mirror of
https://github.com/NVIDIA/cuda-samples.git
synced 2026-10-23 21:28:17 +08:00
add and update samples for CUDA 11.6
This commit is contained in:
Rutwik Choughule
18
Samples/5_Domain_Specific/SobolQRNG/.vscode/c_cpp_properties.json
vendored
Normal file
18
Samples/5_Domain_Specific/SobolQRNG/.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/5_Domain_Specific/SobolQRNG/.vscode/extensions.json
vendored
Normal file
7
Samples/5_Domain_Specific/SobolQRNG/.vscode/extensions.json
vendored
Normal file
@@ -0,0 +1,7 @@
|
||||
{
|
||||
"recommendations": [
|
||||
"nvidia.nsight-vscode-edition",
|
||||
"ms-vscode.cpptools",
|
||||
"ms-vscode.makefile-tools"
|
||||
]
|
||||
}
|
||||
10
Samples/5_Domain_Specific/SobolQRNG/.vscode/launch.json
vendored
Normal file
10
Samples/5_Domain_Specific/SobolQRNG/.vscode/launch.json
vendored
Normal file
@@ -0,0 +1,10 @@
|
||||
{
|
||||
"configurations": [
|
||||
{
|
||||
"name": "CUDA C++: Launch",
|
||||
"type": "cuda-gdb",
|
||||
"request": "launch",
|
||||
"program": "${workspaceFolder}/SobolQRNG"
|
||||
}
|
||||
]
|
||||
}
|
||||
15
Samples/5_Domain_Specific/SobolQRNG/.vscode/tasks.json
vendored
Normal file
15
Samples/5_Domain_Specific/SobolQRNG/.vscode/tasks.json
vendored
Normal file
@@ -0,0 +1,15 @@
|
||||
{
|
||||
"version": "2.0.0",
|
||||
"tasks": [
|
||||
{
|
||||
"label": "sample",
|
||||
"type": "shell",
|
||||
"command": "make dbg=1",
|
||||
"problemMatcher": ["$nvcc"],
|
||||
"group": {
|
||||
"kind": "build",
|
||||
"isDefault": true
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
349
Samples/5_Domain_Specific/SobolQRNG/Makefile
Normal file
349
Samples/5_Domain_Specific/SobolQRNG/Makefile
Normal file
@@ -0,0 +1,349 @@
|
||||
################################################################################
|
||||
# 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 $(TARGET_OS) for TARGET_OS!)
|
||||
endif
|
||||
|
||||
# host compiler
|
||||
ifeq ($(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 += -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 $(TARGET_ARCH),armv7l aarch64 sbsa))
|
||||
SMS ?= 53 61 70 72 75 80 86 87
|
||||
else
|
||||
SMS ?= 35 37 50 52 60 61 70 75 80 86
|
||||
endif
|
||||
|
||||
ifeq ($(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 ($(SAMPLE_ENABLED),0)
|
||||
EXEC ?= @echo "[@]"
|
||||
endif
|
||||
|
||||
################################################################################
|
||||
|
||||
# Target rules
|
||||
all: build
|
||||
|
||||
build: SobolQRNG
|
||||
|
||||
check.deps:
|
||||
ifeq ($(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
|
||||
|
||||
sobol.o:sobol.cpp
|
||||
$(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $<
|
||||
|
||||
sobol_gold.o:sobol_gold.cpp
|
||||
$(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $<
|
||||
|
||||
sobol_gpu.o:sobol_gpu.cu
|
||||
$(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $<
|
||||
|
||||
sobol_primitives.o:sobol_primitives.cpp
|
||||
$(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $<
|
||||
|
||||
SobolQRNG: sobol.o sobol_gold.o sobol_gpu.o sobol_primitives.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
|
||||
$(EXEC) ./SobolQRNG
|
||||
|
||||
testrun: build
|
||||
|
||||
clean:
|
||||
rm -f SobolQRNG sobol.o sobol_gold.o sobol_gpu.o sobol_primitives.o
|
||||
rm -rf ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE)/SobolQRNG
|
||||
|
||||
clobber: clean
|
||||
81
Samples/5_Domain_Specific/SobolQRNG/NsightEclipse.xml
Normal file
81
Samples/5_Domain_Specific/SobolQRNG/NsightEclipse.xml
Normal file
@@ -0,0 +1,81 @@
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<!DOCTYPE entry SYSTEM "SamplesInfo.dtd">
|
||||
<entry>
|
||||
<name>SobolQRNG</name>
|
||||
<cuda_api_list>
|
||||
<toolkit>cudaFree</toolkit>
|
||||
<toolkit>cudaDeviceSynchronize</toolkit>
|
||||
<toolkit>cudaMalloc</toolkit>
|
||||
<toolkit>cudaMemcpy</toolkit>
|
||||
<toolkit>cudaGetErrorString</toolkit>
|
||||
<toolkit>cudaGetDeviceProperties</toolkit>
|
||||
<toolkit>cudaGetDevice</toolkit>
|
||||
</cuda_api_list>
|
||||
<description><![CDATA[This sample implements Sobol Quasirandom Sequence Generator.]]></description>
|
||||
<devicecompilation>whole</devicecompilation>
|
||||
<includepaths>
|
||||
<path>./</path>
|
||||
<path>../</path>
|
||||
<path>../../../Common</path>
|
||||
</includepaths>
|
||||
<keyconcepts>
|
||||
<concept level="advanced">Computational Finance</concept>
|
||||
</keyconcepts>
|
||||
<keywords>
|
||||
<keyword>CUDA</keyword>
|
||||
<keyword>Advanced Topics</keyword>
|
||||
<keyword>Quasi-Monte Carlo Integration</keyword>
|
||||
<keyword>Computational Finance</keyword>
|
||||
</keywords>
|
||||
<libraries>
|
||||
</libraries>
|
||||
<librarypaths>
|
||||
</librarypaths>
|
||||
<nsight_eclipse>true</nsight_eclipse>
|
||||
<primary_file>sobel.cpp</primary_file>
|
||||
<scopes>
|
||||
<scope>1:CUDA Advanced Topics</scope>
|
||||
<scope>3:Computational Finance</scope>
|
||||
</scopes>
|
||||
<sm-arch>sm35</sm-arch>
|
||||
<sm-arch>sm37</sm-arch>
|
||||
<sm-arch>sm50</sm-arch>
|
||||
<sm-arch>sm52</sm-arch>
|
||||
<sm-arch>sm53</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>
|
||||
<sm-arch>sm87</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>sbsa</arch>
|
||||
</env>
|
||||
<env>
|
||||
<arch>ppc64le</arch>
|
||||
<platform>linux</platform>
|
||||
</env>
|
||||
</supported_envs>
|
||||
<supported_sm_architectures>
|
||||
<include>all</include>
|
||||
</supported_sm_architectures>
|
||||
<title>Sobol Quasirandom Number Generator</title>
|
||||
<type>exe</type>
|
||||
</entry>
|
||||
70
Samples/5_Domain_Specific/SobolQRNG/README.md
Normal file
70
Samples/5_Domain_Specific/SobolQRNG/README.md
Normal file
@@ -0,0 +1,70 @@
|
||||
# SobolQRNG - Sobol Quasirandom Number Generator
|
||||
|
||||
## Description
|
||||
|
||||
This sample implements Sobol Quasirandom Sequence Generator.
