 0) { - return (real_dlsym(handle, symbol)); - } - - if (strcmp(symbol, CUDA_SYMBOL_STRING(cuMemAlloc)) == 0) { - return (void*)(&cuMemAlloc); - } else if (strcmp(symbol, CUDA_SYMBOL_STRING(cuMemFree)) == 0) { - return (void*)(&cuMemFree); - } else if (strcmp(symbol, CUDA_SYMBOL_STRING(cuCtxGetCurrent)) == 0) { - return (void*)(&cuCtxGetCurrent); - } else if (strcmp(symbol, CUDA_SYMBOL_STRING(cuCtxSetCurrent)) == 0) { - return (void*)(&cuCtxSetCurrent); - } else if (strcmp(symbol, CUDA_SYMBOL_STRING(cuCtxDestroy)) == 0) { - return (void*)(&cuCtxDestroy); - } - return (real_dlsym(handle, symbol)); -} - -/* -** If the user of this library does not wish to include CUDA specific -*code/headers in the code, -** then all the parameters can be changed and/or simply casted before calling -*the callback. -*/ -#define CU_HOOK_GENERATE_INTERCEPT(hooksymbol, funcname, params, ...) \ - CUresult CUDAAPI funcname params { \ - static void* real_func = \ - (void*)real_dlsym(RTLD_NEXT, CUDA_SYMBOL_STRING(funcname)); \ - CUresult result = CUDA_SUCCESS; \ - \ - if (cuhl.bDebugEnabled) { \ - cuhl.hookedFunctionCalls[hooksymbol]++; \ - } \ - if (cuhl.preHooks[hooksymbol]) { \ - ((CUresult CUDAAPI(*) params)cuhl.preHooks[hooksymbol])(__VA_ARGS__); \ - } \ - result = ((CUresult CUDAAPI(*) params)real_func)(__VA_ARGS__); \ - if (cuhl.postHooks[hooksymbol] && result == CUDA_SUCCESS) { \ - ((CUresult CUDAAPI(*) params)cuhl.postHooks[hooksymbol])(__VA_ARGS__); \ - } \ - return (result); \ - } - -CU_HOOK_GENERATE_INTERCEPT(CU_HOOK_MEM_ALLOC, cuMemAlloc, - (CUdeviceptr * dptr, size_t bytesize), dptr, - bytesize) -CU_HOOK_GENERATE_INTERCEPT(CU_HOOK_MEM_FREE, cuMemFree, (CUdeviceptr dptr), - dptr) -CU_HOOK_GENERATE_INTERCEPT(CU_HOOK_CTX_GET_CURRENT, cuCtxGetCurrent, - (CUcontext * pctx), pctx) -CU_HOOK_GENERATE_INTERCEPT(CU_HOOK_CTX_SET_CURRENT, cuCtxSetCurrent, - (CUcontext ctx), ctx) -CU_HOOK_GENERATE_INTERCEPT(CU_HOOK_CTX_DESTROY, cuCtxDestroy, (CUcontext ctx), - ctx) diff --git a/Samples/2_Concepts_and_Techniques/cuHook/libcuhook.h b/Samples/2_Concepts_and_Techniques/cuHook/libcuhook.h deleted file mode 100644 index d773b430..00000000 --- a/Samples/2_Concepts_and_Techniques/cuHook/libcuhook.h +++ /dev/null @@ -1,63 +0,0 @@ -/* Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * * Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * * Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * * Neither the name of NVIDIA CORPORATION nor the names of its - * contributors may be used to endorse or promote products derived - * from this software without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY - * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR - * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR - * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, - * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, - * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR - * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY - * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT - * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE - * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - */ - -#ifndef _CUHOOK_H_ -#define _CUHOOK_H_ - -typedef enum HookTypesEnum { - PRE_CALL_HOOK, - POST_CALL_HOOK, - CU_HOOK_TYPES, -} HookTypes; - -typedef enum HookSymbolsEnum { - CU_HOOK_MEM_ALLOC, - CU_HOOK_MEM_FREE, - CU_HOOK_CTX_GET_CURRENT, - CU_HOOK_CTX_SET_CURRENT, - CU_HOOK_CTX_DESTROY, - CU_HOOK_SYMBOLS, -} HookSymbols; - -// One and only function to call to register a callback -// You need to dlsym this symbol in your application and call it to register -// callbacks -typedef void (*fnCuHookRegisterCallback)(HookSymbols symbol, HookTypes type, - void* callback); -extern "C" { -void cuHookRegisterCallback(HookSymbols symbol, HookTypes type, void* callback); -} - -// In case you want to intercept, the callbacks need the same type/parameters as -// the real functions -typedef CUresult CUDAAPI (*fnMemAlloc)(CUdeviceptr* dptr, size_t bytesize); -typedef CUresult CUDAAPI (*fnMemFree)(CUdeviceptr dptr); -typedef CUresult CUDAAPI (*fnCtxGetCurrent)(CUcontext* pctx); -typedef CUresult CUDAAPI (*fnCtxSetCurrent)(CUcontext ctx); -typedef CUresult CUDAAPI (*fnCtxDestroy)(CUcontext