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/4_CUDA_Libraries/batchCUBLAS/.vscode/c_cpp_properties.json
vendored
Normal file
18
Samples/4_CUDA_Libraries/batchCUBLAS/.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/4_CUDA_Libraries/batchCUBLAS/.vscode/extensions.json
vendored
Normal file
7
Samples/4_CUDA_Libraries/batchCUBLAS/.vscode/extensions.json
vendored
Normal file
@@ -0,0 +1,7 @@
|
||||
{
|
||||
"recommendations": [
|
||||
"nvidia.nsight-vscode-edition",
|
||||
"ms-vscode.cpptools",
|
||||
"ms-vscode.makefile-tools"
|
||||
]
|
||||
}
|
||||
10
Samples/4_CUDA_Libraries/batchCUBLAS/.vscode/launch.json
vendored
Normal file
10
Samples/4_CUDA_Libraries/batchCUBLAS/.vscode/launch.json
vendored
Normal file
@@ -0,0 +1,10 @@
|
||||
{
|
||||
"configurations": [
|
||||
{
|
||||
"name": "CUDA C++: Launch",
|
||||
"type": "cuda-gdb",
|
||||
"request": "launch",
|
||||
"program": "${workspaceFolder}/batchCUBLAS"
|
||||
}
|
||||
]
|
||||
}
|
||||
15
Samples/4_CUDA_Libraries/batchCUBLAS/.vscode/tasks.json
vendored
Normal file
15
Samples/4_CUDA_Libraries/batchCUBLAS/.vscode/tasks.json
vendored
Normal file
@@ -0,0 +1,15 @@
|
||||
{
|
||||
"version": "2.0.0",
|
||||
"tasks": [
|
||||
{
|
||||
"label": "sample",
|
||||
"type": "shell",
|
||||
"command": "make dbg=1",
|
||||
"problemMatcher": ["$nvcc"],
|
||||
"group": {
|
||||
"kind": "build",
|
||||
"isDefault": true
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
330
Samples/4_CUDA_Libraries/batchCUBLAS/Makefile
Normal file
330
Samples/4_CUDA_Libraries/batchCUBLAS/Makefile
Normal file
@@ -0,0 +1,330 @@
|
||||
################################################################################
|
||||
# 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))
|
||||
|
||||
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
|
||||
SMS ?=
|
||||
|
||||
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_$(sm)))
|
||||
|
||||
ifeq ($(SMS),)
|
||||
ifeq ($(TARGET_ARCH),$(filter $(TARGET_ARCH),armv7l aarch64 sbsa))
|
||||
# Generate PTX code from SM 53
|
||||
GENCODE_FLAGS += -gencode arch=compute_53,code=compute_53
|
||||
else
|
||||
# Generate PTX code from SM 35
|
||||
GENCODE_FLAGS += -gencode arch=compute_35,code=compute_35
|
||||
endif
|
||||
endif
|
||||
|
||||
# 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
|
||||
|
||||
LIBRARIES += -lcublas
|
||||
|
||||
################################################################################
|
||||
|
||||
# Target rules
|
||||
all: build
|
||||
|
||||
build: batchCUBLAS
|
||||
|
||||
batchCUBLAS.o:batchCUBLAS.cpp
|
||||
$(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $<
|
||||
|
||||
batchCUBLAS: batchCUBLAS.o
|
||||
$(NVCC) $(ALL_LDFLAGS) $(GENCODE_FLAGS) -o $@ $+ $(LIBRARIES)
|
||||
mkdir -p ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE)
|
||||
cp $@ ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE)
|
||||
|
||||
run: build
|
||||
./batchCUBLAS
|
||||
|
||||
testrun: build
|
||||
|
||||
clean:
|
||||
rm -f batchCUBLAS batchCUBLAS.o
|
||||
rm -rf ../../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE)/batchCUBLAS
|
||||
|
||||
clobber: clean
|
||||
89
Samples/4_CUDA_Libraries/batchCUBLAS/NsightEclipse.xml
Normal file
89
Samples/4_CUDA_Libraries/batchCUBLAS/NsightEclipse.xml
Normal file
@@ -0,0 +1,89 @@
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<!DOCTYPE entry SYSTEM "SamplesInfo.dtd">
|
||||
<entry>
|
||||
<name>batchCUBLAS</name>
|
||||
<cuda_api_list>
|
||||
<driver>cuRand</driver>
|
||||
<driver>cuEqual</driver>
|
||||
<toolkit>cudaFree</toolkit>
|
||||
<toolkit>cudaStreamCreate</toolkit>
|
||||
<toolkit>cudaDeviceSynchronize</toolkit>