|
||||
|
||||
## Key Concepts
|
||||
|
||||
Computational Finance
|
||||
|
||||
## 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 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)
|
||||
|
||||
## Supported OSes
|
||||
|
||||
Linux, Windows
|
||||
|
||||
## Supported CPU Architecture
|
||||
|
||||
x86_64, ppc64le, armv7l
|
||||
|
||||
## CUDA APIs involved
|
||||
|
||||
### [CUDA Runtime API](http://docs.nvidia.com/cuda/cuda-runtime-api/index.html)
|
||||
cudaFree, cudaDeviceSynchronize, cudaMalloc, cudaMemcpy, cudaGetErrorString, cudaGetDeviceProperties, cudaGetDevice
|
||||
|
||||
## Prerequisites
|
||||
|
||||
Download and install the [CUDA Toolkit 11.6](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)
|
||||
|
||||
20
Samples/5_Domain_Specific/SobolQRNG/SobolQRNG_vs2017.sln
Normal file
20
Samples/5_Domain_Specific/SobolQRNG/SobolQRNG_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}") = "SobolQRNG", "SobolQRNG_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
|
||||
118
Samples/5_Domain_Specific/SobolQRNG/SobolQRNG_vs2017.vcxproj
Normal file
118
Samples/5_Domain_Specific/SobolQRNG/SobolQRNG_vs2017.vcxproj
Normal file
@@ -0,0 +1,118 @@
|
||||
<?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>SobolQRNG_vs2017</RootNamespace>
|
||||
<ProjectName>SobolQRNG</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 11.6.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)/SobolQRNG.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" --threads 0 </AdditionalOptions>
|
||||
<Include>./;../../../Common</Include>
|
||||
<Defines>WIN32</Defines>
|
||||
</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>
|
||||
<ClCompile Include="sobol.cpp" />
|
||||
<ClCompile Include="sobol_gold.cpp" />
|
||||
<CudaCompile Include="sobol_gpu.cu" />
|
||||
<ClCompile Include="sobol_primitives.cpp" />
|
||||
<ClInclude Include="sobol.h" />
|
||||
<ClInclude Include="sobol_gold.h" />
|
||||
<ClInclude Include="sobol_gpu.h" />
|
||||
<ClInclude Include="sobol_primitives.h" />
|
||||
</ItemGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
|
||||
<ImportGroup Label="ExtensionTargets">
|
||||
<Import Project="$(CUDAPropsPath)\CUDA 11.6.targets" />
|
||||
</ImportGroup>
|
||||
</Project>
|
||||
20
Samples/5_Domain_Specific/SobolQRNG/SobolQRNG_vs2019.sln
Normal file
20
Samples/5_Domain_Specific/SobolQRNG/SobolQRNG_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}") = "SobolQRNG", "SobolQRNG_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
|
||||
114
Samples/5_Domain_Specific/SobolQRNG/SobolQRNG_vs2019.vcxproj
Normal file
114
Samples/5_Domain_Specific/SobolQRNG/SobolQRNG_vs2019.vcxproj
Normal file
@@ -0,0 +1,114 @@
|
||||
<?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>SobolQRNG_vs2019</RootNamespace>
|
||||
<ProjectName>SobolQRNG</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="$(CUDAPropsPath)\CUDA 11.6.props" />
|
||||
</ImportGroup>
|
||||
<ImportGroup Label="PropertySheets">
|
||||
<Import Condition="exists('$(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>$(OutDir)/SobolQRNG.exe</OutputFile>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<CodeGeneration>compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;compute_80,sm_80;compute_86,sm_86;</CodeGeneration>
|
||||
<AdditionalOptions>-Xcompiler "/wd 4819" --threads 0 </AdditionalOptions>
|
||||
<Include>./;../../../Common</Include>
|
||||
<Defines>WIN32</Defines>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(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="'$(Configuration)'=='Release'">
|
||||
<ClCompile>
|
||||
<Optimization>MaxSpeed</Optimization>
|
||||
<RuntimeLibrary>MultiThreaded</RuntimeLibrary>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<GenerateDebugInformation>false</GenerateDebugInformation>
|
||||
<LinkTimeCodeGeneration>UseLinkTimeCodeGeneration</LinkTimeCodeGeneration>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<Runtime>MT</Runtime>
|
||||
<TargetMachinePlatform>64</TargetMachinePlatform>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemGroup>
|
||||
<ClCompile Include="sobol.cpp" />
|
||||
<ClCompile Include="sobol_gold.cpp" />
|
||||
<CudaCompile Include="sobol_gpu.cu" />
|
||||
<ClCompile Include="sobol_primitives.cpp" />
|
||||
<ClInclude Include="sobol.h" />
|
||||
<ClInclude Include="sobol_gold.h" />
|
||||
<ClInclude Include="sobol_gpu.h" />
|
||||
<ClInclude Include="sobol_primitives.h" />
|
||||
</ItemGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
|
||||
<ImportGroup Label="ExtensionTargets">
|
||||
<Import Project="$(CUDAPropsPath)\CUDA 11.6.targets" />
|
||||
</ImportGroup>
|
||||
</Project>
|
||||
20
Samples/5_Domain_Specific/SobolQRNG/SobolQRNG_vs2022.sln
Normal file
20
Samples/5_Domain_Specific/SobolQRNG/SobolQRNG_vs2022.sln
Normal file
@@ -0,0 +1,20 @@
|
||||
|
||||
Microsoft Visual Studio Solution File, Format Version 12.00
|
||||
# Visual Studio 2022
|
||||
Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "SobolQRNG", "SobolQRNG_vs2022.vcxproj", "{997E0757-EA74-4A4E-A0FC-47D8C8831A15}"
|
||||
EndProject
|
||||
Global
|
||||
GlobalSection(SolutionConfigurationPlatforms) = preSolution
|
||||
Debug|x64 = Debug|x64
|
||||
Release|x64 = Release|x64
|
||||
EndGlobalSection
|
||||
GlobalSection(ProjectConfigurationPlatforms) = postSolution