ctx); - -#endif /* _CUHOOK_H_ */ diff --git a/Samples/2_Concepts_and_Techniques/dct8x8/CMakeLists.txt b/Samples/2_Concepts_and_Techniques/dct8x8/CMakeLists.txt new file mode 100644 index 00000000..8230154c --- /dev/null +++ b/Samples/2_Concepts_and_Techniques/dct8x8/CMakeLists.txt @@ -0,0 +1,24 @@ +# Include directories and libraries +include_directories(../../../Common) + +# Source file +set(SRC_FILES + dct8x8.cu + BmpUtil.cpp + DCT8x8_Gold.cpp +) + +# Add target for dct8x8 +add_executable(dct8x8 ${SRC_FILES}) +set_target_properties(dct8x8 PROPERTIES CUDA_SEPARABLE_COMPILATION ON) + +target_include_directories(dct8x8 PUBLIC + ${CUDAToolkit_INCLUDE_DIRS} +) + +file(GLOB REF_DATA "teapot512*") +add_custom_command(TARGET dct8x8 POST_BUILD + COMMAND ${CMAKE_COMMAND} -E copy_if_different + ${REF_DATA} + ${CMAKE_CURRENT_BINARY_DIR} +) diff --git a/Samples/2_Concepts_and_Techniques/dct8x8/DCT8x8_Gold.cpp b/Samples/2_Concepts_and_Techniques/dct8x8/DCT8x8_Gold.cpp index 4af956fd..9566eae3 100644 --- a/Samples/2_Concepts_and_Techniques/dct8x8/DCT8x8_Gold.cpp +++ b/Samples/2_Concepts_and_Techniques/dct8x8/DCT8x8_Gold.cpp @@ -42,46 +42,43 @@ #include "BmpUtil.h" /** -* This unitary matrix performs DCT of rows of the matrix to the left -*/ + * This unitary matrix performs DCT of rows of the matrix to the left + */ const float DCTv8matrix[BLOCK_SIZE2] = { - 0.3535533905932738f, 0.4903926402016152f, 0.4619397662556434f, 0.4157348061512726f, 0.3535533905932738f, 0.2777851165098011f, 0.1913417161825449f, 0.0975451610080642f, - 0.3535533905932738f, 0.4157348061512726f, 0.1913417161825449f, -0.0975451610080641f, -0.3535533905932737f, -0.4903926402016152f, -0.4619397662556434f, -0.2777851165098011f, - 0.3535533905932738f, 0.2777851165098011f, -0.1913417161825449f, -0.4903926402016152f, -0.3535533905932738f, 0.0975451610080642f, 0.4619397662556433f, 0.4157348061512727f, - 0.3535533905932738f, 0.0975451610080642f, -0.4619397662556434f, -0.2777851165098011f, 0.3535533905932737f, 0.4157348061512727f, -0.1913417161825450f, -0.4903926402016153f, - 0.3535533905932738f, -0.0975451610080641f, -0.4619397662556434f, 0.2777851165098009f, 0.3535533905932738f, -0.4157348061512726f, -0.1913417161825453f, 0.4903926402016152f, - 0.3535533905932738f, -0.2777851165098010f, -0.1913417161825452f, 0.4903926402016153f, -0.3535533905932733f, -0.0975451610080649f, 0.4619397662556437f, -0.4157348061512720f, - 0.3535533905932738f, -0.4157348061512727f, 0.1913417161825450f, 0.0975451610080640f, -0.3535533905932736f, 0.4903926402016152f, -0.4619397662556435f, 0.2777851165098022f, - 0.3535533905932738f, -0.4903926402016152f, 0.4619397662556433f, -0.4157348061512721f, 0.3535533905932733f, -0.2777851165098008f, 0.1913417161825431f, -0.0975451610080625f -}; + 0.3535533905932738f, 0.4903926402016152f, 0.4619397662556434f, 0.4157348061512726f, 0.3535533905932738f, 0.2777851165098011f, 0.1913417161825449f, 0.0975451610080642f, + 0.3535533905932738f, 0.4157348061512726f, 0.1913417161825449f, -0.0975451610080641f, -0.3535533905932737f, -0.4903926402016152f, -0.4619397662556434f, -0.2777851165098011f, + 0.3535533905932738f, 0.2777851165098011f, -0.1913417161825449f, -0.4903926402016152f, -0.3535533905932738f, 0.0975451610080642f, 0.4619397662556433f, 0.4157348061512727f, + 0.3535533905932738f, 0.0975451610080642f, -0.4619397662556434f, -0.2777851165098011f, 0.3535533905932737f, 0.4157348061512727f, -0.1913417161825450f, -0.4903926402016153f, + 0.3535533905932738f, -0.0975451610080641f, -0.4619397662556434f, 0.2777851165098009f, 0.3535533905932738f, -0.4157348061512726f, -0.1913417161825453f, 0.4903926402016152f, + 0.3535533905932738f, -0.2777851165098010f, -0.1913417161825452f, 0.4903926402016153f, -0.3535533905932733f, -0.0975451610080649f, 0.4619397662556437f, -0.4157348061512720f, + 0.3535533905932738f, -0.4157348061512727f, 0.1913417161825450f, 0.0975451610080640f, -0.3535533905932736f, 0.4903926402016152f, -0.4619397662556435f, 0.2777851165098022f, + 0.3535533905932738f, -0.4903926402016152f, 0.4619397662556433f, -0.4157348061512721f, 0.3535533905932733f, -0.2777851165098008f, 0.1913417161825431f, -0.0975451610080625f}; /** -* This unitary matrix performs DCT of columns of the matrix to the right -*/ + * This unitary