|
||||
<toolkit>cudaMalloc</toolkit>
|
||||
<toolkit>cudaGetLastError</toolkit>
|
||||
<toolkit>cudaMemcpy</toolkit>
|
||||
<toolkit>cudaGetErrorString</toolkit>
|
||||
<toolkit>cudaGetDeviceProperties</toolkit>
|
||||
<toolkit>cudaGetDevice</toolkit>
|
||||
</cuda_api_list>
|
||||
<description><![CDATA[A CUDA Sample that demonstrates how using batched CUBLAS API calls to improve overall performance.]]></description>
|
||||
<devicecompilation>whole</devicecompilation>
|
||||
<fallback_min_ptx>true</fallback_min_ptx>
|
||||
<includepaths>
|
||||
<path>./</path>
|
||||
<path>../</path>
|
||||
<path>../../../Common</path>
|
||||
</includepaths>
|
||||
<keyconcepts>
|
||||
<concept level="basic">Linear Algebra</concept>
|
||||
<concept level="basic">CUBLAS Library</concept>
|
||||
</keyconcepts>
|
||||
<keywords>
|
||||
<keyword>CUBLAS</keyword>
|
||||
<keyword>Linear Algebra</keyword>
|
||||
</keywords>
|
||||
<libraries>
|
||||
<library>cublas</library>
|
||||
</libraries>
|
||||
<librarypaths>
|
||||
</librarypaths>
|
||||
<nsight_eclipse>true</nsight_eclipse>
|
||||
<primary_file>batchCUBLAS.cpp</primary_file>
|
||||
<required_dependencies>
|
||||
<dependency>CUBLAS</dependency>
|
||||
</required_dependencies>
|
||||
<scopes>
|
||||
<scope>1:CUDA Basic Topics</scope>
|
||||
<scope>3:Linear Algebra</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>batchCUBLAS</title>
|
||||
<type>exe</type>
|
||||
</entry>
|
||||
77
Samples/4_CUDA_Libraries/batchCUBLAS/README.md
Normal file
77
Samples/4_CUDA_Libraries/batchCUBLAS/README.md
Normal file
@@ -0,0 +1,77 @@
|
||||
# batchCUBLAS - batchCUBLAS
|
||||
|
||||
## Description
|
||||
|
||||
A CUDA Sample that demonstrates how using batched CUBLAS API calls to improve overall performance.
|
||||
|
||||
## Key Concepts
|
||||
|
||||
Linear Algebra, CUBLAS Library
|
||||
|
||||
## 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 Driver API](http://docs.nvidia.com/cuda/cuda-driver-api/index.html)
|
||||
cuRand, cuEqual
|
||||
|
||||
### [CUDA Runtime API](http://docs.nvidia.com/cuda/cuda-runtime-api/index.html)
|
||||
cudaFree, cudaStreamCreate, cudaDeviceSynchronize, cudaMalloc, cudaGetLastError, cudaMemcpy, cudaGetErrorString, cudaGetDeviceProperties, cudaGetDevice
|
||||
|
||||
## Dependencies needed to build/run
|
||||
[CUBLAS](../../README.md#cublas)
|
||||
|
||||
## Prerequisites
|
||||
|
||||
Download and install the [CUDA Toolkit 11.6](https://developer.nvidia.com/cuda-downloads) for your corresponding platform.
|
||||
Make sure the dependencies mentioned in [Dependencies]() section above are installed.
|
||||
|
||||
## 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)
|
||||
|
||||
665
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS.cpp
Normal file
665
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS.cpp
Normal file
@@ -0,0 +1,665 @@
|
||||
/* Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved.
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
* modification, are permitted provided that the following conditions
|
||||
* are met:
|
||||
* * Redistributions of source code must retain the above copyright
|
||||
* notice, this list of conditions and the following disclaimer.
|
||||
* * Redistributions in binary form must reproduce the above copyright
|
||||
* notice, this list of conditions and the following disclaimer in the
|
||||
* documentation and/or other materials provided with the distribution.
|
||||
* * Neither the name of NVIDIA CORPORATION nor the names of its
|
||||
* contributors may be used to endorse or promote products derived
|
||||
* from this software without specific prior written permission.