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Debug|x64.ActiveCfg = Debug|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Debug|x64.Build.0 = Debug|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Release|x64.ActiveCfg = Release|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Release|x64.Build.0 = Release|x64
|
||||
EndGlobalSection
|
||||
GlobalSection(SolutionProperties) = preSolution
|
||||
HideSolutionNode = FALSE
|
||||
EndGlobalSection
|
||||
EndGlobal
|
||||
114
Samples/5_Domain_Specific/SobolQRNG/SobolQRNG_vs2022.vcxproj
Normal file
114
Samples/5_Domain_Specific/SobolQRNG/SobolQRNG_vs2022.vcxproj
Normal file
@@ -0,0 +1,114 @@
|
||||
<?xml version="1.0" encoding="utf-8"?>
|
||||
<Project DefaultTargets="Build" ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
|
||||
<PropertyGroup>
|
||||
<CUDAPropsPath Condition="'$(CUDAPropsPath)'==''">$(VCTargetsPath)\BuildCustomizations</CUDAPropsPath>
|
||||
</PropertyGroup>
|
||||
<ItemGroup Label="ProjectConfigurations">
|
||||
<ProjectConfiguration Include="Debug|x64">
|
||||
<Configuration>Debug</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
<ProjectConfiguration Include="Release|x64">
|
||||
<Configuration>Release</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
</ItemGroup>
|
||||
<PropertyGroup Label="Globals">
|
||||
<ProjectGuid>{997E0757-EA74-4A4E-A0FC-47D8C8831A15}</ProjectGuid>
|
||||
<RootNamespace>SobolQRNG_vs2022</RootNamespace>
|
||||
<ProjectName>SobolQRNG</ProjectName>
|
||||
<CudaToolkitCustomDir />
|
||||
</PropertyGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
|
||||
<PropertyGroup>
|
||||
<ConfigurationType>Application</ConfigurationType>
|
||||
<CharacterSet>MultiByte</CharacterSet>
|
||||
<PlatformToolset>v143</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 11.6.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)/SobolQRNG.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" --threads 0 </AdditionalOptions>
|
||||
<Include>./;../../../Common</Include>
|
||||
<Defines>WIN32</Defines>
|
||||
</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>
|
||||
<ClCompile Include="sobol.cpp" />
|
||||
<ClCompile Include="sobol_gold.cpp" />
|
||||
<CudaCompile Include="sobol_gpu.cu" />
|
||||
<ClCompile Include="sobol_primitives.cpp" />
|
||||
<ClInclude Include="sobol.h" />
|
||||
<ClInclude Include="sobol_gold.h" />
|
||||
<ClInclude Include="sobol_gpu.h" />
|
||||
<ClInclude Include="sobol_primitives.h" />
|
||||
</ItemGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
|
||||
<ImportGroup Label="ExtensionTargets">
|
||||
<Import Project="$(CUDAPropsPath)\CUDA 11.6.targets" />
|
||||
</ImportGroup>
|
||||
</Project>
|
||||
305
Samples/5_Domain_Specific/SobolQRNG/sobol.cpp
Normal file
305
Samples/5_Domain_Specific/SobolQRNG/sobol.cpp
Normal file
@@ -0,0 +1,305 @@
|
||||
/* 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.
|
||||
*/
|
||||
|
||||
/*
|
||||
* Portions Copyright (c) 2009 Mike Giles, Oxford University. All rights
|
||||
* reserved.
|
||||
* Portions Copyright (c) 2008 Frances Y. Kuo and Stephen Joe. All rights
|
||||
* reserved.
|
||||
*
|
||||
* Sobol Quasi-random Number Generator example
|
||||
*
|
||||
* Based on CUDA code submitted by Mike Giles, Oxford University, United Kingdom
|
||||
* http://people.maths.ox.ac.uk/~gilesm/
|
||||
*
|
||||
* and C code developed by Stephen Joe, University of Waikato, New Zealand
|
||||
* and Frances Kuo, University of New South Wales, Australia
|
||||
* http://web.maths.unsw.edu.au/~fkuo/sobol/
|
||||
*
|
||||
* For theoretical background see:
|
||||
*
|
||||
* P. Bratley and B.L. Fox.
|
||||
* Implementing Sobol's quasirandom sequence generator
|
||||
* http://portal.acm.org/citation.cfm?id=42288
|
||||
* ACM Trans. on Math. Software, 14(1):88-100, 1988
|
||||
*
|
||||
* S. Joe and F. Kuo.
|
||||
* Remark on algorithm 659: implementing Sobol's quasirandom sequence generator.
|
||||
* http://portal.acm.org/citation.cfm?id=641879
|
||||
* ACM Trans. on Math. Software, 29(1):49-57, 2003
|
||||
*/
|
||||
|
||||
#include <iostream>
|
||||
|
||||
#include <cuda_runtime.h> // CUDA Runtime Functions
|
||||
#include <helper_cuda.h> // helper functions for CUDA error checking and initialization
|
||||
#include <helper_functions.h> // helper functions
|
||||
|
||||
#include <stdexcept>
|
||||
#include <math.h>
|
||||
|
||||
#include "sobol.h"
|
||||
#include "sobol_gold.h"
|
||||
#include "sobol_gpu.h"
|
||||
|
||||
#define L1ERROR_TOLERANCE (1e-6)
|
||||
|
||||
const char *sSDKsample = "Sobol Quasi-Random Number Generator";
|
||||
|
||||
void printHelp(int argc, char *argv[]) {
|
||||
if (argc > 0) {
|
||||
std::cout << "\nUsage: " << argv[0] << " <options>\n\n";
|
||||
} else {
|
||||
std::cout << "\nUsage: <program name> <options>\n\n";
|
||||
}
|
||||
|
||||
std::cout << "\t--vectors=M specify number of vectors (required)\n";
|
||||
std::cout << "\t The generator will output M vectors\n\n";
|
||||
std::cout << "\t--dimensions=N specify number of dimensions (required)\n";
|
||||
std::cout << "\t Each vector will consist of N components\n\n";
|
||||
std::cout << std::endl;
|
||||
}
|
||||
|
||||
int main(int argc, char *argv[]) {
|
||||
bool ok = true;
|
||||
|
||||
// We will generate n_vectors vectors of n_dimensions numbers
|
||||
int n_vectors = 100000;
|
||||
int n_dimensions = 100;
|
||||
|
||||
