mirror of
https://github.com/NVIDIA/cuda-samples.git
synced 2026-10-23 21:28:17 +08:00
add and update samples for CUDA 11.5
This commit is contained in:
Rutwik Choughule
18
Samples/histogram/.vscode/c_cpp_properties.json
vendored
Normal file
18
Samples/histogram/.vscode/c_cpp_properties.json
vendored
Normal file
@@ -0,0 +1,18 @@
|
||||
{
|
||||
"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
|
||||
}
|
||||
7
Samples/histogram/.vscode/extensions.json
vendored
Normal file
7
Samples/histogram/.vscode/extensions.json
vendored
Normal file
@@ -0,0 +1,7 @@
|
||||
{
|
||||
"recommendations": [
|
||||
"nvidia.nsight-vscode-edition",
|
||||
"ms-vscode.cpptools",
|
||||
"ms-vscode.makefile-tools"
|
||||
]
|
||||
}
|
||||
10
Samples/histogram/.vscode/launch.json
vendored
Normal file
10
Samples/histogram/.vscode/launch.json
vendored
Normal file
@@ -0,0 +1,10 @@
|
||||
{
|
||||
"configurations": [
|
||||
{
|
||||
"name": "CUDA C++: Launch",
|
||||
"type": "cuda-gdb",
|
||||
"request": "launch",
|
||||
"program": "${workspaceFolder}/histogram"
|
||||
}
|
||||
]
|
||||
}
|
||||
15
Samples/histogram/.vscode/tasks.json
vendored
Normal file
15
Samples/histogram/.vscode/tasks.json
vendored
Normal file
@@ -0,0 +1,15 @@
|
||||
{
|
||||
"version": "2.0.0",
|
||||
"tasks": [
|
||||
{
|
||||
"label": "sample",
|
||||
"type": "shell",
|
||||
"command": "make dbg=1",
|
||||
"problemMatcher": ["$nvcc"],
|
||||
"group": {
|
||||
"kind": "build",
|
||||
"isDefault": true
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
371
Samples/histogram/Makefile
Normal file
371
Samples/histogram/Makefile
Normal file
@@ -0,0 +1,371 @@
|
||||
################################################################################
|
||||
# Copyright (c) 2019, 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 $(TARGET_OS) for TARGET_OS!)
|
||||
endif
|
||||
|
||||
# host compiler
|
||||
ifeq ($(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
|
||||
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 += --qpp-config 5.4.0,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 :=
|
||||
|
||||
################################################################################
|
||||
|
||||
#Detect if installed version of GCC supports required C++11
|
||||
ifeq ($(TARGET_OS),linux)
|
||||
empty :=
|
||||
space := $(empty) $(empty)
|
||||
GCCVERSIONSTRING := $(shell expr `$(HOST_COMPILER) -dumpversion`)
|
||||
#Create version number without "."
|
||||
GCCVERSION := $(shell expr `echo $(GCCVERSIONSTRING)` | cut -f1 -d.)
|
||||
GCCVERSION += $(shell expr `echo $(GCCVERSIONSTRING)` | cut -f2 -d.)
|
||||
GCCVERSION += $(shell expr `echo $(GCCVERSIONSTRING)` | cut -f3 -d.)
|
||||
# Make sure the version number has at least 3 decimals
|
||||
GCCVERSION += 00
|
||||
# Remove spaces from the version number
|
||||
GCCVERSION := $(subst $(space),$(empty),$(GCCVERSION))
|
||||
#$(warning $(GCCVERSION))
|
||||
|
||||
IS_MIN_VERSION := $(shell expr `echo $(GCCVERSION)` \>= 51000)
|
||||
|
||||
ifeq ($(IS_MIN_VERSION), 1)
|
||||
$(info >>> GCC Version is greater or equal to 5.1.0 <<<)
|
||||
else
|
||||
$(info >>> Waiving build. Minimum GCC version required is 5.1.0<<<)
|
||||
SAMPLE_ENABLED := 0
|
||||
endif
|
||||
endif
|
||||
|
||||
# Gencode arguments
|
||||
ifeq ($(TARGET_ARCH),$(filter $(TARGET_ARCH),armv7l aarch64))
|
||||
SMS ?= 35 37 50 52 60 61 70 72 75 80 86
|
||||
else
|
||||
SMS ?= 35 37 50 52 60 61 70 75 80 86
|
||||
endif
|
||||
|
||||
ifeq ($(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 ($(SAMPLE_ENABLED),0)
|
||||
EXEC ?= @echo "[@]"
|
||||
endif
|
||||
|
||||
################################################################################
|
||||
|
||||
# Target rules
|
||||
all: build
|
||||
|
||||
build: histogram
|
||||
|
||||
check.deps:
|
||||
ifeq ($(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
|
||||
|
||||
histogram256.o:histogram256.cu
|
||||
$(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $<
|
||||
|
||||
histogram64.o:histogram64.cu
|
||||
$(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $<
|
||||
|
||||
histogram_gold.o:histogram_gold.cpp
|
||||
$(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $<
|
||||
|
||||
main.o:main.cpp
|
||||
$(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $<
|
||||
|
||||
histogram: histogram256.o histogram64.o histogram_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
|
||||
$(EXEC) ./histogram
|
||||
|
||||
clean:
|
||||
rm -f histogram histogram256.o histogram64.o histogram_gold.o main.o
|
||||
rm -rf ../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE)/histogram
|
||||
|
||||
clobber: clean
|
||||
72
Samples/histogram/NsightEclipse.xml
Normal file
72
Samples/histogram/NsightEclipse.xml
Normal file
@@ -0,0 +1,72 @@
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<!DOCTYPE entry SYSTEM "SamplesInfo.dtd">
|
||||
<entry>
|
||||
<name>histogram</name>
|
||||
<description><![CDATA[This sample demonstrates efficient implementation of 64-bin and 256-bin histogram.]]></description>
|
||||
<devicecompilation>whole</devicecompilation>
|
||||
<includepaths>
|
||||
<path>./</path>
|
||||
<path>../</path>
|
||||
<path>../../Common</path>
|
||||
</includepaths>
|
||||
<keyconcepts>
|
||||
<concept level="advanced">Image Processing</concept>
|
||||
<concept level="advanced">Data Parallel Algorithms</concept>
|
||||
</keyconcepts>
|
||||
<keywords>
|
||||
<keyword>CUDA</keyword>
|
||||
<keyword>Data-Parallel Algorithms</keyword>
|
||||
<keyword>Image Processing</keyword>
|
||||
<keyword>histogram</keyword>
|
||||
</keywords>
|
||||
<libraries>
|
||||
</libraries>
|
||||
<librarypaths>
|
||||
</librarypaths>
|
||||
<nsight_eclipse>true</nsight_eclipse>
|
||||
<primary_file>main.cpp</primary_file>
|
||||
<scopes>
|
||||
<scope>1:CUDA Advanced Topics</scope>
|
||||
<scope>1:Data-Parallel Algorithms</scope>
|
||||
<scope>2:Image Processing</scope>
|
||||
<scope>2:Data Compression</scope>
|
||||
<scope>2:Computer Vision</scope>
|
||||
</scopes>
|
||||
<sm-arch>sm35</sm-arch>
|
||||
<sm-arch>sm37</sm-arch>
|
||||
<sm-arch>sm50</sm-arch>
|
||||
<sm-arch>sm52</sm-arch>
|
||||
<sm-arch>sm60</sm-arch>
|
||||
<sm-arch>sm61</sm-arch>
|
||||
<sm-arch>sm70</sm-arch>
|
||||
<sm-arch>sm72</sm-arch>
|
||||
<sm-arch>sm75</sm-arch>
|
||||
<sm-arch>sm80</sm-arch>
|
||||
<sm-arch>sm86</sm-arch>
|
||||
<supported_envs>
|
||||
<env>
|
||||
<arch>x86_64</arch>
|
||||
<platform>linux</platform>
|
||||
</env>
|
||||
<env>
|
||||
<platform>windows7</platform>
|
||||
</env>
|
||||
<env>
|
||||
<arch>x86_64</arch>
|
||||
<platform>macosx</platform>
|
||||
</env>
|
||||
<env>
|
||||
<arch>arm</arch>
|
||||
</env>
|
||||
<env>
|
||||
<arch>ppc64le</arch>
|
||||
<platform>linux</platform>
|
||||
</env>
|
||||
</supported_envs>
|
||||
<supported_sm_architectures>
|
||||
<include>all</include>
|
||||
</supported_sm_architectures>
|
||||
<title>CUDA Histogram</title>
|
||||
<type>exe</type>
|
||||
<whitepaper>doc\histogram.pdf</whitepaper>
|
||||
</entry>
|
||||
67
Samples/histogram/README.md
Normal file
67
Samples/histogram/README.md
Normal file
@@ -0,0 +1,67 @@
|
||||
# histogram - CUDA Histogram
|
||||
|
||||
## Description
|
||||
|
||||
This sample demonstrates efficient implementation of 64-bin and 256-bin histogram.