|
||||
*
|
||||
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
|
||||
* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
||||
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
|
||||
* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
|
||||
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
|
||||
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
|
||||
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
|
||||
* OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
||||
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
||||
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||||
*/
|
||||
|
||||
/*
|
||||
* This example demonstrates how to get better performance by
|
||||
* batching CUBLAS calls with the use of using streams
|
||||
*/
|
||||
|
||||
#include <ctype.h>
|
||||
#include <math.h>
|
||||
#include <stdio.h>
|
||||
#include <stdlib.h>
|
||||
#include <string.h>
|
||||
#if defined(WIN32) || defined(_WIN32) || defined(WIN64) || defined(_WIN64)
|
||||
#include <float.h>
|
||||
#endif
|
||||
|
||||
/* Using updated (v2) interfaces to cublas and cusparse */
|
||||
#include <cublas_v2.h>
|
||||
#include <cuda_runtime.h>
|
||||
|
||||
// Utilities and system includes
|
||||
#include <helper_cuda.h>
|
||||
|
||||
#include "batchCUBLAS.h"
|
||||
|
||||
const char *sSDKname = "batchCUBLAS";
|
||||
|
||||
//==============================================================================
|
||||
// Device information utilities
|
||||
//==============================================================================
|
||||
|
||||
#if defined(__cplusplus)
|
||||
extern "C" {
|
||||
#endif /* __cplusplus */
|
||||
|
||||
int getDeviceVersion(void) {
|
||||
int device;
|
||||
struct cudaDeviceProp properties;
|
||||
|
||||
if (cudaGetDevice(&device) != cudaSuccess) {
|
||||
printf("failed to get device\n");
|
||||
return 0;
|
||||
}
|
||||
|
||||
if (cudaGetDeviceProperties(&properties, device) != cudaSuccess) {
|
||||
printf("failed to get properties\n");
|
||||
return 0;
|
||||
}
|
||||
|
||||
return properties.major * 100 + properties.minor * 10;
|
||||
}
|
||||
|
||||
size_t getDeviceMemory(void) {
|
||||
struct cudaDeviceProp properties;
|
||||
int device;
|
||||
|
||||
if (cudaGetDevice(&device) != cudaSuccess) {
|
||||
return 0;
|
||||
}
|
||||
|
||||
if (cudaGetDeviceProperties(&properties, device) != cudaSuccess) {
|
||||
return 0;
|
||||
}
|
||||
|
||||
return properties.totalGlobalMem;
|
||||
}
|
||||
#if defined(__cplusplus)
|
||||
}
|
||||
#endif /* __cplusplus */
|
||||
|
||||
//==============================================================================
|
||||
// random utilities
|
||||
//==============================================================================
|
||||
|
||||
template <typename T_ELEM>
|
||||
void fillupMatrix(T_ELEM *A, int lda, int rows, int cols, int seed = 0);
|
||||
|
||||
template <typename T_ELEM>
|
||||
void fillupMatrix(T_ELEM *A, int lda, int rows, int cols, int seed) {
|
||||
for (int j = 0; j < cols; j++) {
|
||||
for (int i = 0; i < rows; i++) {
|
||||
A[i + lda * j] = cuGet<T_ELEM>(
|
||||
((double)(((lda * i + j + seed) % 253) + 1)) / 256.0,
|
||||
((double)((((cols * i + j) + 123 + seed) % 253) + 1)) / 256.0);
|
||||
}
|
||||
}
|
||||
}
|
||||