printf("%s Starting...\n\n", sSDKsample);
|
||||
|
||||
// Print help if requested
|
||||
if (checkCmdLineFlag(argc, (const char **)argv, "help")) {
|
||||
printHelp(argc, argv);
|
||||
return 0;
|
||||
}
|
||||
|
||||
if (checkCmdLineFlag(argc, (const char **)argv, "qatest")) {
|
||||
// For QA testing set a default number of vectors and dimensions
|
||||
n_vectors = 100000;
|
||||
n_dimensions = 100;
|
||||
} else {
|
||||
// Parse the command line to determine the required number of vectors
|
||||
if (checkCmdLineFlag(argc, (const char **)argv, "vectors")) {
|
||||
n_vectors = getCmdLineArgumentInt(argc, (const char **)argv, "vectors");
|
||||
|
||||
if (n_vectors < 1) {
|
||||
std::cerr << "Illegal argument: number of vectors must be positive "
|
||||
"(--vectors=N)"
|
||||
<< std::endl;
|
||||
ok = false;
|
||||
}
|
||||
}
|
||||
|
||||
std::cout << "> number of vectors = " << n_vectors << std::endl;
|
||||
|
||||
// Parse the command line to determine the number of dimensions in each
|
||||
// vector
|
||||
if (checkCmdLineFlag(argc, (const char **)argv, "dimensions")) {
|
||||
n_dimensions =
|
||||
getCmdLineArgumentInt(argc, (const char **)argv, "dimensions");
|
||||
|
||||
if (n_dimensions < 1) {
|
||||
std::cerr << "Illegal argument: number of dimensions must be positive "
|
||||
"(--dimensions=N)"
|
||||
<< std::endl;
|
||||
ok = false;
|
||||
}
|
||||
}
|
||||
|
||||
std::cout << "> number of dimensions = " << n_dimensions << std::endl;
|
||||
}
|
||||
|
||||
// If any of the command line checks failed, exit
|
||||
if (!ok) {
|
||||
return -1;
|
||||
}
|
||||
|
||||
// Use command-line specified CUDA device, otherwise use device with highest
|
||||
// Gflops/s
|
||||
findCudaDevice(argc, (const char **)argv);
|
||||
|
||||
// Create a timer to measure performance
|
||||
StopWatchInterface *hTimer = NULL;
|
||||
double time;
|
||||
sdkCreateTimer(&hTimer);
|
||||
|
||||
// Allocate memory for the arrays
|
||||
std::cout << "Allocating CPU memory..." << std::endl;
|
||||
unsigned int *h_directions = 0;
|
||||
float *h_outputCPU = 0;
|
||||
float *h_outputGPU = 0;
|
||||
|
||||
try {
|
||||
h_directions = new unsigned int[n_dimensions * n_directions];
|
||||
h_outputCPU = new float[n_vectors * n_dimensions];
|
||||
h_outputGPU = new float[n_vectors * n_dimensions];
|
||||
} catch (std::exception e) {
|
||||
std::cerr << "Caught exception: " << e.what() << std::endl;
|
||||
std::cerr << "Unable to allocate CPU memory (try running with fewer "
|
||||
"vectors/dimensions)"
|
||||
<< std::endl;
|
||||
exit(EXIT_FAILURE);
|
||||
}
|
||||
|
||||
std::cout << "Allocating GPU memory..." << std::endl;
|
||||
unsigned int *d_directions;
|
||||
float *d_output;
|
||||
|
||||
try {
|
||||
cudaError_t cudaResult;
|
||||
cudaResult = cudaMalloc((void **)&d_directions,
|
||||
n_dimensions * n_directions * sizeof(unsigned int));
|
||||
|
||||
if (cudaResult != cudaSuccess) {
|
||||
throw std::runtime_error(cudaGetErrorString(cudaResult));
|
||||
}
|
||||
|
||||
cudaResult = cudaMalloc((void **)&d_output,
|
||||
n_vectors * n_dimensions * sizeof(float));
|
||||
|
||||
if (cudaResult != cudaSuccess) {
|
||||
throw std::runtime_error(cudaGetErrorString(cudaResult));
|
||||
}
|
||||
} catch (std::runtime_error e) {
|
||||
std::cerr << "Caught exception: " << e.what() << std::endl;
|
||||
std::cerr << "Unable to allocate GPU memory (try running with fewer "
|
||||
"vectors/dimensions)"
|
||||
<< std::endl;
|
||||
exit(EXIT_FAILURE);
|
||||
}
|
||||
|
||||
// Initialize the direction numbers (done on the host)
|
||||
std::cout << "Initializing direction numbers..." << std::endl;
|
||||
initSobolDirectionVectors(n_dimensions, h_directions);
|
||||
|
||||
// Copy the direction numbers to the device
|
||||
std::cout << "Copying direction numbers to device..." << std::endl;
|
||||
checkCudaErrors(cudaMemcpy(d_directions, h_directions,
|
||||
n_dimensions * n_directions * sizeof(unsigned int),
|
||||
cudaMemcpyHostToDevice));
|
||||
checkCudaErrors(cudaDeviceSynchronize());
|
||||
|
||||
// Execute the QRNG on the device
|
||||
std::cout << "Executing QRNG on GPU..." << std::endl;
|
||||
sdkResetTimer(&hTimer);
|
||||
sdkStartTimer(&hTimer);
|
||||
sobolGPU(n_vectors, n_dimensions, d_directions, d_output);
|
||||
checkCudaErrors(cudaDeviceSynchronize());
|
||||
sdkStopTimer(&hTimer);
|
||||
time = sdkGetTimerValue(&hTimer);
|
||||
|
||||
if (time < 1e-6) {
|
||||
std::cout << "Gsamples/s: problem size too small to measure, try "
|
||||
"increasing number of vectors or dimensions"
|
||||
<< std::endl;
|
||||
} else {
|
||||
std::cout << "Gsamples/s: "
|
||||
<< (double)n_vectors * (double)n_dimensions * 1E-9 / (time * 1E-3)
|
||||
<< std::endl;
|
||||
}
|
||||
|
||||
std::cout << "Reading results from GPU..." << std::endl;
|
||||
checkCudaErrors(cudaMemcpy(h_outputGPU, d_output,
|
||||
n_vectors * n_dimensions * sizeof(float),
|
||||
cudaMemcpyDeviceToHost));
|
||||
|
||||
std::cout << std::endl;
|
||||
// Execute the QRNG on the host
|
||||
std::cout << "Executing QRNG on CPU..." << std::endl;
|
||||
sdkResetTimer(&hTimer);
|
||||
sdkStartTimer(&hTimer);
|
||||
sobolCPU(n_vectors, n_dimensions, h_directions, h_outputCPU);
|
||||
sdkStopTimer(&hTimer);
|
||||
time = sdkGetTimerValue(&hTimer);
|
||||
|
||||
if (time < 1e-6) {
|
||||
std::cout << "Gsamples/s: problem size too small to measure, try "