|
||||
|
||||
## Key Concepts
|
||||
|
||||
Image Processing, Data Parallel Algorithms
|
||||
|
||||
## Supported SM Architectures
|
||||
|
||||
[SM 3.5 ](https://developer.nvidia.com/cuda-gpus) [SM 3.7 ](https://developer.nvidia.com/cuda-gpus) [SM 5.0 ](https://developer.nvidia.com/cuda-gpus) [SM 5.2 ](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)
|
||||
|
||||
## Supported OSes
|
||||
|
||||
Linux, Windows
|
||||
|
||||
## Supported CPU Architecture
|
||||
|
||||
x86_64, ppc64le, armv7l
|
||||
|
||||
## CUDA APIs involved
|
||||
|
||||
## Prerequisites
|
||||
|
||||
Download and install the [CUDA Toolkit 11.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform.
|
||||
|
||||
## Build and Run
|
||||
|
||||
### Windows
|
||||
The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format:
|
||||
```
|
||||
*_vs<version>.sln - for Visual Studio <version>
|
||||
```
|
||||
Each individual sample has its own set of solution files in its directory:
|
||||
|
||||
To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used.
|
||||
> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details."
|
||||
|
||||
### 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 <sample_dir>
|
||||
$ make
|
||||
```
|
||||
The samples makefiles can take advantage of certain options:
|
||||
* **TARGET_ARCH=<arch>** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l.
|
||||
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.<br/>
|
||||
`$ make TARGET_ARCH=x86_64` <br/> `$ make TARGET_ARCH=ppc64le` <br/> `$ make TARGET_ARCH=armv7l` <br/>
|
||||
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=<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)
|
||||
|
||||
BIN
Samples/histogram/doc/histogram.doc
Normal file
BIN
Samples/histogram/doc/histogram.doc
Normal file
Binary file not shown.
BIN
Samples/histogram/doc/histogram.pdf
Normal file
BIN
Samples/histogram/doc/histogram.pdf
Normal file
Binary file not shown.
BIN
Samples/histogram/doc/histogram.vsd
Normal file
BIN
Samples/histogram/doc/histogram.vsd
Normal file
Binary file not shown.
163
Samples/histogram/histogram256.cu
Normal file
163
Samples/histogram/histogram256.cu
Normal file
@@ -0,0 +1,163 @@
|
||||
/* Copyright (c) 2021, 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.
|
||||
*/
|
||||
|
||||
#include <assert.h>
|
||||
#include <stdio.h>
|
||||
#include <stdlib.h>
|
||||
#include <string.h>
|
||||
#include <cooperative_groups.h>
|
||||
|
||||
namespace cg = cooperative_groups;
|
||||
#include <helper_cuda.h>
|
||||
#include "histogram_common.h"
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// Shortcut shared memory atomic addition functions
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
|
||||
#define TAG_MASK 0xFFFFFFFFU
|
||||
inline __device__ void addByte(uint *s_WarpHist, uint data, uint threadTag) {
|
||||
atomicAdd(s_WarpHist + data, 1);
|
||||
}
|
||||
|
||||
inline __device__ void addWord(uint *s_WarpHist, uint data, uint tag) {
|
||||
addByte(s_WarpHist, (data >> 0) & 0xFFU, tag);
|
||||
addByte(s_WarpHist, (data >> 8) & 0xFFU, tag);
|
||||
addByte(s_WarpHist, (data >> 16) & 0xFFU, tag);
|
||||
addByte(s_WarpHist, (data >> 24) & 0xFFU, tag);
|
||||
}
|
||||
|
||||
__global__ void histogram256Kernel(uint *d_PartialHistograms, uint *d_Data,
|
||||
uint dataCount) {
|
||||
// Handle to thread block group
|
||||
cg::thread_block cta = cg::this_thread_block();
|
||||
// Per-warp subhistogram storage
|
||||
__shared__ uint s_Hist[HISTOGRAM256_THREADBLOCK_MEMORY];
|
||||
uint *s_WarpHist =
|
||||
s_Hist + (threadIdx.x >> LOG2_WARP_SIZE) * HISTOGRAM256_BIN_COUNT;
|
||||
|
||||
// Clear shared memory storage for current threadblock before processing
|
||||
#pragma unroll
|
||||
|
||||
for (uint i = 0;
|
||||
i < (HISTOGRAM256_THREADBLOCK_MEMORY / HISTOGRAM256_THREADBLOCK_SIZE);
|
||||
i++) {
|
||||
s_Hist[threadIdx.x + i * HISTOGRAM256_THREADBLOCK_SIZE] = 0;
|
||||
}
|
||||
|
||||
// Cycle through the entire data set, update subhistograms for each warp
|
||||
const uint tag = threadIdx.x << (UINT_BITS - LOG2_WARP_SIZE);
|
||||
|
||||
cg::sync(cta);
|
||||
|
||||
for (uint pos = UMAD(blockIdx.x, blockDim.x, threadIdx.x); pos < dataCount;
|
||||
pos += UMUL(blockDim.x, gridDim.x)) {
|
||||
uint data = d_Data[pos];
|
||||
addWord(s_WarpHist, data, tag);
|
||||
}
|
||||
|
||||
// Merge per-warp histograms into per-block and write to global memory
|
||||
cg::sync(cta);
|
||||
|
||||
for (uint bin = threadIdx.x; bin < HISTOGRAM256_BIN_COUNT;
|
||||
bin += HISTOGRAM256_THREADBLOCK_SIZE) {
|
||||
uint sum = 0;
|
||||
|
||||
for (uint i = 0; i < WARP_COUNT; i++) {
|
||||
sum += s_Hist[bin + i * HISTOGRAM256_BIN_COUNT] & TAG_MASK;
|
||||
}
|
||||
|
||||
d_PartialHistograms[blockIdx.x * HISTOGRAM256_BIN_COUNT + bin] = sum;
|
||||
}
|
||||
}
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// Merge histogram256() output
|
||||
// Run one threadblock per bin; each threadblock adds up the same bin counter
|
||||
// from every partial histogram. Reads are uncoalesced, but mergeHistogram256
|
||||
// takes only a fraction of total processing time
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
#define MERGE_THREADBLOCK_SIZE 256
|
||||
|
||||
__global__ void mergeHistogram256Kernel(uint *d_Histogram,
|
||||
uint *d_PartialHistograms,
|
||||
uint histogramCount) {
|
||||
// Handle to thread block group
|
||||
cg::thread_block cta = cg::this_thread_block();
|
||||
|