/* Explicit instantiation */
|
||||
template void fillupMatrix<float>(float *A, int lda, int rows, int cols,
|
||||
int seed);
|
||||
template void fillupMatrix<double>(double *A, int lda, int rows, int cols,
|
||||
int seed);
|
||||
|
||||
/* For debugging */
|
||||
void printCuType(const char *str, float A) {
|
||||
fprintf(stdout, "%s (0x%08x, %g)", str, floatAsUInt(A), A);
|
||||
}
|
||||
|
||||
void printCuType(const char *str, double A) {
|
||||
fprintf(stdout, "%s (0x%016llx, %g)", str, doubleAsULL(A), A);
|
||||
}
|
||||
|
||||
//==============================================================================
|
||||
// defines and structures
|
||||
//==============================================================================
|
||||
|
||||
#define CUBLAS_SGEMM_MAX_ULP_ERR (.3)
|
||||
#define CUBLAS_DGEMM_MAX_ULP_ERR (1.e-3)
|
||||
#define CUBLAS_SGEMM_MAX_RELATIVE_ERR (6.e-6)
|
||||
#define CUBLAS_DGEMM_MAX_RELATIVE_ERR (0.0)
|
||||
#define CUBLAS_GEMM_TEST_COUNT (30)
|
||||
#define BENCH_MATRIX_M (128)
|
||||
#define BENCH_MATRIX_K (128)
|
||||
#define BENCH_MATRIX_N (128)
|
||||
|
||||
#define CLEANUP() \
|
||||
do { \
|
||||
if (A) free(A); \
|
||||
if (B) free(B); \
|
||||
if (C) free(C); \
|
||||
for (int i = 0; i < opts.N; ++i) { \
|
||||
if (devPtrA[i]) cudaFree(devPtrA[i]); \
|
||||
if (devPtrB[i]) cudaFree(devPtrB[i]); \
|
||||
if (devPtrC[i]) cudaFree(devPtrC[i]); \
|
||||
} \
|
||||
if (devPtrA) free(devPtrA); \
|
||||
if (devPtrB) free(devPtrB); \
|
||||
if (devPtrC) free(devPtrC); \
|
||||
if (devPtrA_dev) cudaFree(devPtrA_dev); \
|
||||
if (devPtrB_dev) cudaFree(devPtrB_dev); \
|
||||
if (devPtrC_dev) cudaFree(devPtrC_dev); \
|
||||
fflush(stdout); \
|
||||
} while (0)
|
||||
|
||||
enum testMethod { tmRegular, tmStream, tmBatched };
|
||||
|
||||
struct gemmOpts {
|
||||
int m;
|
||||
int n;
|
||||
int k;
|
||||
testMethod test_method;
|
||||
char *elem_type;
|
||||
int N; // number of multiplications
|
||||
};
|
||||
|
||||
template <typename T_ELEM>
|
||||
struct gemmTestParams {
|
||||
cublasOperation_t transa;
|
||||
cublasOperation_t transb;
|
||||
int m;
|
||||
int n;
|
||||
int k;
|
||||
T_ELEM alpha;
|
||||
T_ELEM beta;
|
||||
};
|
||||
|
||||
//==============================================================================
|
||||
// template wrappers for cuda functions
|
||||
//==============================================================================
|
||||
|
||||
static inline cublasStatus_t cublasXgemm(cublasHandle_t handle,
|
||||
cublasOperation_t transa,
|
||||
cublasOperation_t transb, int m, int n,
|
||||
int k, float *alpha, const float *A,
|
||||
int lda, float *B, int ldb,
|
||||
float *beta, float *C, int ldc) {
|
||||
return cublasSgemm(handle, transa, transb, m, n, k, alpha, A, lda, B, ldb,
|
||||
beta, C, ldc);
|
||||
}
|
||||
|
||||
static inline cublasStatus_t cublasXgemm(cublasHandle_t handle,
|
||||
cublasOperation_t transa,
|
||||
cublasOperation_t transb, int m, int n,
|
||||
int k, double *alpha, const double *A,
|
||||
int lda, double *B, int ldb,
|
||||
double *beta, double *C, int ldc) {
|
||||
return cublasDgemm(handle, transa, transb, m, n, k, alpha, A, lda, B, ldb,
|
||||
beta, C, ldc);
|
||||
}
|
||||
|
||||
static inline cublasStatus_t cublasXgemmBatched(
|
||||