|
||||
"increasing number of vectors or dimensions"
|
||||
<< std::endl;
|
||||
} else {
|
||||
std::cout << "Gsamples/s: "
|
||||
<< (double)n_vectors * (double)n_dimensions * 1E-9 / (time * 1E-3)
|
||||
<< std::endl;
|
||||
}
|
||||
|
||||
// Check the results
|
||||
std::cout << "Checking results..." << std::endl;
|
||||
float l1norm_diff = 0.0F;
|
||||
float l1norm_ref = 0.0F;
|
||||
float l1error;
|
||||
|
||||
// Special case if n_vectors is 1, when the vector should be exactly 0
|
||||
if (n_vectors == 1) {
|
||||
for (int d = 0, v = 0; d < n_dimensions; d++) {
|
||||
float ref = h_outputCPU[d * n_vectors + v];
|
||||
l1norm_diff += fabs(h_outputGPU[d * n_vectors + v] - ref);
|
||||
l1norm_ref += fabs(ref);
|
||||
}
|
||||
|
||||
// Output the L1-Error
|
||||
l1error = l1norm_diff;
|
||||
|
||||
if (l1norm_ref != 0) {
|
||||
std::cerr << "Error: L1-Norm of the reference is not zero (for single "
|
||||
"vector), golden generator appears broken\n";
|
||||
} else {
|
||||
std::cout << "L1-Error: " << l1error << std::endl;
|
||||
}
|
||||
} else {
|
||||
for (int d = 0; d < n_dimensions; d++) {
|
||||
for (int v = 0; v < n_vectors; v++) {
|
||||
float ref = h_outputCPU[d * n_vectors + v];
|
||||
l1norm_diff += fabs(h_outputGPU[d * n_vectors + v] - ref);
|
||||
l1norm_ref += fabs(ref);
|
||||
}
|
||||
}
|
||||
|
||||
// Output the L1-Error
|
||||
l1error = l1norm_diff / l1norm_ref;
|
||||
|
||||
if (l1norm_ref == 0) {
|
||||
std::cerr << "Error: L1-Norm of the reference is zero, golden generator "
|
||||
"appears broken\n";
|
||||
} else {
|
||||
std::cout << "L1-Error: " << l1error << std::endl;
|
||||
}
|
||||
}
|
||||
|
||||
// Cleanup and terminate
|
||||
std::cout << "Shutting down..." << std::endl;
|
||||
sdkDeleteTimer(&hTimer);
|
||||
delete h_directions;
|
||||
delete h_outputCPU;
|
||||
delete h_outputGPU;
|
||||
checkCudaErrors(cudaFree(d_directions));
|
||||
checkCudaErrors(cudaFree(d_output));
|
||||
|
||||
// Check pass/fail using L1 error
|
||||
exit(l1error < L1ERROR_TOLERANCE ? EXIT_SUCCESS : EXIT_FAILURE);
|
||||
}
|
||||
62
Samples/5_Domain_Specific/SobolQRNG/sobol.h
Normal file
62
Samples/5_Domain_Specific/SobolQRNG/sobol.h
Normal file
@@ -0,0 +1,62 @@
|
||||
/* 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.
|
||||
*/
|
||||
|
||||
/*
|
||||
* Portions Copyright (c) 2009 Mike Giles, Oxford University. All rights
|
||||
* reserved.
|
||||
* Portions Copyright (c) 2008 Frances Y. Kuo and Stephen Joe. All rights
|
||||
* reserved.
|
||||
*
|
||||
* Sobol Quasi-random Number Generator example
|
||||
*
|
||||
* Based on CUDA code submitted by Mike Giles, Oxford University, United Kingdom
|
||||
* http://people.maths.ox.ac.uk/~gilesm/
|
||||
*
|
||||
* and C code developed by Stephen Joe, University of Waikato, New Zealand
|
||||
* and Frances Kuo, University of New South Wales, Australia
|
||||
* http://web.maths.unsw.edu.au/~fkuo/sobol/
|
||||
*
|
||||
* For theoretical background see:
|
||||
*
|
||||
* P. Bratley and B.L. Fox.
|
||||
* Implementing Sobol's quasirandom sequence generator
|
||||
* http://portal.acm.org/citation.cfm?id=42288
|
||||
* ACM Trans. on Math. Software, 14(1):88-100, 1988
|
||||
*
|
||||
* S. Joe and F. Kuo.
|
||||
* Remark on algorithm 659: implementing Sobol's quasirandom sequence generator.
|
||||
* http://portal.acm.org/citation.cfm?id=641879
|
||||
* ACM Trans. on Math. Software, 29(1):49-57, 2003
|
||||
*/
|
||||
|
||||
#ifndef SOBOL_H
|
||||
#define SOBOL_H
|
||||
|
||||
// Number of direction vectors is fixed to 32
|
||||
#define n_directions 32
|
||||
|
||||
#endif
|
||||
173
Samples/5_Domain_Specific/SobolQRNG/sobol_gold.cpp
Normal file
173
Samples/5_Domain_Specific/SobolQRNG/sobol_gold.cpp
Normal file
@@ -0,0 +1,173 @@
|
||||
/* 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.
|
||||
*/
|
||||
|
||||
/*
|
||||
* Portions Copyright (c) 2009 Mike Giles, Oxford University. All rights
|
||||
* reserved.
|
||||
* Portions Copyright (c) 2008 Frances Y. Kuo and Stephen Joe. All rights
|
||||
* reserved.
|
||||
*
|
||||
* Sobol Quasi-random Number Generator example
|
||||
*
|
||||
* Based on CUDA code submitted by Mike Giles, Oxford University, United Kingdom
|
||||
* http://people.maths.ox.ac.uk/~gilesm/
|
||||
*
|
||||
* and C code developed by Stephen Joe, University of Waikato, New Zealand
|
||||
* and Frances Kuo, University of New South Wales, Australia
|
||||
* http://web.maths.unsw.edu.au/~fkuo/sobol/
|
||||
*
|
||||
* For theoretical background see:
|
||||
*
|
||||
* P. Bratley and B.L. Fox.
|
||||
* Implementing Sobol's quasirandom sequence generator
|
||||
* http://portal.acm.org/citation.cfm?id=42288
|
||||
* ACM Trans. on Math. Software, 14(1):88-100, 1988
|
||||
*
|
||||
* S. Joe and F. Kuo.
|
||||
* Remark on algorithm 659: implementing Sobol's quasirandom sequence generator.
|
||||
* http://portal.acm.org/citation.cfm?id=641879
|
||||
* ACM Trans. on Math. Software, 29(1):49-57, 2003
|
||||
*/
|
||||
|
||||
#include <stdio.h>
|
||||
#include <stdlib.h>
|
||||
#include <math.h>
|
||||
#include <string.h>
|
||||
|
||||
#include "sobol.h"
|
||||
#include "sobol_gold.h"
|
||||
#include "sobol_primitives.h"
|
||||
|
||||
#define k_2powneg32 2.3283064E-10F
|
||||
|
||||
// Windows does not provide ffs (find first set) so here is a
|
||||
// fairly simple implementation.