||||
uint sum = 0;
|
||||
|
||||
for (uint i = threadIdx.x; i < histogramCount; i += MERGE_THREADBLOCK_SIZE) {
|
||||
sum += d_PartialHistograms[blockIdx.x + i * HISTOGRAM256_BIN_COUNT];
|
||||
}
|
||||
|
||||
__shared__ uint data[MERGE_THREADBLOCK_SIZE];
|
||||
data[threadIdx.x] = sum;
|
||||
|
||||
for (uint stride = MERGE_THREADBLOCK_SIZE / 2; stride > 0; stride >>= 1) {
|
||||
cg::sync(cta);
|
||||
|
||||
if (threadIdx.x < stride) {
|
||||
data[threadIdx.x] += data[threadIdx.x + stride];
|
||||
}
|
||||
}
|
||||
|
||||
if (threadIdx.x == 0) {
|
||||
d_Histogram[blockIdx.x] = data[0];
|
||||
}
|
||||
}
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// Host interface to GPU histogram
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// histogram256kernel() intermediate results buffer
|
||||
static const uint PARTIAL_HISTOGRAM256_COUNT = 240;
|
||||
static uint *d_PartialHistograms;
|
||||
|
||||
// Internal memory allocation
|
||||
extern "C" void initHistogram256(void) {
|
||||
checkCudaErrors(cudaMalloc(
|
||||
(void **)&d_PartialHistograms,
|
||||
PARTIAL_HISTOGRAM256_COUNT * HISTOGRAM256_BIN_COUNT * sizeof(uint)));
|
||||
}
|
||||
|
||||
// Internal memory deallocation
|
||||
extern "C" void closeHistogram256(void) {
|
||||
checkCudaErrors(cudaFree(d_PartialHistograms));
|
||||
}
|
||||
|
||||
extern "C" void histogram256(uint *d_Histogram, void *d_Data, uint byteCount) {
|
||||
assert(byteCount % sizeof(uint) == 0);
|
||||
histogram256Kernel<<<PARTIAL_HISTOGRAM256_COUNT,
|
||||
HISTOGRAM256_THREADBLOCK_SIZE>>>(
|
||||
d_PartialHistograms, (uint *)d_Data, byteCount / sizeof(uint));
|
||||
getLastCudaError("histogram256Kernel() execution failed\n");
|
||||
|
||||
mergeHistogram256Kernel<<<HISTOGRAM256_BIN_COUNT, MERGE_THREADBLOCK_SIZE>>>(
|
||||
d_Histogram, d_PartialHistograms, PARTIAL_HISTOGRAM256_COUNT);
|
||||
getLastCudaError("mergeHistogram256Kernel() execution failed\n");
|
||||
}
|
||||
205
Samples/histogram/histogram64.cu
Normal file
205
Samples/histogram/histogram64.cu
Normal file
@@ -0,0 +1,205 @@
|
||||
/* Copyright (c) 2021, 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.
|
||||
*/
|
||||
|
||||
#include <assert.h>
|
||||
#include <stdio.h>
|
||||
#include <stdlib.h>
|
||||
#include <string.h>
|
||||
#include <cooperative_groups.h>
|
||||
|
||||
namespace cg = cooperative_groups;
|
||||
#include <helper_cuda.h>
|
||||
#include "histogram_common.h"
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// GPU-specific common definitions
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// Data type used for input data fetches
|
||||
typedef uint4 data_t;
|
||||
|
||||
// May change on future hardware, so better parametrize the code
|
||||
#define SHARED_MEMORY_BANKS 16
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// Main computation pass: compute gridDim.x partial histograms
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// Count a byte into shared-memory storage
|
||||
inline __device__ void addByte(uchar *s_ThreadBase, uint data) {
|
||||
s_ThreadBase[UMUL(data, HISTOGRAM64_THREADBLOCK_SIZE)]++;
|
||||
}
|
||||
|
||||
// Count four bytes of a word
|
||||
inline __device__ void addWord(uchar *s_ThreadBase, uint data) {
|
||||
// Only higher 6 bits of each byte matter, as this is a 64-bin histogram
|
||||
addByte(s_ThreadBase, (data >> 2) & 0x3FU);
|
||||
addByte(s_ThreadBase, (data >> 10) & 0x3FU);
|
||||
addByte(s_ThreadBase, (data >> 18) & 0x3FU);
|
||||
addByte(s_ThreadBase, (data >> 26) & 0x3FU);
|
||||
}
|
||||
|
||||
__global__ void histogram64Kernel(uint *d_PartialHistograms, data_t *d_Data,
|
||||
uint dataCount) {
|
||||
// Handle to thread block group
|
||||
cg::thread_block cta = cg::this_thread_block();
|
||||
// Encode thread index in order to avoid bank conflicts in s_Hist[] access:
|
||||
// each group of SHARED_MEMORY_BANKS threads accesses consecutive shared
|
||||
// memory banks
|
||||
// and the same bytes [0..3] within the banks
|
||||
// Because of this permutation block size should be a multiple of 4 *
|
||||
// SHARED_MEMORY_BANKS
|
||||
const uint threadPos = ((threadIdx.x & ~(SHARED_MEMORY_BANKS * 4 - 1)) << 0) |
|
||||
((threadIdx.x & (SHARED_MEMORY_BANKS - 1)) << 2) |
|
||||
((threadIdx.x & (SHARED_MEMORY_BANKS * 3)) >> 4);
|
||||
|
||||
// Per-thread histogram storage
|
||||
__shared__ uchar s_Hist[HISTOGRAM64_THREADBLOCK_SIZE * HISTOGRAM64_BIN_COUNT];
|
||||
uchar *s_ThreadBase = s_Hist + threadPos;
|
||||
|
||||
// Initialize shared memory (writing 32-bit words)
|
||||
#pragma unroll
|
||||
|
||||
for (uint i = 0; i < (HISTOGRAM64_BIN_COUNT / 4); i++) {
|
||||
((uint *)s_Hist)[threadIdx.x + i * HISTOGRAM64_THREADBLOCK_SIZE] = 0;
|
||||
}
|
||||
|
||||
// Read data from global memory and submit to the shared-memory histogram
|
||||
// Since histogram counters are byte-sized, every single thread can't do more
|
||||
// than 255 submission
|
||||
cg::sync(cta);
|
||||
|
||||
for (uint pos = UMAD(blockIdx.x, blockDim.x, threadIdx.x); pos < dataCount;
|
||||
pos += UMUL(blockDim.x, gridDim.x)) {
|
||||
data_t data = d_Data[pos];
|
||||
addWord(s_ThreadBase, data.x);
|
||||
addWord(s_ThreadBase, data.y);
|
||||
addWord(s_ThreadBase, data.z);
|
||||
addWord(s_ThreadBase, data.w);
|
||||
}
|
||||
|
||||
// Accumulate per-thread histograms into per-block and write to global memory
|
||||
cg::sync(cta);
|
||||
|
||||
if (threadIdx.x < HISTOGRAM64_BIN_COUNT) {
|
||||