cublasHandle_t handle, cublasOperation_t transa, cublasOperation_t transb,
|
||||
int m, int n, int k, float *alpha, const float *Aarray[], int lda,
|
||||
const float *Barray[], int ldb, float *beta, float *Carray[], int ldc,
|
||||
int batchCount) {
|
||||
#if CUDART_VERSION >= 4010
|
||||
return cublasSgemmBatched(handle, transa, transb, m, n, k, alpha, Aarray, lda,
|
||||
Barray, ldb, beta, Carray, ldc, batchCount);
|
||||
#else
|
||||
return CUBLAS_STATUS_SUCCESS;
|
||||
#endif
|
||||
}
|
||||
|
||||
static inline cublasStatus_t cublasXgemmBatched(
|
||||
cublasHandle_t handle, cublasOperation_t transa, cublasOperation_t transb,
|
||||
int m, int n, int k, double *alpha, const double *Aarray[], int lda,
|
||||
const double *Barray[], int ldb, double *beta, double *Carray[], int ldc,
|
||||
int batchCount) {
|
||||
#if CUDART_VERSION >= 4010
|
||||
return cublasDgemmBatched(handle, transa, transb, m, n, k, alpha, Aarray, lda,
|
||||
Barray, ldb, beta, Carray, ldc, batchCount);
|
||||
#else
|
||||
return CUBLAS_STATUS_SUCCESS;
|
||||
#endif
|
||||
}
|
||||
|
||||
//==============================================================================
|
||||
// Primary Application code
|
||||
//==============================================================================
|
||||
|
||||
static int processArgs(int argc, char *argv[], struct gemmOpts *opts) {
|
||||
int error = 0;
|
||||
int oldError;
|
||||
memset(opts, 0, sizeof(*opts));
|
||||
static char default_type[] = "d"; // default double
|
||||
opts->elem_type = default_type;
|
||||
opts->N = 10;
|
||||
|
||||
while (argc) {
|
||||
oldError = error;
|
||||
|
||||
if (*argv[0] == SWITCH_CHAR) {
|
||||
switch (*(argv[0] + 1)) {
|
||||
case 'm':
|
||||
opts->m = (int)atol(argv[0] + 2);
|
||||
break;
|
||||
|
||||
case 'n':
|
||||
opts->n = (int)atol(argv[0] + 2);
|
||||
break;
|
||||
|
||||
case 'k':
|
||||
opts->k = (int)atol(argv[0] + 2);
|
||||
break;
|
||||
|
||||
case 'N':
|
||||
opts->N = (int)atol(argv[0] + 2);
|
||||
break;
|
||||
|
||||
default:
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
if (error > oldError) {
|
||||
fprintf(stderr, "Invalid switch '%c%s'\n", SWITCH_CHAR, argv[0] + 1);
|
||||
}
|
||||
|
||||
argc -= 1;
|
||||
argv++;
|
||||
}
|
||||
|
||||
return error;
|
||||
}
|
||||
|
||||
template <typename T_ELEM>
|
||||
static int TESTGEN(gemm)(const struct gemmOpts *opts, int matrixM, int matrixN,
|
||||
int matrixK, int &numTests,
|
||||
struct gemmTestParams<T_ELEM> *params) {
|
||||
static T_ELEM alpha[] = {cuGet<T_ELEM>(0, 0), cuGet<T_ELEM>(-1, -1),
|
||||
cuGet<T_ELEM>(1, -2), cuGet<T_ELEM>(2, -1),
|
||||
cuGet<T_ELEM>(0, -3)};
|
||||
static T_ELEM beta[] = {cuGet<T_ELEM>(0, 0), cuGet<T_ELEM>(-1, -1),
|
||||
cuGet<T_ELEM>(1, -2), cuGet<T_ELEM>(2, -1),
|
||||
cuGet<T_ELEM>(0, -3)};
|
||||
|
||||
#define NBR_ALPHAS (sizeof(alpha) / sizeof(alpha[0]))
|
||||
#define NBR_BETAS (sizeof(beta) / sizeof(beta[0]))
|
||||
static T_ELEM theAlpha;
|
||||
static T_ELEM theBeta;
|
||||
static int state;
|
||||
static int m;
|
||||
static int n;
|
||||
static int k;
|
||||
|
||||
if (numTests-- <= 0) {
|
||||
return -1;
|
||||
}
|
||||
|
||||
theAlpha = alpha[cuRand() % NBR_ALPHAS];
|
||||
theBeta = beta[cuRand() % NBR_BETAS];
|
||||
params->transa = CUBLAS_OP_N;
|
||||