|
||||
// WIN32 is defined on 32 and 64 bit Windows
|
||||
#if defined(WIN32) || defined(_WIN32) || defined(WIN64) || defined(_WIN64)
|
||||
int ffs(const unsigned int &i) {
|
||||
unsigned int v = i;
|
||||
unsigned int count;
|
||||
|
||||
if (!v) {
|
||||
count = 0;
|
||||
} else {
|
||||
count = 2;
|
||||
|
||||
if ((v & 0xffff) == 0) {
|
||||
v >>= 16;
|
||||
count += 16;
|
||||
}
|
||||
|
||||
if ((v & 0xff) == 0) {
|
||||
v >>= 8;
|
||||
count += 8;
|
||||
}
|
||||
|
||||
if ((v & 0xf) == 0) {
|
||||
v >>= 4;
|
||||
count += 4;
|
||||
}
|
||||
|
||||
if ((v & 0x3) == 0) {
|
||||
v >>= 2;
|
||||
count += 2;
|
||||
}
|
||||
|
||||
count -= v & 0x1;
|
||||
}
|
||||
|
||||
return count;
|
||||
}
|
||||
#endif
|
||||
|
||||
// Create the direction numbers, based on the primitive polynomials.
|
||||
void initSobolDirectionVectors(int n_dimensions, unsigned int *directions) {
|
||||
unsigned int *v = directions;
|
||||
|
||||
for (int dim = 0; dim < n_dimensions; dim++) {
|
||||
// First dimension is a special case
|
||||
if (dim == 0) {
|
||||
for (int i = 0; i < n_directions; i++) {
|
||||
// All m's are 1
|
||||
v[i] = 1 << (31 - i);
|
||||
}
|
||||
} else {
|
||||
int d = sobol_primitives[dim].degree;
|
||||
|
||||
// The first direction numbers (up to the degree of the polynomial)
|
||||
// are simply v[i] = m[i] / 2^i (stored in Q0.32 format)
|
||||
for (int i = 0; i < d; i++) {
|
||||
v[i] = sobol_primitives[dim].m[i] << (31 - i);
|
||||
}
|
||||
|
||||
// The remaining direction numbers are computed as described in
|
||||
// the Bratley and Fox paper.
|
||||
// v[i] = a[1]v[i-1] ^ a[2]v[i-2] ^ ... ^ a[v-1]v[i-d+1] ^ v[i-d] ^
|
||||
// v[i-d]/2^d
|
||||
for (int i = d; i < n_directions; i++) {
|
||||
// First do the v[i-d] ^ v[i-d]/2^d part
|
||||
v[i] = v[i - d] ^ (v[i - d] >> d);
|
||||
|
||||
// Now do the a[1]v[i-1] ^ a[2]v[i-2] ^ ... part
|
||||
// Note that the coefficients a[] are zero or one and for compactness in
|
||||
// the input tables they are stored as bits of a single integer. To
|
||||
// extract the relevant bit we use right shift and mask with 1.
|
||||
// For example, for a 10 degree polynomial there are ten useful bits in
|
||||
// a, so to get a[2] we need to right shift 7 times (to get the 8th bit
|
||||
// into the LSB) and then mask with 1.
|
||||
for (int j = 1; j < d; j++) {
|
||||
v[i] ^= (((sobol_primitives[dim].a >> (d - 1 - j)) & 1) * v[i - j]);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
v += n_directions;
|
||||
}
|
||||
}
|
||||
|
||||
// Reference model for generating Sobol numbers on the host
|
||||
void sobolCPU(int n_vectors, int n_dimensions, unsigned int *directions,
|
||||
float *output) {
|
||||
unsigned int *v = directions;
|
||||
|
||||
for (int d = 0; d < n_dimensions; d++) {
|
||||
unsigned int X = 0;
|
||||
// x[0] is zero (in all dimensions)
|
||||
output[n_vectors * d] = 0.0;
|
||||
|
||||
for (int i = 1; i < n_vectors; i++) {
|
||||
// x[i] = x[i-1] ^ v[c]
|
||||
// where c is the index of the rightmost zero bit in i
|
||||
// minus 1 (since C arrays count from zero)
|
||||
// In the Bratley and Fox paper this is equation (**)
|
||||
X ^= v[ffs(~(i - 1)) - 1];
|
||||
output[i + n_vectors * d] = (float)X * k_2powneg32;
|
||||
}
|
||||
|
||||
v += n_directions;
|
||||
}
|
||||
}
|
||||
64
Samples/5_Domain_Specific/SobolQRNG/sobol_gold.h
Normal file
64
Samples/5_Domain_Specific/SobolQRNG/sobol_gold.h
Normal file
@@ -0,0 +1,64 @@
|
||||
/* 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.
|
||||
*/
|
||||
|
||||
/*
|
||||
* Portions Copyright (c) 2009 Mike Giles, Oxford University. All rights
|
||||
* reserved.
|
||||
* Portions Copyright (c) 2008 Frances Y. Kuo and Stephen Joe. All rights
|
||||
* reserved.
|
||||
*
|
||||
* Sobol Quasi-random Number Generator example
|
||||
*
|
||||
* Based on CUDA code submitted by Mike Giles, Oxford University, United Kingdom
|
||||
* http://people.maths.ox.ac.uk/~gilesm/
|
||||
*
|
||||
* and C code developed by Stephen Joe, University of Waikato, New Zealand
|
||||
* and Frances Kuo, University of New South Wales, Australia
|
||||
* http://web.maths.unsw.edu.au/~fkuo/sobol/
|
||||
*
|
||||
* For theoretical background see:
|
||||
*
|
||||
* P. Bratley and B.L. Fox.
|
||||
* Implementing Sobol's quasirandom sequence generator
|
||||
* http://portal.acm.org/citation.cfm?id=42288
|
||||
* ACM Trans. on Math. Software, 14(1):88-100, 1988
|
||||
*
|
||||
* S. Joe and F. Kuo.
|
||||
* Remark on algorithm 659: implementing Sobol's quasirandom sequence generator.
|
||||
* http://portal.acm.org/citation.cfm?id=641879
|
||||
* ACM Trans. on Math. Software, 29(1):49-57, 2003
|
||||
*
|
||||
*/
|
||||
|
||||
#ifndef SOBOL_GOLD_H
|
||||
#define SOBOL_GOLD_H
|
||||
|
||||
void initSobolDirectionVectors(int n_dimensions, unsigned int *directions);
|
||||
void sobolCPU(int n_vectors, int n_dimensions, unsigned int *directions,
|
||||
float *output);
|
||||
|
||||
#endif
|
||||
207
Samples/5_Domain_Specific/SobolQRNG/sobol_gpu.cu
Normal file
207
Samples/5_Domain_Specific/SobolQRNG/sobol_gpu.cu
Normal file
@@ -0,0 +1,207 @@
|
||||
/* 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.
|
||||
*/
|
||||
|
||||
/*
|
||||
* Portions Copyright (c) 2009 Mike Giles, Oxford University. All rights
|
||||
* reserved.