uchar *s_HistBase =
|
||||
s_Hist + UMUL(threadIdx.x, HISTOGRAM64_THREADBLOCK_SIZE);
|
||||
|
||||
uint sum = 0;
|
||||
uint pos = 4 * (threadIdx.x & (SHARED_MEMORY_BANKS - 1));
|
||||
|
||||
#pragma unroll
|
||||
|
||||
for (uint i = 0; i < (HISTOGRAM64_THREADBLOCK_SIZE / 4); i++) {
|
||||
sum += s_HistBase[pos + 0] + s_HistBase[pos + 1] + s_HistBase[pos + 2] +
|
||||
s_HistBase[pos + 3];
|
||||
pos = (pos + 4) & (HISTOGRAM64_THREADBLOCK_SIZE - 1);
|
||||
}
|
||||
|
||||
d_PartialHistograms[blockIdx.x * HISTOGRAM64_BIN_COUNT + threadIdx.x] = sum;
|
||||
}
|
||||
}
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// Merge histogram64() output
|
||||
// Run one threadblock per bin; each threadbock adds up the same bin counter
|
||||
// from every partial histogram. Reads are uncoalesced, but mergeHistogram64
|
||||
// takes only a fraction of total processing time
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
#define MERGE_THREADBLOCK_SIZE 256
|
||||
|
||||
__global__ void mergeHistogram64Kernel(uint *d_Histogram,
|
||||
uint *d_PartialHistograms,
|
||||
uint histogramCount) {
|
||||
// Handle to thread block group
|
||||
cg::thread_block cta = cg::this_thread_block();
|
||||
__shared__ uint data[MERGE_THREADBLOCK_SIZE];
|
||||
|
||||
uint sum = 0;
|
||||
|
||||
for (uint i = threadIdx.x; i < histogramCount; i += MERGE_THREADBLOCK_SIZE) {
|
||||
sum += d_PartialHistograms[blockIdx.x + i * HISTOGRAM64_BIN_COUNT];
|
||||
}
|
||||
|
||||
data[threadIdx.x] = sum;
|
||||
|
||||
for (uint stride = MERGE_THREADBLOCK_SIZE / 2; stride > 0; stride >>= 1) {
|
||||
cg::sync(cta);
|
||||
|
||||
if (threadIdx.x < stride) {
|
||||
data[threadIdx.x] += data[threadIdx.x + stride];
|
||||
}
|
||||
}
|
||||
|
||||
if (threadIdx.x == 0) {
|
||||
d_Histogram[blockIdx.x] = data[0];
|
||||
}
|
||||
}
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// CPU interface to GPU histogram calculator
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// histogram64kernel() intermediate results buffer
|
||||
// MAX_PARTIAL_HISTOGRAM64_COUNT == 32768 and HISTOGRAM64_THREADBLOCK_SIZE == 64
|
||||
// amounts to max. 480MB of input data
|
||||
static const uint MAX_PARTIAL_HISTOGRAM64_COUNT = 32768;
|
||||
static uint *d_PartialHistograms;
|
||||
|
||||
// Internal memory allocation
|
||||
extern "C" void initHistogram64(void) {
|
||||
assert(HISTOGRAM64_THREADBLOCK_SIZE % (4 * SHARED_MEMORY_BANKS) == 0);
|
||||
checkCudaErrors(cudaMalloc(
|
||||
(void **)&d_PartialHistograms,
|
||||
MAX_PARTIAL_HISTOGRAM64_COUNT * HISTOGRAM64_BIN_COUNT * sizeof(uint)));
|
||||
}
|
||||
|
||||
// Internal memory deallocation
|
||||
extern "C" void closeHistogram64(void) {
|
||||
checkCudaErrors(cudaFree(d_PartialHistograms));
|
||||
}
|
||||
|
||||
// Round a / b to nearest higher integer value
|
||||
inline uint iDivUp(uint a, uint b) {
|
||||
return (a % b != 0) ? (a / b + 1) : (a / b);
|
||||
}
|
||||
|
||||
// Snap a to nearest lower multiple of b
|
||||
inline uint iSnapDown(uint a, uint b) { return a - a % b; }
|
||||
|
||||
extern "C" void histogram64(uint *d_Histogram, void *d_Data, uint byteCount) {
|
||||
const uint histogramCount = iDivUp(
|
||||
byteCount, HISTOGRAM64_THREADBLOCK_SIZE * iSnapDown(255, sizeof(data_t)));
|
||||
|
||||
assert(byteCount % sizeof(data_t) == 0);
|
||||
assert(histogramCount <= MAX_PARTIAL_HISTOGRAM64_COUNT);
|
||||
|
||||
histogram64Kernel<<<histogramCount, HISTOGRAM64_THREADBLOCK_SIZE>>>(
|
||||
d_PartialHistograms, (data_t *)d_Data, byteCount / sizeof(data_t));
|
||||
getLastCudaError("histogram64Kernel() execution failed\n");
|
||||
|
||||
mergeHistogram64Kernel<<<HISTOGRAM64_BIN_COUNT, MERGE_THREADBLOCK_SIZE>>>(
|
||||
d_Histogram, d_PartialHistograms, histogramCount);
|
||||
getLastCudaError("mergeHistogram64() execution failed\n");
|
||||
}
|
||||
85
Samples/histogram/histogram_common.h
Normal file
85
Samples/histogram/histogram_common.h
Normal file
@@ -0,0 +1,85 @@
|
||||
/* Copyright (c) 2021, 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 HISTOGRAM_COMMON_H
|
||||
#define HISTOGRAM_COMMON_H
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// Common definitions
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
#define HISTOGRAM64_BIN_COUNT 64
|
||||
#define HISTOGRAM256_BIN_COUNT 256
|
||||
#define UINT_BITS 32
|
||||
typedef unsigned int uint;
|
||||
typedef unsigned char uchar;
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// GPU-specific common definitions
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
#define LOG2_WARP_SIZE 5U
|
||||
#define WARP_SIZE (1U << LOG2_WARP_SIZE)
|
||||
|
||||
// May change on future hardware, so better parametrize the code
|
||||
#define SHARED_MEMORY_BANKS 16
|
||||
|
||||
// Threadblock size: must be a multiple of (4 * SHARED_MEMORY_BANKS)
|
||||
// because of the bit permutation of threadIdx.x
|
||||
#define HISTOGRAM64_THREADBLOCK_SIZE (4 * SHARED_MEMORY_BANKS)
|
||||
|
||||
// Warps ==subhistograms per threadblock
|
||||
#define WARP_COUNT 6
|
||||
|
||||
// Threadblock size
|
||||
#define HISTOGRAM256_THREADBLOCK_SIZE (WARP_COUNT * WARP_SIZE)
|
||||
|
||||
// Shared memory per threadblock
|
||||
#define HISTOGRAM256_THREADBLOCK_MEMORY (WARP_COUNT * HISTOGRAM256_BIN_COUNT)
|
||||
|
||||
#define UMUL(a, b) ((a) * (b))
|
||||
#define UMAD(a, b, c) (UMUL((a), (b)) + (c))
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// Reference CPU histogram