params->transb = CUBLAS_OP_N;
|
||||
m = matrixM;
|
||||
n = matrixN;
|
||||
k = matrixK;
|
||||
params->m = m;
|
||||
params->n = n;
|
||||
params->k = k;
|
||||
params->alpha = theAlpha;
|
||||
params->beta = theBeta;
|
||||
|
||||
printf("#### args: ta=%d tb=%d m=%d n=%d k=%d ", (unsigned int)params->transa,
|
||||
(unsigned int)params->transb, params->m, params->n, params->k);
|
||||
printCuType(" alpha =", params->alpha);
|
||||
printCuType(" beta=", params->beta);
|
||||
printf("\n");
|
||||
|
||||
m = cuRand() % matrixM;
|
||||
n = cuRand() % matrixN;
|
||||
k = cuRand() % matrixK;
|
||||
|
||||
state = cuRand() % 9;
|
||||
return 0;
|
||||
}
|
||||
|
||||
template <typename T_ELEM>
|
||||
void fillupMatrixDebug(T_ELEM *A, int lda, int rows, int cols) {
|
||||
for (int j = 0; j < cols; j++) {
|
||||
for (int i = 0; i < rows; i++) {
|
||||
A[i + lda * j] = cuGet<T_ELEM>(i + j);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
template <typename T_ELEM>
|
||||
int test_gemm_loop(struct gemmOpts &opts, float err, double max_relative_error,
|
||||
cublasHandle_t handle) {
|
||||
struct gemmTestParams<T_ELEM> params;
|
||||
cudaStream_t *streamArray = 0;
|
||||
cublasStatus_t status1, status2, status3;
|
||||
T_ELEM *A = NULL;
|
||||
T_ELEM *B = NULL;
|
||||
T_ELEM *C = NULL;
|
||||
T_ELEM **devPtrA = 0;
|
||||
T_ELEM **devPtrB = 0;
|
||||
T_ELEM **devPtrC = 0;
|
||||
T_ELEM **devPtrA_dev = NULL;
|
||||
T_ELEM **devPtrB_dev = NULL;
|
||||
T_ELEM **devPtrC_dev = NULL;
|
||||
int matrixM, matrixN, matrixK;
|
||||
int rowsA, rowsB, rowsC;
|
||||
int colsA, colsB, colsC;
|
||||
int matrixSizeA, matrixSizeB, matrixSizeC;
|
||||
int errors;
|
||||
double start, stop;
|
||||
|
||||
printf("Testing %cgemm\n", *opts.elem_type);
|
||||
|
||||
matrixM = (opts.m) ? opts.m : BENCH_MATRIX_M;
|
||||
matrixN = (opts.n) ? opts.n : BENCH_MATRIX_N;
|
||||
matrixK = (opts.k) ? opts.k : BENCH_MATRIX_K;
|
||||
|
||||
rowsA = imax(1, matrixM);
|
||||
colsA = imax(1, matrixK);
|
||||
rowsB = imax(1, matrixK);
|
||||
colsB = imax(1, matrixN);
|
||||
rowsC = imax(1, matrixM);
|
||||
colsC = imax(1, matrixN);
|
||||
|
||||
matrixSizeA = rowsA * colsA;
|
||||
matrixSizeB = rowsB * colsB;
|
||||
matrixSizeC = rowsC * colsC;
|
||||
|
||||
devPtrA = (T_ELEM **)malloc(opts.N * sizeof(*devPtrA));
|
||||
devPtrB = (T_ELEM **)malloc(opts.N * sizeof(*devPtrB));
|
||||
devPtrC = (T_ELEM **)malloc(opts.N * sizeof(*devPtrC));
|
||||
|
||||
for (int i = 0; i < opts.N; i++) {
|
||||
cudaError_t err1 =
|
||||
cudaMalloc((void **)&devPtrA[i], matrixSizeA * sizeof(devPtrA[0][0]));
|
||||
cudaError_t err2 =
|
||||
cudaMalloc((void **)&devPtrB[i], matrixSizeB * sizeof(devPtrB[0][0]));
|
||||
cudaError_t err3 =
|
||||
cudaMalloc((void **)&devPtrC[i], matrixSizeC * sizeof(devPtrC[0][0]));
|
||||
|
||||
if ((err1 != cudaSuccess) || (err2 != cudaSuccess) ||
|
||||
(err3 != cudaSuccess)) {
|
||||
CLEANUP();
|
||||
fprintf(stderr, "!!!! GPU memory allocation error\n");
|
||||
return CUBLASTEST_FAILED;
|
||||
}
|
||||
}
|
||||
|
||||
// For batched processing we need those arrays on the device
|
||||
if (opts.test_method == tmBatched) {
|
||||
cudaError_t err1 =
|
||||
cudaMalloc((void **)&devPtrA_dev, opts.N * sizeof(*devPtrA));
|
||||
cudaError_t err2 =
|
||||