matrix performs DCT of columns of the matrix to the right + */ const float DCTv8matrixT[BLOCK_SIZE2] = { - 0.3535533905932738f, 0.3535533905932738f, 0.3535533905932738f, 0.3535533905932738f, 0.3535533905932738f, 0.3535533905932738f, 0.3535533905932738f, 0.3535533905932738f, - 0.4903926402016152f, 0.4157348061512726f, 0.2777851165098011f, 0.0975451610080642f, -0.0975451610080641f, -0.2777851165098010f, -0.4157348061512727f, -0.4903926402016152f, - 0.4619397662556434f, 0.1913417161825449f, -0.1913417161825449f, -0.4619397662556434f, -0.4619397662556434f, -0.1913417161825452f, 0.1913417161825450f, 0.4619397662556433f, - 0.4157348061512726f, -0.0975451610080641f, -0.4903926402016152f, -0.2777851165098011f, 0.2777851165098009f, 0.4903926402016153f, 0.0975451610080640f, -0.4157348061512721f, - 0.3535533905932738f, -0.3535533905932737f, -0.3535533905932738f, 0.3535533905932737f, 0.3535533905932738f, -0.3535533905932733f, -0.3535533905932736f, 0.3535533905932733f, - 0.2777851165098011f, -0.4903926402016152f, 0.0975451610080642f, 0.4157348061512727f, -0.4157348061512726f, -0.0975451610080649f, 0.4903926402016152f, -0.2777851165098008f, - 0.1913417161825449f, -0.4619397662556434f, 0.4619397662556433f, -0.1913417161825450f, -0.1913417161825453f, 0.4619397662556437f, -0.4619397662556435f, 0.1913417161825431f, - 0.0975451610080642f, -0.2777851165098011f, 0.4157348061512727f, -0.4903926402016153f, 0.4903926402016152f, -0.4157348061512720f, 0.2777851165098022f, -0.0975451610080625f -}; + 0.3535533905932738f, 0.3535533905932738f, 0.3535533905932738f, 0.3535533905932738f, 0.3535533905932738f, 0.3535533905932738f, 0.3535533905932738f, 0.3535533905932738f, + 0.4903926402016152f, 0.4157348061512726f, 0.2777851165098011f, 0.0975451610080642f, -0.0975451610080641f, -0.2777851165098010f, -0.4157348061512727f, -0.4903926402016152f, + 0.4619397662556434f, 0.1913417161825449f, -0.1913417161825449f, -0.4619397662556434f, -0.4619397662556434f, -0.1913417161825452f, 0.1913417161825450f, 0.4619397662556433f, + 0.4157348061512726f, -0.0975451610080641f, -0.4903926402016152f, -0.2777851165098011f, 0.2777851165098009f, 0.4903926402016153f, 0.0975451610080640f, -0.4157348061512721f, + 0.3535533905932738f, -0.3535533905932737f, -0.3535533905932738f, 0.3535533905932737f, 0.3535533905932738f, -0.3535533905932733f, -0.3535533905932736f, 0.3535533905932733f, + 0.2777851165098011f, -0.4903926402016152f, 0.0975451610080642f, 0.4157348061512727f, -0.4157348061512726f, -0.0975451610080649f, 0.4903926402016152f, -0.2777851165098008f, + 0.1913417161825449f, -0.4619397662556434f, 0.4619397662556433f, -0.1913417161825450f, -0.1913417161825453f, 0.4619397662556437f, -0.4619397662556435f, 0.1913417161825431f, + 0.0975451610080642f, -0.2777851165098011f, 0.4157348061512727f, -0.4903926402016153f, 0.4903926402016152f, -0.4157348061512720f, 0.2777851165098022f, -0.0975451610080625f}; /** -* JPEG quality=0_of_12 quantization matrix -*/ -float Q[BLOCK_SIZE2] = { - 32.f, 33.f, 51.f, 81.f, 66.f, 39.f, 34.f, 17.f, - 33.f, 36.f, 48.f, 47.f, 28.f, 23.f, 12.f, 12.f, - 51.f, 48.f, 47.f, 28.f, 23.f, 12.f, 12.f, 12.f, - 81.f, 47.f, 28.f, 23.f, 12.f, 12.f, 12.f, 12.f, - 66.f, 28.f, 23.f, 12.f, 12.f, 12.f, 12.f, 12.f, - 39.f, 23.f, 12.f, 12.f, 12.f, 12.f, 12.f, 12.f, - 34.f, 12.f, 12.f, 12.f, 12.f, 12.f, 12.f, 12.f, - 17.f, 12.f, 12.f, 12.f, 12.f, 12.f, 12.f, 12.f -}; + * JPEG quality=0_of_12 quantization matrix + */ +float Q_GOLD[BLOCK_SIZE2] = { + 32.f, 33.f, 51.f, 81.f, 66.f, 39.f, 34.f, 17.f, + 33.f, 36.f, 48.f, 47.f, 28.f, 23.f, 12.f, 12.f, + 51.f, 48.f, 47.f, 28.f, 23.f, 12.f, 12.f, 12.f, + 81.f, 47.f, 28.f, 23.f, 12.f, 12.f, 12.f, 12.f, + 66.f, 28.f, 23.f, 12.f, 12.f, 12.f, 12.f, 12.f, + 39.f, 23.f, 12.f, 12.f, 12.f, 12.f, 12.f, 12.f, + 34.f, 12.f, 12.f, 12.f, 12.f, 12.f, 12.f, 12.f, + 17.f, 12.f, 12.f, 12.f, 12.f, 12.f, 12.f, 12.f}; /** ************************************************************************** @@ -97,12 +94,16 @@ float Q[BLOCK_SIZE2] = { * \return None */ void mult8x8(const float *M1, int M1Stride, const float *M2, int M2Stride, - float *Mres, int MresStride) { - for (int i = 0; i < BLOCK_SIZE; i++) { - for (int j = 0; j < BLOCK_SIZE; j++) { + float *Mres, int MresStride) +{ + for (int i = 0; i < BLOCK_SIZE; i++) + { + for (int j = 0; j < BLOCK_SIZE; j++) + { float accumul = 0; - for (int k = 0; k < BLOCK_SIZE; k++) { + for (int k = 0; k < BLOCK_SIZE; k++) + { accumul += M1[i * M1Stride + k] * M2[k * M2Stride + j]; } @@ -125,13 +126,16 @@ void mult8x8(const float *M1, int M1Stride, const float *M2, int M2Stride, * \return None */ extern "C" void computeDCT8x8Gold1(const float *fSrc, float *fDst, int Stride, - ROI Size) { + ROI Size) +{ float tmpblock[BLOCK_SIZE2]; // perform block wise DCT // DCT(A) = DCTv8matrixT * A * DCTv8matrix - for (int i = 0; i + BLOCK_SIZE - 1 < Size.height; i += BLOCK_SIZE) { - for (int j = 0; j + BLOCK_SIZE - 1 < Size.width; j += BLOCK_SIZE) { + for (int i = 0; i + BLOCK_SIZE - 1 < Size.height; i += BLOCK_SIZE) + { + for (int j = 0; j + BLOCK_SIZE - 1 < Size.width; j += BLOCK_SIZE) + { // tmpblock = DCTv8matrixT * A mult8x8(DCTv8matrixT, BLOCK_SIZE, fSrc + i * Stride + j, Stride, tmpblock, BLOCK_SIZE); @@ -156,13 +160,16 @@ extern "C" void computeDCT8x8Gold1(const float *fSrc, float *fDst, int Stride, * \return None */ extern "C" void computeIDCT8x8Gold1(const float *fSrc, float *fDst, int Stride, - ROI Size) { + ROI Size) +{ float tmpblock[BLOCK_SIZE2]; // perform block wise IDCT // IDCT(A) = DCTv8matrix * A * DCTv8matrixT - for (int i = 0; i + BLOCK_SIZE - 1 < Size.height; i += BLOCK_SIZE) { - for (int j = 0; j + BLOCK_SIZE - 1 < Size.width; j += BLOCK_SIZE) { + for (int i = 0; i + BLOCK_SIZE - 1 < Size.height; i += BLOCK_SIZE) + { + for (int j = 0; j + BLOCK_SIZE - 1 < Size.width; j += BLOCK_SIZE) + { // tmpblock = DCTv8matrix * A mult8x8(DCTv8matrix, BLOCK_SIZE, fSrc + i * Stride + j, Stride, tmpblock, BLOCK_SIZE); @@ -184,16 +191,19 @@ extern "C" void computeIDCT8x8Gold1(const float *fSrc, float *fDst, int Stride, * * \return None */ -extern "C" void quantizeGoldFloat(float *fSrcDst, int Stride, ROI Size) { - // perform block wise in-place quantization using Q - // Q(A) = round(A ./ Q) .* Q; - for (int i = 0; i < Size.height; i++) { - for (int j = 0; j < Size.width; j++) { +extern "C" void quantizeGoldFloat(float *fSrcDst, int Stride, ROI Size) +{ + // perform block wise in-place quantization using Q_GOLD + // Q_GOLD(A) = round(A ./ Q_GOLD) .* Q_GOLD; + for (int i = 0; i < Size.height; i++) + { + for (int j = 0; j < Size.width; j++) + { int qx = j % BLOCK_SIZE; int qy = i % BLOCK_SIZE; float quantized = - round_f(fSrcDst[i * Stride + j] / Q[(qy << BLOCK_SIZE_LOG2) + qx]); - fSrcDst[i * Stride + j] = quantized * Q[(qy << BLOCK_SIZE_LOG2) + qx]; + round_f(fSrcDst[i * Stride + j] / Q_GOLD[(qy << BLOCK_SIZE_LOG2) + qx]); + fSrcDst[i * Stride + j] = quantized * Q_GOLD[(qy << BLOCK_SIZE_LOG2) + qx]; } } } @@ -209,22 +219,28 @@ extern "C" void quantizeGoldFloat(float *fSrcDst, int Stride, ROI Size) { * * \return None */ -void quantizeGoldShort(short *fSrcDst, int Stride, ROI Size) { - // perform block wise in-place quantization using Q - // Q(A) = round(A ./ Q) .* Q; - for (int i = 0; i < Size.height; i++) { - for (int j = 0; j < Size.width; j++) { +void quantizeGoldShort(short *fSrcDst, int Stride, ROI Size) +{ + // perform block wise in-place quantization using Q_GOLD + // Q_GOLD(A) = round(A ./ Q_GOLD) .* Q_GOLD; + for (int i = 0; i < Size.height; i++) + { + for (int j = 0; j < Size.width; j++) + { int qx = j % BLOCK_SIZE; int qy = i % BLOCK_SIZE; short temp = fSrcDst[i * Stride + j]; - short quant = (short)(Q[(qy << BLOCK_SIZE_LOG2) + qx]); + short quant = (short)(Q_GOLD[(qy << BLOCK_SIZE_LOG2) + qx]); - if (temp < 0) { + if (temp < 0) + { temp = -temp; temp += quant >> 1; temp /= quant; temp = -temp; - } else { + } + else + { temp += quant >> 1; temp /= quant; } @@ -235,17 +251,17 @@ void quantizeGoldShort(short *fSrcDst, int Stride, ROI Size) { } // Used in forward and inverse DCT. -float C_a = 1.387039845322148f; //!< a = (2^0.5) * cos( pi / 16); -float C_b = 1.306562964876377f; //!< b = (2^0.5) * cos( pi / 8); -float C_c = 1.175875602419359f; //!< c = (2^0.5) * cos(3 * pi / 16); -float C_d = 0.785694958387102f; //!< d = (2^0.5) * cos(5 * pi / 16); -float C_e = 0.541196100146197f; //!< e = (2^0.5) * cos(3 * pi / 8); -float C_f = 0.275899379282943f; //!< f = (2^0.5) * cos(7 * pi / 16); +float C_a = 1.387039845322148f; //!