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
extern "C" void histogram64CPU(uint *h_Histogram, void *h_Data, uint byteCount);
|
||||
|
||||
extern "C" void histogram256CPU(uint *h_Histogram, void *h_Data,
|
||||
uint byteCount);
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
// GPU histogram
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
extern "C" void initHistogram64(void);
|
||||
extern "C" void initHistogram256(void);
|
||||
extern "C" void closeHistogram64(void);
|
||||
extern "C" void closeHistogram256(void);
|
||||
|
||||
extern "C" void histogram64(uint *d_Histogram, void *d_Data, uint byteCount);
|
||||
|
||||
extern "C" void histogram256(uint *d_Histogram, void *d_Data, uint byteCount);
|
||||
|
||||
#endif
|
||||
59
Samples/histogram/histogram_gold.cpp
Normal file
59
Samples/histogram/histogram_gold.cpp
Normal file
@@ -0,0 +1,59 @@
|
||||
/* Copyright (c) 2021, 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.
|
||||
*/
|
||||
|
||||
#include <assert.h>
|
||||
#include "histogram_common.h"
|
||||
|
||||
extern "C" void histogram64CPU(uint *h_Histogram, void *h_Data,
|
||||
uint byteCount) {
|
||||
for (uint i = 0; i < HISTOGRAM64_BIN_COUNT; i++) h_Histogram[i] = 0;
|
||||
|
||||
assert(sizeof(uint) == 4 && (byteCount % 4) == 0);
|
||||
|
||||
for (uint i = 0; i < (byteCount / 4); i++) {
|
||||
uint data = ((uint *)h_Data)[i];
|
||||
h_Histogram[(data >> 2) & 0x3FU]++;
|
||||
h_Histogram[(data >> 10) & 0x3FU]++;
|
||||
h_Histogram[(data >> 18) & 0x3FU]++;
|
||||
h_Histogram[(data >> 26) & 0x3FU]++;
|
||||
}
|
||||
}
|
||||
|
||||
extern "C" void histogram256CPU(uint *h_Histogram, void *h_Data,
|
||||
uint byteCount) {
|
||||
for (uint i = 0; i < HISTOGRAM256_BIN_COUNT; i++) h_Histogram[i] = 0;
|
||||
|
||||
assert(sizeof(uint) == 4 && (byteCount % 4) == 0);
|
||||
|
||||
for (uint i = 0; i < (byteCount / 4); i++) {
|
||||
uint data = ((uint *)h_Data)[i];
|
||||
h_Histogram[(data >> 0) & 0xFFU]++;
|
||||
h_Histogram[(data >> 8) & 0xFFU]++;
|
||||
h_Histogram[(data >> 16) & 0xFFU]++;
|
||||
h_Histogram[(data >> 24) & 0xFFU]++;
|
||||
}
|
||||
}
|
||||
20
Samples/histogram/histogram_vs2017.sln
Normal file
20
Samples/histogram/histogram_vs2017.sln
Normal file
@@ -0,0 +1,20 @@
|
||||
|
||||
Microsoft Visual Studio Solution File, Format Version 12.00
|
||||
# Visual Studio 2017
|
||||
Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "histogram", "histogram_vs2017.vcxproj", "{997E0757-EA74-4A4E-A0FC-47D8C8831A15}"
|
||||
EndProject
|
||||
Global
|
||||
GlobalSection(SolutionConfigurationPlatforms) = preSolution
|
||||
Debug|x64 = Debug|x64
|
||||
Release|x64 = Release|x64
|
||||
EndGlobalSection
|
||||
GlobalSection(ProjectConfigurationPlatforms) = postSolution
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Debug|x64.ActiveCfg = Debug|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Debug|x64.Build.0 = Debug|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Release|x64.ActiveCfg = Release|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Release|x64.Build.0 = Release|x64
|
||||
EndGlobalSection
|
||||
GlobalSection(SolutionProperties) = preSolution
|
||||
HideSolutionNode = FALSE
|
||||
EndGlobalSection
|
||||
EndGlobal
|
||||
116
Samples/histogram/histogram_vs2017.vcxproj
Normal file
116
Samples/histogram/histogram_vs2017.vcxproj
Normal file
@@ -0,0 +1,116 @@
|
||||
<?xml version="1.0" encoding="utf-8"?>
|
||||
<Project DefaultTargets="Build" ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
|
||||
<PropertyGroup>
|
||||
<CUDAPropsPath Condition="'$(CUDAPropsPath)'==''">$(VCTargetsPath)\BuildCustomizations</CUDAPropsPath>
|
||||
</PropertyGroup>
|
||||
<ItemGroup Label="ProjectConfigurations">
|
||||
<ProjectConfiguration Include="Debug|x64">
|
||||
<Configuration>Debug</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
<ProjectConfiguration Include="Release|x64">
|
||||
<Configuration>Release</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
</ItemGroup>
|
||||
<PropertyGroup Label="Globals">
|
||||
<ProjectGuid>{997E0757-EA74-4A4E-A0FC-47D8C8831A15}</ProjectGuid>
|
||||
<RootNamespace>histogram_vs2017</RootNamespace>
|
||||
<ProjectName>histogram</ProjectName>
|
||||
<CudaToolkitCustomDir />
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(WindowsTargetPlatformVersion)'==''">
|
||||
<LatestTargetPlatformVersion>$([Microsoft.Build.Utilities.ToolLocationHelper]::GetLatestSDKTargetPlatformVersion('Windows', '10.0'))</LatestTargetPlatformVersion>
|
||||
<WindowsTargetPlatformVersion Condition="'$(WindowsTargetPlatformVersion)' == ''">$(LatestTargetPlatformVersion)</WindowsTargetPlatformVersion>
|
||||
<TargetPlatformVersion>$(WindowsTargetPlatformVersion)</TargetPlatformVersion>
|
||||
</PropertyGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
|
||||
<PropertyGroup>
|
||||
<ConfigurationType>Application</ConfigurationType>
|
||||
<CharacterSet>MultiByte</CharacterSet>
|
||||
<PlatformToolset>v141</PlatformToolset>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)'=='Debug'">
|
||||
<UseDebugLibraries>true</UseDebugLibraries>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)'=='Release'">
|
||||
<WholeProgramOptimization>true</WholeProgramOptimization>
|
||||
</PropertyGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
|
||||
<ImportGroup Label="ExtensionSettings">
|
||||
<Import Project="$(CUDAPropsPath)\CUDA 11.5.props" />
|
||||
</ImportGroup>
|
||||
<ImportGroup Label="PropertySheets">