|
||||
* Portions Copyright (c) 2008 Frances Y. Kuo and Stephen Joe. All rights
|
||||
* reserved.
|
||||
*
|
||||
* Sobol Quasi-random Number Generator example
|
||||
*
|
||||
* Based on CUDA code submitted by Mike Giles, Oxford University, United Kingdom
|
||||
* http://people.maths.ox.ac.uk/~gilesm/
|
||||
*
|
||||
* and C code developed by Stephen Joe, University of Waikato, New Zealand
|
||||
* and Frances Kuo, University of New South Wales, Australia
|
||||
* http://web.maths.unsw.edu.au/~fkuo/sobol/
|
||||
*
|
||||
* For theoretical background see:
|
||||
*
|
||||
* P. Bratley and B.L. Fox.
|
||||
* Implementing Sobol's quasirandom sequence generator
|
||||
* http://portal.acm.org/citation.cfm?id=42288
|
||||
* ACM Trans. on Math. Software, 14(1):88-100, 1988
|
||||
*
|
||||
* S. Joe and F. Kuo.
|
||||
* Remark on algorithm 659: implementing Sobol's quasirandom sequence generator.
|
||||
* http://portal.acm.org/citation.cfm?id=641879
|
||||
* ACM Trans. on Math. Software, 29(1):49-57, 2003
|
||||
*
|
||||
*/
|
||||
|
||||
#include "sobol.h"
|
||||
#include "sobol_gpu.h"
|
||||
#include <cooperative_groups.h>
|
||||
|
||||
namespace cg = cooperative_groups;
|
||||
#include <helper_cuda.h>
|
||||
|
||||
#define k_2powneg32 2.3283064E-10F
|
||||
|
||||
__global__ void sobolGPU_kernel(unsigned n_vectors, unsigned n_dimensions,
|
||||
unsigned *d_directions, float *d_output) {
|
||||
// Handle to thread block group
|
||||
cg::thread_block cta = cg::this_thread_block();
|
||||
__shared__ unsigned int v[n_directions];
|
||||
|
||||
// Offset into the correct dimension as specified by the
|
||||
// block y coordinate
|
||||
d_directions = d_directions + n_directions * blockIdx.y;
|
||||
d_output = d_output + n_vectors * blockIdx.y;
|
||||
|
||||
// Copy the direction numbers for this dimension into shared
|
||||
// memory - there are only 32 direction numbers so only the
|
||||
// first 32 (n_directions) threads need participate.
|
||||
if (threadIdx.x < n_directions) {
|
||||
v[threadIdx.x] = d_directions[threadIdx.x];
|
||||
}
|
||||
|
||||
cg::sync(cta);
|
||||
|
||||
// Set initial index (i.e. which vector this thread is
|
||||
// computing first) and stride (i.e. step to the next vector
|
||||
// for this thread)
|
||||
int i0 = threadIdx.x + blockIdx.x * blockDim.x;
|
||||
int stride = gridDim.x * blockDim.x;
|
||||
|
||||
// Get the gray code of the index
|
||||
// c.f. Numerical Recipes in C, chapter 20
|
||||
// http://www.nrbook.com/a/bookcpdf/c20-2.pdf
|
||||
unsigned int g = i0 ^ (i0 >> 1);
|
||||
|
||||
// Initialisation for first point x[i0]
|
||||
// In the Bratley and Fox paper this is equation (*), where
|
||||
// we are computing the value for x[n] without knowing the
|
||||
// value of x[n-1].
|
||||
unsigned int X = 0;
|
||||
unsigned int mask;
|
||||
|
||||
for (unsigned int k = 0; k < __ffs(stride) - 1; k++) {
|
||||
// We want X ^= g_k * v[k], where g_k is one or zero.
|
||||
// We do this by setting a mask with all bits equal to
|
||||
// g_k. In reality we keep shifting g so that g_k is the
|
||||
// LSB of g. This way we avoid multiplication.
|
||||
mask = -(g & 1);
|
||||
X ^= mask & v[k];
|
||||
g = g >> 1;
|
||||
}
|
||||
|
||||
if (i0 < n_vectors) {
|
||||
d_output[i0] = (float)X * k_2powneg32;
|
||||
}
|
||||
|
||||
// Now do rest of points, using the stride
|
||||
// Here we want to generate x[i] from x[i-stride] where we
|
||||
// don't have any of the x in between, therefore we have to
|
||||
// revisit the equation (**), this is easiest with an example
|
||||
// so assume stride is 16.
|
||||
// From x[n] to x[n+16] there will be:
|
||||
// 8 changes in the first bit
|
||||
// 4 changes in the second bit
|
||||
// 2 changes in the third bit
|
||||
// 1 change in the fourth
|
||||
// 1 change in one of the remaining bits
|
||||
//
|
||||
// What this means is that in the equation:
|
||||
// x[n+1] = x[n] ^ v[p]
|
||||
// x[n+2] = x[n+1] ^ v[q] = x[n] ^ v[p] ^ v[q]
|
||||
// ...
|
||||
// We will apply xor with v[1] eight times, v[2] four times,
|
||||
// v[3] twice, v[4] once and one other direction number once.
|
||||
// Since two xors cancel out, we can skip even applications
|
||||
// and just apply xor with v[4] (i.e. log2(16)) and with
|
||||
// the current applicable direction number.
|
||||
// Note that all these indices count from 1, so we need to
|
||||
// subtract 1 from them all to account for C arrays counting
|
||||
// from zero.
|
||||
unsigned int v_log2stridem1 = v[__ffs(stride) - 2];
|
||||
unsigned int v_stridemask = stride - 1;
|
||||
|
||||
for (unsigned int i = i0 + stride; i < n_vectors; i += stride) {
|
||||
// x[i] = x[i-stride] ^ v[b] ^ v[c]
|
||||
// where b is log2(stride) minus 1 for C array indexing
|
||||
// where c is the index of the rightmost zero bit in i,
|
||||
// not including the bottom log2(stride) bits, minus 1
|
||||
// for C array indexing
|
||||
// In the Bratley and Fox paper this is equation (**)
|
||||
X ^= v_log2stridem1 ^ v[__ffs(~((i - stride) | v_stridemask)) - 1];
|
||||
d_output[i] = (float)X * k_2powneg32;
|
||||
}
|
||||
}
|
||||
|
||||
extern "C" void sobolGPU(int n_vectors, int n_dimensions,
|
||||
unsigned int *d_directions, float *d_output) {
|
||||
const int threadsperblock = 64;
|
||||
|
||||
// Set up the execution configuration
|
||||
dim3 dimGrid;
|
||||
dim3 dimBlock;
|
||||
|
||||
int device;
|
||||
cudaDeviceProp prop;
|
||||
checkCudaErrors(cudaGetDevice(&device));
|
||||
checkCudaErrors(cudaGetDeviceProperties(&prop, device));
|
||||
|
||||
// This implementation of the generator outputs all the draws for
|
||||
// one dimension in a contiguous region of memory, followed by the
|
||||
// next dimension and so on.