< a = (2^0.5) * cos( pi / 16); +float C_b = 1.306562964876377f; //!< b = (2^0.5) * cos( pi / 8); +float C_c = 1.175875602419359f; //!< c = (2^0.5) * cos(3 * pi / 16); +float C_d = 0.785694958387102f; //!< d = (2^0.5) * cos(5 * pi / 16); +float C_e = 0.541196100146197f; //!< e = (2^0.5) * cos(3 * pi / 8); +float C_f = 0.275899379282943f; //!< f = (2^0.5) * cos(7 * pi / 16); /** -* Normalization constant that is used in forward and inverse DCT -*/ -float C_norm = 0.3535533905932737f; // 1 / (8^0.5) + * Normalization constant that is used in forward and inverse DCT + */ +float C_norm = 0.3535533905932737f; // 1 / (8^0.5) /** ************************************************************************** @@ -261,7 +277,8 @@ float C_norm = 0.3535533905932737f; // 1 / (8^0.5) * \return None */ void SubroutineDCTvector(float *FirstIn, int StepIn, float *FirstOut, - int StepOut) { + int StepOut) +{ float X07P = FirstIn[0 * StepIn] + FirstIn[7 * StepIn]; float X16P = FirstIn[1 * StepIn] + FirstIn[6 * StepIn]; float X25P = FirstIn[2 * StepIn] + FirstIn[5 * StepIn]; @@ -306,7 +323,8 @@ void SubroutineDCTvector(float *FirstIn, int StepIn, float *FirstOut, * \return None */ void SubroutineIDCTvector(float *FirstIn, int StepIn, float *FirstOut, - int StepOut) { + int StepOut) +{ float Y04P = FirstIn[0 * StepIn] + FirstIn[4 * StepIn]; float Y2b6eP = C_b * FirstIn[2 * StepIn] + C_e * FirstIn[6 * StepIn]; @@ -352,17 +370,22 @@ void SubroutineIDCTvector(float *FirstIn, int StepIn, float *FirstOut, * \return None */ extern "C" void computeDCT8x8Gold2(const float *fSrc, float *fDst, int Stride, - ROI Size) { - for (int i = 0; i + BLOCK_SIZE - 1 < Size.height; i += BLOCK_SIZE) { - for (int j = 0; j + BLOCK_SIZE - 1 < Size.width; j += BLOCK_SIZE) { + ROI Size) +{ + for (int i = 0; i + BLOCK_SIZE - 1 < Size.height; i += BLOCK_SIZE) + { + for (int j = 0; j + BLOCK_SIZE - 1 < Size.width; j += BLOCK_SIZE) + { // process rows - for (int k = 0; k < BLOCK_SIZE; k++) { + for (int k = 0; k < BLOCK_SIZE; k++) + { SubroutineDCTvector((float *)fSrc + (i + k) * Stride + j, 1, fDst + (i + k) * Stride + j, 1); } // process columns - for (int k = 0; k < BLOCK_SIZE; k++) { + for (int k = 0; k < BLOCK_SIZE; k++) + { SubroutineDCTvector(fDst + i * Stride + (j + k), Stride, fDst + i * Stride + (j + k), Stride); } @@ -384,17 +407,22 @@ extern "C" void computeDCT8x8Gold2(const float *fSrc, float *fDst, int Stride, * \return None */ extern "C" void computeIDCT8x8Gold2(const float *fSrc, float *fDst, int Stride, - ROI Size) { - for (int i = 0; i + BLOCK_SIZE - 1 < Size.height; i += BLOCK_SIZE) { - for (int j = 0; j + BLOCK_SIZE - 1 < Size.width; j += BLOCK_SIZE) { + ROI Size) +{ + for (int i = 0; i + BLOCK_SIZE - 1 < Size.height; i += BLOCK_SIZE) + { + for (int j = 0; j + BLOCK_SIZE - 1 < Size.width; j += BLOCK_SIZE) + { // process rows - for (int k = 0; k < BLOCK_SIZE; k++) { + for (int k = 0; k < BLOCK_SIZE; k++) + { SubroutineIDCTvector((float *)fSrc + (i + k) * Stride + j, 1, fDst + (i + k) * Stride + j, 1); } // process columns - for (int k = 0; k < BLOCK_SIZE; k++) { + for (int k = 0; k < BLOCK_SIZE; k++) + { SubroutineIDCTvector(fDst + i * Stride + (j + k), Stride, fDst + i * Stride + (j + k), Stride); } diff --git a/Samples/2_Concepts_and_Techniques/dct8x8/Makefile b/Samples/2_Concepts_and_Techniques/dct8x8/Makefile deleted file mode 100644 index 181007c9..00000000 --- a/Samples/2_Concepts_and_Techniques/dct8x8/Makefile +++ /dev/null @@ -1,363 +0,0 @@ -################################################################################ -# Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved. -# -# Redistribution and use in source and binary forms, with or without -# modification, are permitted provided that the following conditions -# are met: -# * Redistributions of source code must retain the above copyright -# notice, this list of conditions and the following disclaimer. -# * Redistributions in binary form must reproduce the above copyright -# notice, this list of conditions and the following disclaimer in the -# documentation and/or other materials provided with the distribution. -# * Neither the name of NVIDIA CORPORATION nor the names of its -# contributors may be used to endorse or promote products derived -# from this software without specific prior written permission. -# -# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY -# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE -# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR -# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR -# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, -# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, -# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR -# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY -# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -# -################################################################################ -# -# Makefile project only supported on Mac OS X and Linux Platforms) -# -################################################################################ - -# Location of the CUDA Toolkit -CUDA_PATH ?= /usr/local/cuda - -############################## -# start deprecated interface # -############################## -ifeq ($(x86_64),1) - $(info WARNING - x86_64 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=x86_64 instead) - TARGET_ARCH ?= x86_64 -endif -ifeq ($(ARMv7),1) - $(info WARNING - ARMv7 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=armv7l instead) - TARGET_ARCH ?= armv7l -endif -ifeq ($(aarch64),1) - $(info WARNING - aarch64 variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=aarch64 instead) - TARGET_ARCH ?= aarch64 -endif -ifeq ($(ppc64le),1) - $(info WARNING - ppc64le variable has been deprecated) - $(info WARNING - please use TARGET_ARCH=ppc64le instead) - TARGET_ARCH ?= ppc64le -endif -ifneq ($(GCC),) - $(info WARNING - GCC variable has been deprecated) - $(info WARNING - please use HOST_COMPILER=$(GCC) instead) - HOST_COMPILER ?= $(GCC) -endif -ifneq ($(abi),) - $(error ERROR - abi variable has been removed) -endif -############################ -# end deprecated interface # -############################ - -# architecture -HOST_ARCH := $(shell uname -m) -TARGET_ARCH ?= $(HOST_ARCH) -ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 sbsa ppc64le armv7l)) - ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 sbsa ppc64le)) - TARGET_SIZE := 64 - else ifneq (,$(filter $(TARGET_ARCH),armv7l)) - TARGET_SIZE := 32 - endif - else - TARGET_SIZE := $(shell getconf LONG_BIT) - endif -else - $(error ERROR - unsupported value $(TARGET_ARCH) for TARGET_ARCH!) -endif - -# sbsa and aarch64 systems look similar. Need to differentiate them at host level for now. -ifeq ($(HOST_ARCH),aarch64) - ifeq ($(CUDA_PATH)/targets/sbsa-linux,$(shell ls -1d $(CUDA_PATH)/targets/sbsa-linux 2>/dev/null)) - HOST_ARCH := sbsa - TARGET_ARCH := sbsa - endif -endif - -ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq (,$(filter $(HOST_ARCH)-$(TARGET_ARCH),aarch64-armv7l x86_64-armv7l x86_64-aarch64 x86_64-sbsa x86_64-ppc64le)) - $(error ERROR - cross compiling from $(HOST_ARCH) to $(TARGET_ARCH) is not supported!) - endif -endif - -# When on native aarch64 system with userspace of 32-bit, change TARGET_ARCH to armv7l -ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_SIZE),aarch64-aarch64-32) - TARGET_ARCH = armv7l -endif - -# operating system -HOST_OS := $(shell uname -s 2>/dev/null | tr "[:upper:]" "[:lower:]") -TARGET_OS ?= $(HOST_OS) -ifeq (,$(filter $(TARGET_OS),linux darwin qnx android)) - $(error ERROR - unsupported value $(TARGET_OS) for TARGET_OS!) -endif - -# host compiler -ifdef HOST_COMPILER - CUSTOM_HOST_COMPILER = 1 -endif - -ifeq ($(TARGET_OS),darwin) - ifeq ($(shell expr `xcodebuild -version | grep -i xcode | awk '{print $$2}' | cut -d'.' -f1` \>= 5),1) - HOST_COMPILER ?= clang++ - endif -else ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq ($(HOST_ARCH)-$(TARGET_ARCH),x86_64-armv7l) - ifeq ($(TARGET_OS),linux) - HOST_COMPILER ?= arm-linux-gnueabihf-g++ - else ifeq ($(TARGET_OS),qnx) - ifeq ($(QNX_HOST),) - $(error ERROR - QNX_HOST must be passed to the QNX host toolchain) - endif - ifeq ($(QNX_TARGET),) - $(error ERROR - QNX_TARGET must be passed to the QNX target toolchain) - endif - export QNX_HOST - export QNX_TARGET - HOST_COMPILER ?= $(QNX_HOST)/usr/bin/arm-unknown-nto-qnx6.6.0eabi-g++ - else ifeq ($(TARGET_OS),android) - HOST_COMPILER ?