|
||||
<Import Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" />
|
||||
</ImportGroup>
|
||||
<PropertyGroup Label="UserMacros" />
|
||||
<PropertyGroup>
|
||||
<IntDir>$(Platform)/$(Configuration)/</IntDir>
|
||||
<IncludePath>$(IncludePath)</IncludePath>
|
||||
<CodeAnalysisRuleSet>AllRules.ruleset</CodeAnalysisRuleSet>
|
||||
<CodeAnalysisRules />
|
||||
<CodeAnalysisRuleAssemblies />
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Platform)'=='x64'">
|
||||
<OutDir>../../bin/win64/$(Configuration)/</OutDir>
|
||||
</PropertyGroup>
|
||||
<ItemDefinitionGroup>
|
||||
<ClCompile>
|
||||
<WarningLevel>Level3</WarningLevel>
|
||||
<PreprocessorDefinitions>WIN32;_MBCS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
|
||||
<AdditionalIncludeDirectories>./;$(CudaToolkitDir)/include;../../Common;</AdditionalIncludeDirectories>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<SubSystem>Console</SubSystem>
|
||||
<AdditionalDependencies>cudart_static.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
|
||||
<AdditionalLibraryDirectories>$(CudaToolkitLibDir);</AdditionalLibraryDirectories>
|
||||
<OutputFile>$(OutDir)/histogram.exe</OutputFile>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<CodeGeneration>compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;compute_80,sm_80;compute_86,sm_86;</CodeGeneration>
|
||||
<AdditionalOptions>-Xcompiler "/wd 4819" %(AdditionalOptions)</AdditionalOptions>
|
||||
<Include>./;../../Common</Include>
|
||||
<Defines>WIN32</Defines>
|
||||
<AdditionalOptions>--threads 0</AdditionalOptions>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(Configuration)'=='Debug'">
|
||||
<ClCompile>
|
||||
<Optimization>Disabled</Optimization>
|
||||
<RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<GenerateDebugInformation>true</GenerateDebugInformation>
|
||||
<LinkTimeCodeGeneration>Default</LinkTimeCodeGeneration>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<Runtime>MTd</Runtime>
|
||||
<TargetMachinePlatform>64</TargetMachinePlatform>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(Configuration)'=='Release'">
|
||||
<ClCompile>
|
||||
<Optimization>MaxSpeed</Optimization>
|
||||
<RuntimeLibrary>MultiThreaded</RuntimeLibrary>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<GenerateDebugInformation>false</GenerateDebugInformation>
|
||||
<LinkTimeCodeGeneration>UseLinkTimeCodeGeneration</LinkTimeCodeGeneration>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<Runtime>MT</Runtime>
|
||||
<TargetMachinePlatform>64</TargetMachinePlatform>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemGroup>
|
||||
<CudaCompile Include="histogram256.cu" />
|
||||
<CudaCompile Include="histogram64.cu" />
|
||||
<ClCompile Include="histogram_gold.cpp" />
|
||||
<ClCompile Include="main.cpp" />
|
||||
<ClInclude Include="histogram_common.h" />
|
||||
</ItemGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
|
||||
<ImportGroup Label="ExtensionTargets">
|
||||
<Import Project="$(CUDAPropsPath)\CUDA 11.5.targets" />
|
||||
</ImportGroup>
|
||||
</Project>
|
||||
20
Samples/histogram/histogram_vs2019.sln
Normal file
20
Samples/histogram/histogram_vs2019.sln
Normal file
@@ -0,0 +1,20 @@
|
||||
|
||||
Microsoft Visual Studio Solution File, Format Version 12.00
|
||||
# Visual Studio 2019
|
||||
Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "histogram", "histogram_vs2019.vcxproj", "{997E0757-EA74-4A4E-A0FC-47D8C8831A15}"
|
||||
EndProject
|
||||
Global
|
||||
GlobalSection(SolutionConfigurationPlatforms) = preSolution
|
||||
Debug|x64 = Debug|x64
|
||||
Release|x64 = Release|x64
|
||||
EndGlobalSection
|
||||
GlobalSection(ProjectConfigurationPlatforms) = postSolution
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Debug|x64.ActiveCfg = Debug|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Debug|x64.Build.0 = Debug|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Release|x64.ActiveCfg = Release|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Release|x64.Build.0 = Release|x64
|
||||
EndGlobalSection
|
||||
GlobalSection(SolutionProperties) = preSolution
|
||||
HideSolutionNode = FALSE
|
||||
EndGlobalSection
|
||||
EndGlobal
|
||||
112
Samples/histogram/histogram_vs2019.vcxproj
Normal file
112
Samples/histogram/histogram_vs2019.vcxproj
Normal file
@@ -0,0 +1,112 @@
|
||||
<?xml version="1.0" encoding="utf-8"?>
|
||||
<Project DefaultTargets="Build" ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
|
||||
<PropertyGroup>
|
||||
<CUDAPropsPath Condition="'$(CUDAPropsPath)'==''">$(VCTargetsPath)\BuildCustomizations</CUDAPropsPath>
|
||||
</PropertyGroup>
|
||||
<ItemGroup Label="ProjectConfigurations">
|
||||
<ProjectConfiguration Include="Debug|x64">
|
||||
<Configuration>Debug</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
<ProjectConfiguration Include="Release|x64">
|
||||
<Configuration>Release</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
</ItemGroup>
|
||||
<PropertyGroup Label="Globals">
|
||||
<ProjectGuid>{997E0757-EA74-4A4E-A0FC-47D8C8831A15}</ProjectGuid>
|
||||
<RootNamespace>histogram_vs2019</RootNamespace>
|
||||
<ProjectName>histogram</ProjectName>
|
||||
<CudaToolkitCustomDir />
|
||||
</PropertyGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
|
||||
<PropertyGroup>
|
||||
<ConfigurationType>Application</ConfigurationType>
|
||||
<CharacterSet>MultiByte</CharacterSet>
|
||||
<PlatformToolset>v142</PlatformToolset>