cudaMalloc((void **)&devPtrB_dev, opts.N * sizeof(*devPtrB));
|
||||
cudaError_t err3 =
|
||||
cudaMalloc((void **)&devPtrC_dev, opts.N * sizeof(*devPtrC));
|
||||
|
||||
if ((err1 != cudaSuccess) || (err2 != cudaSuccess) ||
|
||||
(err3 != cudaSuccess)) {
|
||||
CLEANUP();
|
||||
fprintf(stderr, "!!!! GPU memory allocation error\n");
|
||||
return CUBLASTEST_FAILED;
|
||||
}
|
||||
|
||||
err1 = cudaMemcpy(devPtrA_dev, devPtrA, opts.N * sizeof(*devPtrA),
|
||||
cudaMemcpyHostToDevice);
|
||||
err2 = cudaMemcpy(devPtrB_dev, devPtrB, opts.N * sizeof(*devPtrB),
|
||||
cudaMemcpyHostToDevice);
|
||||
err3 = cudaMemcpy(devPtrC_dev, devPtrC, opts.N * sizeof(*devPtrC),
|
||||
cudaMemcpyHostToDevice);
|
||||
|
||||
if ((err1 != cudaSuccess) || (err2 != cudaSuccess) ||
|
||||
(err3 != cudaSuccess)) {
|
||||
CLEANUP();
|
||||
fprintf(stderr, "!!!! cannot copy pointer array to device\n");
|
||||
return CUBLASTEST_FAILED;
|
||||
}
|
||||
}
|
||||
|
||||
A = (T_ELEM *)malloc(matrixSizeA * sizeof(A[0]));
|
||||
B = (T_ELEM *)malloc(matrixSizeB * sizeof(B[0]));
|
||||
C = (T_ELEM *)malloc(matrixSizeC * sizeof(C[0]));
|
||||
|
||||
if ((!A) || (!B) || (!C)) {
|
||||
CLEANUP();
|
||||
fprintf(stderr, "!!!! system memory allocation error\n");
|
||||
return CUBLASTEST_FAILED;
|
||||
}
|
||||
|
||||
streamArray = (cudaStream_t *)malloc(opts.N * sizeof(cudaStream_t *));
|
||||
|
||||
for (int i = 0; i < opts.N; i++) {
|
||||
if (opts.test_method == tmStream) {
|
||||
cudaError_t cudaErr = cudaStreamCreate(&streamArray[i]);
|
||||
|
||||
if (cudaErr != cudaSuccess) {
|
||||
CLEANUP();
|
||||
fprintf(stderr, "!!!! cannot create stream\n");
|
||||
return CUBLASTEST_FAILED;
|
||||
}
|
||||
} else {
|
||||
streamArray[i] = 0;
|
||||
}
|
||||
}
|
||||
|
||||
errors = 0;
|
||||
int numTests = 1;
|
||||
|
||||
while (TESTGEN(gemm)(&opts, matrixM, matrixN, matrixK, numTests, ¶ms) ==
|
||||
0) {
|
||||
printf("#### args: lda=%d ldb=%d ldc=%d\n", rowsA, rowsB, rowsC);
|
||||
|
||||
// fillup with Nan first (so lda padding is full on Nan)
|
||||
memset(A, 0xFF, matrixSizeA * sizeof(A[0]));
|
||||
fillupMatrixDebug(A, rowsA, params.m, params.k);
|
||||
memset(B, 0xFF, matrixSizeB * sizeof(B[0]));
|
||||
fillupMatrix(B, rowsB, params.k, params.n, 121);
|
||||
|
||||
if (!cuEqual(params.beta, cuGet<T_ELEM>(0))) {
|
||||
fillupMatrix(C, rowsC, params.m, params.n);
|
||||
} else {
|
||||
/* fill with SNaNs to make sure ZGEMM doesn't access C */
|
||||
memset(C, 0xFF, matrixSizeC * sizeof(C[0]));
|
||||
}
|
||||
|
||||
double flopsCoef = 2.0;
|
||||
|
||||
for (int i = 0; i < opts.N; i++) {
|
||||
status1 = cublasSetMatrix(rowsA, colsA, sizeof(A[0]), A, rowsA,
|
||||
devPtrA[i], rowsA);
|
||||
status2 = cublasSetMatrix(rowsB, colsB, sizeof(B[0]), B, rowsB,
|
||||
devPtrB[i], rowsB);
|
||||
status3 = cublasSetMatrix(rowsC, colsC, sizeof(C[0]), C, rowsC,
|
||||
devPtrC[i], rowsC);
|
||||
|
||||
if ((status1 != CUBLAS_STATUS_SUCCESS) || (status2 != status1) ||
|
||||
(status3 != status1)) {
|
||||
CLEANUP();
|
||||
fprintf(stderr, "!!!! GPU access error (write)\n");
|
||||
return CUBLASTEST_FAILED;
|
||||
}
|
||||
}
|
||||
|
||||
start = second();
|
||||
|
||||
if (opts.test_method == tmBatched) {
|
||||
cublasSetStream(handle, streamArray[0]);
|
||||