|
||||
// Therefore all threads within a block will be processing different
|
||||
// vectors from the same dimension. As a result we want the total
|
||||
// number of blocks to be a multiple of the number of dimensions.
|
||||
dimGrid.y = n_dimensions;
|
||||
|
||||
// If the number of dimensions is large then we will set the number
|
||||
// of blocks to equal the number of dimensions (i.e. dimGrid.x = 1)
|
||||
// but if the number of dimensions is small (e.g. less than four per
|
||||
// multiprocessor) then we'll partition the vectors across blocks
|
||||
// (as well as threads).
|
||||
if (n_dimensions < (4 * prop.multiProcessorCount)) {
|
||||
dimGrid.x = 4 * prop.multiProcessorCount;
|
||||
} else {
|
||||
dimGrid.x = 1;
|
||||
}
|
||||
|
||||
// Cap the dimGrid.x if the number of vectors is small
|
||||
if (dimGrid.x > (unsigned int)(n_vectors / threadsperblock)) {
|
||||
dimGrid.x = (n_vectors + threadsperblock - 1) / threadsperblock;
|
||||
}
|
||||
|
||||
// Round up to a power of two, required for the algorithm so that
|
||||
// stride is a power of two.
|
||||
unsigned int targetDimGridX = dimGrid.x;
|
||||
|
||||
for (dimGrid.x = 1; dimGrid.x < targetDimGridX; dimGrid.x *= 2)
|
||||
;
|
||||
|
||||
// Fix the number of threads
|
||||
dimBlock.x = threadsperblock;
|
||||
|
||||
// Execute GPU kernel
|
||||
sobolGPU_kernel<<<dimGrid, dimBlock>>>(n_vectors, n_dimensions, d_directions,
|
||||
d_output);
|
||||
}
|
||||
63
Samples/5_Domain_Specific/SobolQRNG/sobol_gpu.h
Normal file
63
Samples/5_Domain_Specific/SobolQRNG/sobol_gpu.h
Normal file
@@ -0,0 +1,63 @@
|
||||
/* 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.
|
||||
*/
|
||||
|
||||
/*
|
||||
* Portions Copyright (c) 2009 Mike Giles, Oxford University. All rights
|
||||
* reserved.
|
||||
* Portions Copyright (c) 2008 Frances Y. Kuo and Stephen Joe. All rights
|
||||
* reserved.
|
||||
*
|
||||
* Sobol Quasi-random Number Generator example
|
||||
*
|
||||
* Based on CUDA code submitted by Mike Giles, Oxford University, United Kingdom
|
||||
* http://people.maths.ox.ac.uk/~gilesm/
|
||||
*
|
||||
* and C code developed by Stephen Joe, University of Waikato, New Zealand
|
||||
* and Frances Kuo, University of New South Wales, Australia
|
||||
* http://web.maths.unsw.edu.au/~fkuo/sobol/
|
||||
*
|
||||
* For theoretical background see:
|
||||
*
|
||||
* P. Bratley and B.L. Fox.
|
||||
* Implementing Sobol's quasirandom sequence generator
|
||||
* http://portal.acm.org/citation.cfm?id=42288
|
||||
* ACM Trans. on Math. Software, 14(1):88-100, 1988
|
||||
*
|
||||
* S. Joe and F. Kuo.
|
||||
* Remark on algorithm 659: implementing Sobol's quasirandom sequence generator.
|
||||
* http://portal.acm.org/citation.cfm?id=641879
|
||||
* ACM Trans. on Math. Software, 29(1):49-57, 2003
|
||||
*
|
||||
*/
|
||||
|
||||
#ifndef SOBOL_GPU_H
|
||||
#define SOBOL_GPU_H
|
||||
|
||||
extern "C" void sobolGPU(int n_vectors, int n_dimensions,
|
||||
unsigned int *d_directions, float *d_output);
|
||||
|
||||
#endif
|
||||
10273
Samples/5_Domain_Specific/SobolQRNG/sobol_primitives.cpp
Normal file
10273
Samples/5_Domain_Specific/SobolQRNG/sobol_primitives.cpp
Normal file
File diff suppressed because it is too large
Load Diff
76
Samples/5_Domain_Specific/SobolQRNG/sobol_primitives.h
Normal file
76
Samples/5_Domain_Specific/SobolQRNG/sobol_primitives.h
Normal file
@@ -0,0 +1,76 @@
|
||||
/* 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.
|
||||
*/
|
||||
|
||||
/*
|
||||
* Portions Copyright (c) 2009 Mike Giles, Oxford University. All rights
|
||||
* reserved.
|
||||
* Portions Copyright (c) 2008 Frances Y. Kuo and Stephen Joe. All rights
|
||||
* reserved.
|
||||
*
|
||||
* Sobol Quasi-random Number Generator example
|
||||
*
|
||||
* Based on CUDA code submitted by Mike Giles, Oxford University, United Kingdom
|
||||
* http://people.maths.ox.ac.uk/~gilesm/
|
||||
*
|
||||
* and C code developed by Stephen Joe, University of Waikato, New Zealand
|
||||
* and Frances Kuo, University of New South Wales, Australia
|
||||
* http://web.maths.unsw.edu.au/~fkuo/sobol/
|
||||
*
|
||||
* For theoretical background see:
|
||||
*
|
||||
* P. Bratley and B.L. Fox.
|
||||
* Implementing Sobol's quasirandom sequence generator
|
||||
* http://portal.acm.org/citation.cfm?id=42288
|
||||
* ACM Trans. on Math. Software, 14(1):88-100, 1988
|
||||
*
|
||||
* S. Joe and F. Kuo.
|
||||
* Remark on algorithm 659: implementing Sobol's quasirandom sequence generator.
|
||||
* http://portal.acm.org/citation.cfm?id=641879
|
||||
* ACM Trans. on Math. Software, 29(1):49-57, 2003
|
||||
*
|
||||
*/
|
||||
|
||||
#ifndef SOBOL_PRIMITIVES_H
|
||||
#define SOBOL_PRIMITIVES_H
|
||||
|
||||
#define max_m 17
|
||||
|
||||
// Each primitive is stored as a struct where
|
||||
// dimension is the dimension number of the polynomial (unused)
|
||||
// degree is the degree of the polynomial
|
||||
// a is a binary word representing the coefficients
|
||||
// m is the array of m values
|
||||
struct primitive {
|
||||
unsigned int dimension;
|
||||
unsigned int degree;
|
||||
unsigned int a;
|
||||
unsigned int m[max_m];
|
||||
};
|
||||
|
||||
extern const struct primitive sobol_primitives[];
|
||||
|
||||
#endif
|
||||
Reference in New Issue
Block a user