|
||||
<WindowsTargetPlatformVersion>10.0</WindowsTargetPlatformVersion>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)'=='Debug'">
|
||||
<UseDebugLibraries>true</UseDebugLibraries>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)'=='Release'">
|
||||
<WholeProgramOptimization>true</WholeProgramOptimization>
|
||||
</PropertyGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
|
||||
<ImportGroup Label="ExtensionSettings">
|
||||
<Import Project="$(CUDAPropsPath)\CUDA 11.5.props" />
|
||||
</ImportGroup>
|
||||
<ImportGroup Label="PropertySheets">
|
||||
<Import Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" />
|
||||
</ImportGroup>
|
||||
<PropertyGroup Label="UserMacros" />
|
||||
<PropertyGroup>
|
||||
<IntDir>$(Platform)/$(Configuration)/</IntDir>
|
||||
<IncludePath>$(IncludePath)</IncludePath>
|
||||
<CodeAnalysisRuleSet>AllRules.ruleset</CodeAnalysisRuleSet>
|
||||
<CodeAnalysisRules />
|
||||
<CodeAnalysisRuleAssemblies />
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Platform)'=='x64'">
|
||||
<OutDir>../../bin/win64/$(Configuration)/</OutDir>
|
||||
</PropertyGroup>
|
||||
<ItemDefinitionGroup>
|
||||
<ClCompile>
|
||||
<WarningLevel>Level3</WarningLevel>
|
||||
<PreprocessorDefinitions>WIN32;_MBCS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
|
||||
<AdditionalIncludeDirectories>./;$(CudaToolkitDir)/include;../../Common;</AdditionalIncludeDirectories>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<SubSystem>Console</SubSystem>
|
||||
<AdditionalDependencies>cudart_static.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
|
||||
<AdditionalLibraryDirectories>$(CudaToolkitLibDir);</AdditionalLibraryDirectories>
|
||||
<OutputFile>$(OutDir)/histogram.exe</OutputFile>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<CodeGeneration>compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;compute_80,sm_80;compute_86,sm_86;</CodeGeneration>
|
||||
<AdditionalOptions>-Xcompiler "/wd 4819" %(AdditionalOptions)</AdditionalOptions>
|
||||
<Include>./;../../Common</Include>
|
||||
<Defines>WIN32</Defines>
|
||||
<AdditionalOptions>--threads 0</AdditionalOptions>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(Configuration)'=='Debug'">
|
||||
<ClCompile>
|
||||
<Optimization>Disabled</Optimization>
|
||||
<RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<GenerateDebugInformation>true</GenerateDebugInformation>
|
||||
<LinkTimeCodeGeneration>Default</LinkTimeCodeGeneration>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<Runtime>MTd</Runtime>
|
||||
<TargetMachinePlatform>64</TargetMachinePlatform>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(Configuration)'=='Release'">
|
||||
<ClCompile>
|
||||
<Optimization>MaxSpeed</Optimization>
|
||||
<RuntimeLibrary>MultiThreaded</RuntimeLibrary>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<GenerateDebugInformation>false</GenerateDebugInformation>
|
||||
<LinkTimeCodeGeneration>UseLinkTimeCodeGeneration</LinkTimeCodeGeneration>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<Runtime>MT</Runtime>
|
||||
<TargetMachinePlatform>64</TargetMachinePlatform>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemGroup>
|
||||
<CudaCompile Include="histogram256.cu" />
|
||||
<CudaCompile Include="histogram64.cu" />
|
||||
<ClCompile Include="histogram_gold.cpp" />
|
||||
<ClCompile Include="main.cpp" />
|
||||
<ClInclude Include="histogram_common.h" />
|
||||
</ItemGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
|
||||
<ImportGroup Label="ExtensionTargets">
|
||||
<Import Project="$(CUDAPropsPath)\CUDA 11.5.targets" />
|
||||
</ImportGroup>
|
||||
</Project>
|
||||
230
Samples/histogram/main.cpp
Normal file
230
Samples/histogram/main.cpp
Normal file
@@ -0,0 +1,230 @@
|
||||
/* Copyright (c) 2021, 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 implements 64-bin histogram calculation
|
||||
* of arbitrary-sized 8-bit data array
|
||||
*/
|
||||
|
||||
// CUDA Runtime
|
||||
#include <cuda_runtime.h>
|
||||
|
||||
// Utility and system includes
|
||||
#include <helper_cuda.h>
|
||||
#include <helper_functions.h> // helper for shared that are common to CUDA Samples
|
||||
|
||||
// project include
|
||||
#include "histogram_common.h"
|
||||
|
||||
const int numRuns = 16;
|
||||
const static char *sSDKsample = "[histogram]\0";
|
||||
|
||||
int main(int argc, char **argv) {
|
||||
uchar *h_Data;
|
||||
uint *h_HistogramCPU, *h_HistogramGPU;
|
||||
uchar *d_Data;
|
||||
uint *d_Histogram;
|
||||
StopWatchInterface *hTimer = NULL;
|
||||
int PassFailFlag = 1;
|
||||
uint byteCount = 64 * 1048576;
|
||||
uint uiSizeMult = 1;
|
||||
|
||||
cudaDeviceProp deviceProp;
|
||||
deviceProp.major = 0;
|
||||
deviceProp.minor = 0;
|
||||
|
||||
// set logfile name and start logs
|
||||
printf("[%s] - Starting...\n", sSDKsample);
|
||||
|
||||
// Use command-line specified CUDA device, otherwise use device with highest
|
||||
// Gflops/s
|
||||
int dev = findCudaDevice(argc, (const char **)argv);
|
||||
|
||||
checkCudaErrors(cudaGetDeviceProperties(&deviceProp, dev));
|
||||
|
||||
printf("CUDA device [%s] has %d Multi-Processors, Compute %d.%d\n",
|
||||
deviceProp.name, deviceProp.multiProcessorCount, deviceProp.major,
|
||||
deviceProp.minor);
|
||||
|
||||
sdkCreateTimer(&hTimer);
|
||||
|
||||
// Optional Command-line multiplier to increase size of array to histogram
|
||||
if (checkCmdLineFlag(argc, (const char **)argv, "sizemult")) {