= arm-linux-androideabi-g++ - endif - else ifeq ($(TARGET_ARCH),aarch64) - ifeq ($(TARGET_OS), linux) - HOST_COMPILER ?= aarch64-linux-gnu-g++ - else ifeq ($(TARGET_OS),qnx) - ifeq ($(QNX_HOST),) - $(error ERROR - QNX_HOST must be passed to the QNX host toolchain) - endif - ifeq ($(QNX_TARGET),) - $(error ERROR - QNX_TARGET must be passed to the QNX target toolchain) - endif - export QNX_HOST - export QNX_TARGET - HOST_COMPILER ?= $(QNX_HOST)/usr/bin/q++ - else ifeq ($(TARGET_OS), android) - HOST_COMPILER ?= aarch64-linux-android-clang++ - endif - else ifeq ($(TARGET_ARCH),sbsa) - HOST_COMPILER ?= aarch64-linux-gnu-g++ - else ifeq ($(TARGET_ARCH),ppc64le) - HOST_COMPILER ?= powerpc64le-linux-gnu-g++ - endif -endif -HOST_COMPILER ?= g++ -NVCC := $(CUDA_PATH)/bin/nvcc -ccbin $(HOST_COMPILER) - -# internal flags -NVCCFLAGS := -m${TARGET_SIZE} -CCFLAGS := -LDFLAGS := - -# build flags - -# Link flag for customized HOST_COMPILER with gcc realpath -GCC_PATH := $(shell which gcc) -ifeq ($(CUSTOM_HOST_COMPILER),1) - ifneq ($(filter /%,$(HOST_COMPILER)),) - ifneq ($(findstring gcc,$(HOST_COMPILER)),) - ifneq ($(GCC_PATH),$(HOST_COMPILER)) - LDFLAGS += -lstdc++ - endif - endif - endif -endif - -ifeq ($(TARGET_OS),darwin) - LDFLAGS += -rpath $(CUDA_PATH)/lib - CCFLAGS += -arch $(HOST_ARCH) -else ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_OS),x86_64-armv7l-linux) - LDFLAGS += --dynamic-linker=/lib/ld-linux-armhf.so.3 - CCFLAGS += -mfloat-abi=hard -else ifeq ($(TARGET_OS),android) - LDFLAGS += -pie - CCFLAGS += -fpie -fpic -fexceptions -endif - -ifneq ($(TARGET_ARCH),$(HOST_ARCH)) - ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-linux) - ifneq ($(TARGET_FS),) - GCCVERSIONLTEQ46 := $(shell expr `$(HOST_COMPILER) -dumpversion` \<= 4.6) - ifeq ($(GCCVERSIONLTEQ46),1) - CCFLAGS += --sysroot=$(TARGET_FS) - endif - LDFLAGS += --sysroot=$(TARGET_FS) - LDFLAGS += -rpath-link=$(TARGET_FS)/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib/arm-linux-gnueabihf - endif - endif - ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-linux) - ifneq ($(TARGET_FS),) - GCCVERSIONLTEQ46 := $(shell expr `$(HOST_COMPILER) -dumpversion` \<= 4.6) - ifeq ($(GCCVERSIONLTEQ46),1) - CCFLAGS += --sysroot=$(TARGET_FS) - endif - LDFLAGS += --sysroot=$(TARGET_FS) - LDFLAGS += -rpath-link=$(TARGET_FS)/lib -L$(TARGET_FS)/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/lib/aarch64-linux-gnu -L$(TARGET_FS)/lib/aarch64-linux-gnu - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib -L$(TARGET_FS)/usr/lib - LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib/aarch64-linux-gnu -L$(TARGET_FS)/usr/lib/aarch64-linux-gnu - LDFLAGS += --unresolved-symbols=ignore-in-shared-libs - CCFLAGS += -isystem=$(TARGET_FS)/usr/include -I$(TARGET_FS)/usr/include -I$(TARGET_FS)/usr/include/libdrm - CCFLAGS += -isystem=$(TARGET_FS)/usr/include/aarch64-linux-gnu -I$(TARGET_FS)/usr/include/aarch64-linux-gnu - endif - endif - ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx) - NVCCFLAGS += -D_QNX_SOURCE - NVCCFLAGS += --qpp-config 8.3.0,gcc_ntoaarch64le - CCFLAGS += -DWIN_INTERFACE_CUSTOM -I/usr/include/aarch64-qnx-gnu - LDFLAGS += -lsocket - LDFLAGS += -L/usr/lib/aarch64-qnx-gnu - CCFLAGS += "-Wl\,-rpath-link\,/usr/lib/aarch64-qnx-gnu" - ifdef TARGET_OVERRIDE - LDFLAGS += -lslog2 - endif - - ifneq ($(TARGET_FS),) - LDFLAGS += -L$(TARGET_FS)/usr/lib - CCFLAGS += "-Wl\,-rpath-link\,$(TARGET_FS)/usr/lib" - LDFLAGS += -L$(TARGET_FS)/usr/libnvidia - CCFLAGS += "-Wl\,-rpath-link\,$(TARGET_FS)/usr/libnvidia" - CCFLAGS += -I$(TARGET_FS)/../include - endif - endif -endif - -ifdef TARGET_OVERRIDE # cuda toolkit targets override - NVCCFLAGS += -target-dir $(TARGET_OVERRIDE) -endif - -# Install directory of different arch -CUDA_INSTALL_TARGET_DIR := -ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-linux) - CUDA_INSTALL_TARGET_DIR = targets/armv7-linux-gnueabihf/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-linux) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-linux/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),sbsa-linux) - CUDA_INSTALL_TARGET_DIR = targets/sbsa-linux/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-android) - CUDA_INSTALL_TARGET_DIR = targets/armv7-linux-androideabi/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-android) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-linux-androideabi/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-qnx) - CUDA_INSTALL_TARGET_DIR = targets/ARMv7-linux-QNX/ -else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx) - CUDA_INSTALL_TARGET_DIR = targets/aarch64-qnx/ -else ifeq ($(TARGET_ARCH),ppc64le) - CUDA_INSTALL_TARGET_DIR = targets/ppc64le-linux/ -endif - -# Debug build flags -ifeq ($(dbg),1) - NVCCFLAGS += -g -G - BUILD_TYPE := debug -else - BUILD_TYPE := release -endif - -ALL_CCFLAGS := -ALL_CCFLAGS += $(NVCCFLAGS) -ALL_CCFLAGS += $(EXTRA_NVCCFLAGS) -ALL_CCFLAGS += $(addprefix -Xcompiler ,$(CCFLAGS)) -ALL_CCFLAGS += $(addprefix -Xcompiler ,$(EXTRA_CCFLAGS)) - -SAMPLE_ENABLED := 1 - -ALL_LDFLAGS := -ALL_LDFLAGS += $(ALL_CCFLAGS) -ALL_LDFLAGS += $(addprefix -Xlinker ,$(LDFLAGS)) -ALL_LDFLAGS += $(addprefix -Xlinker ,$(EXTRA_LDFLAGS)) - -# Common includes and paths for CUDA -INCLUDES := -I../../../Common -LIBRARIES := - -################################################################################ - -# Gencode arguments -ifeq ($(TARGET_ARCH),$(filter $(TARGET_ARCH),armv7l aarch64 sbsa)) -SMS ?= 53 61 70 72 75 80 86 87 90 -else -SMS ?= 50 52 60 61 70 75 80 86 89 90 -endif - -ifeq ($(SMS),) -$(info >>> WARNING - no SM architectures have been specified - waiving sample <<<) -SAMPLE_ENABLED := 0 -endif - -ifeq ($(GENCODE_FLAGS),) -# Generate SASS code for each SM architecture listed in $(SMS) -$(foreach sm,$(SMS),$(eval GENCODE_FLAGS += -gencode arch=compute_$(sm),code=sm_$(sm))) - -# Generate PTX code from the highest SM architecture in $(SMS) to guarantee forward-compatibility -HIGHEST_SM := $(lastword $(sort $(SMS))) -ifneq ($(HIGHEST_SM),) -GENCODE_FLAGS += -gencode arch=compute_$(HIGHEST_SM),code=compute_$(HIGHEST_SM) -endif -endif - -ALL_CCFLAGS += --threads 0 --std=c++11 - -ifeq ($(SAMPLE_ENABLED),0) -EXEC ?= @echo "[@]" -endif - -################################################################################ - -# Target rules -all: build - -build: dct8x8 - -check.deps: -ifeq ($(SAMPLE_ENABLED),0) - @echo "Sample will be waived due to the above missing dependencies" -else - @echo "Sample is ready - all dependencies have been met" -endif - -BmpUtil.o:BmpUtil.cpp - $(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $< - -DCT8x8_Gold.o:DCT8x8_Gold.cpp - $(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $< - -dct8x8.o:dct8x8.cu - $(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $< - -dct8x8: BmpUtil.o DCT8x8_Gold.o dct8x8.o - $(EXEC) $(NVCC) $(ALL_LDFLAGS) $(GENCODE_FLAGS) -o $@ $+ $(LIBRARIES) - $(EXEC) mkdir -p ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE) - $(EXEC) cp $@ ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE) - -run: build - $(EXEC) ./dct8x8 - -testrun: build - -clean: - rm -f dct8x8 BmpUtil.o DCT8x8_Gold.o dct8x8.o - rm -rf ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE)/dct8x8 - -clobber: clean diff --git a/Samples/2_Concepts_and_Techniques/dct8x8/NsightEclipse.xml b/Samples/2_Concepts_and_Techniques/dct8x8/NsightEclipse.xml deleted file mode 100644 index 1231b7a3..00000000 --- a/Samples/2_Concepts_and_Techniques/dct8x8/NsightEclipse.xml +++ /dev/null @@ -1,89 +0,0 @@ - - - - dct8x8 - - cudaMallocArray - cudaFreeArray - cudaFree - cudaMallocPitch - cudaDestroyTextureObject - cudaDeviceSynchronize - cudaCreateTextureObject - - - whole - - data\teapot512.bmp - - - ./ - ../ - ../../../Common - - - Image Processing - Video Compression - - - CUDA - Image Processing - DCT - Discrete Cosine Transfrom - - - - - - true - dct8x8.cu - - 1:CUDA Advanced Topics - 2:Image Processing - 2:Data Compression - 2:Video Codecs - 2:Computer Vision - - sm50 - sm52 - sm53 - sm60 - sm61 - sm70 - sm72 - sm75 - sm80 - sm86 - sm87 - sm89 - sm90 - - - x86_64 - linux - - - windows7 - - - x86_64 - macosx - - - arm - - - sbsa - - - ppc64le - linux - - - - all - - DCT8x8 - exe - doc\dct8x8.pdf - diff --git a/Samples/CMakeLists.txt b/Samples/CMakeLists.txt index c4667bc1..b01437a4 100644 --- a/Samples/CMakeLists.txt +++ b/Samples/CMakeLists.txt @@ -1,2 +1,3 @@ add_subdirectory(0_Introduction) add_subdirectory(1_Utilities) +add_subdirectory(2_Concepts_and_Techniques)