|
||||
uiSizeMult = getCmdLineArgumentInt(argc, (const char **)argv, "sizemult");
|
||||
uiSizeMult = MAX(1, MIN(uiSizeMult, 10));
|
||||
byteCount *= uiSizeMult;
|
||||
}
|
||||
|
||||
printf("Initializing data...\n");
|
||||
printf("...allocating CPU memory.\n");
|
||||
h_Data = (uchar *)malloc(byteCount);
|
||||
h_HistogramCPU = (uint *)malloc(HISTOGRAM256_BIN_COUNT * sizeof(uint));
|
||||
h_HistogramGPU = (uint *)malloc(HISTOGRAM256_BIN_COUNT * sizeof(uint));
|
||||
|
||||
printf("...generating input data\n");
|
||||
srand(2009);
|
||||
|
||||
for (uint i = 0; i < byteCount; i++) {
|
||||
h_Data[i] = rand() % 256;
|
||||
}
|
||||
|
||||
printf("...allocating GPU memory and copying input data\n\n");
|
||||
checkCudaErrors(cudaMalloc((void **)&d_Data, byteCount));
|
||||
checkCudaErrors(
|
||||
cudaMalloc((void **)&d_Histogram, HISTOGRAM256_BIN_COUNT * sizeof(uint)));
|
||||
checkCudaErrors(
|
||||
cudaMemcpy(d_Data, h_Data, byteCount, cudaMemcpyHostToDevice));
|
||||
|
||||
{
|
||||
printf("Starting up 64-bin histogram...\n\n");
|
||||
initHistogram64();
|
||||
|
||||
printf("Running 64-bin GPU histogram for %u bytes (%u runs)...\n\n",
|
||||
byteCount, numRuns);
|
||||
|
||||
for (int iter = -1; iter < numRuns; iter++) {
|
||||
// iter == -1 -- warmup iteration
|
||||
if (iter == 0) {
|
||||
cudaDeviceSynchronize();
|
||||
sdkResetTimer(&hTimer);
|
||||
sdkStartTimer(&hTimer);
|
||||
}
|
||||
|
||||
histogram64(d_Histogram, d_Data, byteCount);
|
||||
}
|
||||
|
||||
cudaDeviceSynchronize();
|
||||
sdkStopTimer(&hTimer);
|
||||
double dAvgSecs =
|
||||
1.0e-3 * (double)sdkGetTimerValue(&hTimer) / (double)numRuns;
|
||||
printf("histogram64() time (average) : %.5f sec, %.4f MB/sec\n\n", dAvgSecs,
|
||||
((double)byteCount * 1.0e-6) / dAvgSecs);
|
||||
printf(
|
||||
"histogram64, Throughput = %.4f MB/s, Time = %.5f s, Size = %u Bytes, "
|
||||
"NumDevsUsed = %u, Workgroup = %u\n",
|
||||
(1.0e-6 * (double)byteCount / dAvgSecs), dAvgSecs, byteCount, 1,
|
||||
HISTOGRAM64_THREADBLOCK_SIZE);
|
||||
|
||||
printf("\nValidating GPU results...\n");
|
||||
printf(" ...reading back GPU results\n");
|
||||
checkCudaErrors(cudaMemcpy(h_HistogramGPU, d_Histogram,
|
||||
HISTOGRAM64_BIN_COUNT * sizeof(uint),
|
||||
cudaMemcpyDeviceToHost));
|
||||
|
||||
printf(" ...histogram64CPU()\n");
|
||||
histogram64CPU(h_HistogramCPU, h_Data, byteCount);
|
||||
|
||||
printf(" ...comparing the results...\n");
|
||||
|
||||
for (uint i = 0; i < HISTOGRAM64_BIN_COUNT; i++)
|
||||
if (h_HistogramGPU[i] != h_HistogramCPU[i]) {
|
||||
PassFailFlag = 0;
|
||||
}
|
||||
|
||||
printf(PassFailFlag ? " ...64-bin histograms match\n\n"
|
||||
: " ***64-bin histograms do not match!!!***\n\n");
|
||||
|
||||
printf("Shutting down 64-bin histogram...\n\n\n");
|
||||
closeHistogram64();
|
||||
}
|
||||
|
||||
{
|
||||
printf("Initializing 256-bin histogram...\n");
|
||||
initHistogram256();
|
||||
|
||||
printf("Running 256-bin GPU histogram for %u bytes (%u runs)...\n\n",
|
||||
byteCount, numRuns);
|
||||
|
||||
for (int iter = -1; iter < numRuns; iter++) {
|
||||
// iter == -1 -- warmup iteration
|
||||
if (iter == 0) {
|
||||
checkCudaErrors(cudaDeviceSynchronize());
|
||||
sdkResetTimer(&hTimer);
|
||||
sdkStartTimer(&hTimer);
|
||||
}
|
||||
|
||||
histogram256(d_Histogram, d_Data, byteCount);
|
||||
}
|
||||
|
||||
cudaDeviceSynchronize();
|
||||
sdkStopTimer(&hTimer);
|
||||
double dAvgSecs =
|
||||
1.0e-3 * (double)sdkGetTimerValue(&hTimer) / (double)numRuns;
|
||||
printf("histogram256() time (average) : %.5f sec, %.4f MB/sec\n\n",
|
||||
dAvgSecs, ((double)byteCount * 1.0e-6) / dAvgSecs);
|
||||
printf(
|
||||
"histogram256, Throughput = %.4f MB/s, Time = %.5f s, Size = %u Bytes, "
|
||||
"NumDevsUsed = %u, Workgroup = %u\n",
|
||||
(1.0e-6 * (double)byteCount / dAvgSecs), dAvgSecs, byteCount, 1,
|
||||
HISTOGRAM256_THREADBLOCK_SIZE);
|
||||
|
||||
printf("\nValidating GPU results...\n");
|
||||
printf(" ...reading back GPU results\n");
|
||||
checkCudaErrors(cudaMemcpy(h_HistogramGPU, d_Histogram,
|
||||
HISTOGRAM256_BIN_COUNT * sizeof(uint),
|
||||
cudaMemcpyDeviceToHost));
|
||||
|
||||
printf(" ...histogram256CPU()\n");
|
||||
histogram256CPU(h_HistogramCPU, h_Data, byteCount);
|
||||
|
||||
printf(" ...comparing the results\n");
|
||||
|
||||
for (uint i = 0; i < HISTOGRAM256_BIN_COUNT; i++)
|
||||
if (h_HistogramGPU[i] != h_HistogramCPU[i]) {
|
||||
PassFailFlag = 0;
|
||||
}
|
||||
|
||||
printf(PassFailFlag ? " ...256-bin histograms match\n\n"
|
||||
: " ***256-bin histograms do not match!!!***\n\n");
|
||||
|
||||
printf("Shutting down 256-bin histogram...\n\n\n");
|
||||
closeHistogram256();
|
||||
}
|
||||
|
||||
printf("Shutting down...\n");
|
||||
sdkDeleteTimer(&hTimer);
|
||||
checkCudaErrors(cudaFree(d_Histogram));
|
||||
checkCudaErrors(cudaFree(d_Data));
|
||||
free(h_HistogramGPU);
|
||||
free(h_HistogramCPU);
|
||||
free(h_Data);
|
||||
|
||||
printf(
|
||||
"\nNOTE: The CUDA Samples are not meant for performance measurements. "
|
||||
"Results may vary when GPU Boost is enabled.\n\n");
|
||||
|
||||
printf("%s - Test Summary\n", sSDKsample);
|
||||
|
||||
// pass or fail (for both 64 bit and 256 bit histograms)
|
||||
if (!PassFailFlag) {
|
||||
printf("Test failed!\n");
|
||||
exit(EXIT_FAILURE);
|
||||
}
|
||||
|
||||
printf("Test passed\n");
|
||||
exit(EXIT_SUCCESS);
|
||||
}
|
||||
Reference in New Issue
Block a user