status1 = cublasXgemmBatched(handle, params.transa, params.transb,
|
||||
params.m, params.n, params.k, ¶ms.alpha,
|
||||
(const T_ELEM **)devPtrA_dev, rowsA,
|
||||
(const T_ELEM **)devPtrB_dev, rowsB,
|
||||
¶ms.beta, devPtrC_dev, rowsC, opts.N);
|
||||
|
||||
if (status1 != CUBLAS_STATUS_SUCCESS) {
|
||||
cudaError_t cudaStatus = cudaGetLastError();
|
||||
CLEANUP();
|
||||
fprintf(stderr,
|
||||
"!!!! GPU program execution error : cublas Error=%d, cuda "
|
||||
"Error=%d,(%s)\n",
|
||||
status1, cudaStatus, cudaGetErrorString(cudaStatus));
|
||||
return CUBLASTEST_FAILED;
|
||||
}
|
||||
} else {
|
||||
for (int i = 0; i < opts.N; i++) {
|
||||
cublasSetStream(handle, streamArray[i]);
|
||||
status1 =
|
||||
cublasXgemm(handle, params.transa, params.transb, params.m,
|
||||
params.n, params.k, ¶ms.alpha, devPtrA[i], rowsA,
|
||||
devPtrB[i], rowsB, ¶ms.beta, devPtrC[i], rowsC);
|
||||
|
||||
if (status1 != CUBLAS_STATUS_SUCCESS) {
|
||||
cudaError_t cudaStatus = cudaGetLastError();
|
||||
CLEANUP();
|
||||
fprintf(stderr,
|
||||
"!!!! GPU program execution error : cublas Error=%d, cuda "
|
||||
"Error=%d,(%s)\n",
|
||||
status1, cudaStatus, cudaGetErrorString(cudaStatus));
|
||||
return CUBLASTEST_FAILED;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
cudaError_t cudaStatus = cudaDeviceSynchronize();
|
||||
|
||||
if (cudaStatus != cudaSuccess) {
|
||||
CLEANUP();
|
||||
fprintf(stderr,
|
||||
"!!!! GPU program execution error on cudaDeviceSynchronize : "
|
||||
"cudaError=%d,(%s)\n",
|
||||
cudaStatus, cudaGetErrorString(cudaStatus));
|
||||
return CUBLASTEST_FAILED;
|
||||
}
|
||||
|
||||
stop = second();
|
||||
|
||||
fprintf(stdout, "^^^^ elapsed = %10.8f sec GFLOPS=%g\n", (stop - start),
|
||||
opts.N * (1e-9 * flopsCoef * params.m * params.n * params.k) /
|
||||
(stop - start));
|
||||
|
||||
} // end while (TESTGEN..
|
||||
|
||||
CLEANUP();
|
||||
fprintf(stdout, "@@@@ %cgemm test %s\n", *opts.elem_type,
|
||||
errors ? "FAIL" : "OK");
|
||||
return CUBLASTEST_PASSED;
|
||||
}
|
||||
|
||||
int main(int argc, char *argv[]) {
|
||||
struct gemmOpts opts;
|
||||
int errors, nTimes, nTotalErrors = 0;
|
||||
int status = CUBLASTEST_PASSED;
|
||||
|
||||
printf("%s Starting...\n\n", sSDKname);
|
||||
|
||||
int dev = findCudaDevice(argc, (const char **)argv);
|
||||
|
||||
if (dev == -1) {
|
||||
return CUBLASTEST_FAILED;
|
||||
}
|
||||
|
||||
errors = processArgs(argc, argv, &opts);
|
||||
|
||||
if (errors) {
|
||||
fprintf(stdout,
|
||||
"\n Usage: batchcublas [-mSIZE_M] [-nSIZE_N] [-kSIZE_N] "
|
||||
"[-NSIZE_NUM_ITERATIONS] [-qatest] [-noprompt]\n");
|
||||
return CUBLASTEST_FAILED;
|
||||
}
|
||||
|
||||
cublasHandle_t handle;
|
||||
|
||||
if (cublasCreate(&handle) != CUBLAS_STATUS_SUCCESS) {
|
||||
fprintf(stdout, "CUBLAS initialization failed!\n");
|
||||
exit(EXIT_FAILURE);
|
||||
}
|
||||
|
||||
// Run single kernels
|
||||
fprintf(stdout, "\n ==== Running single kernels ==== \n\n");
|
||||
nTimes = opts.N;
|
||||
opts.N = 1;
|
||||
*(opts.elem_type) = 's';
|
||||
status = test_gemm_loop<float>(opts, (float)CUBLAS_SGEMM_MAX_ULP_ERR,
|
||||
(double)CUBLAS_SGEMM_MAX_RELATIVE_ERR, handle);
|
||||
|
||||
// Run Double version
|
||||
*(opts.elem_type) = 'd';
|
||||
|
||||
if (getDeviceVersion() < DEV_VER_DBL_SUPPORT) {
|
||||
fprintf(stdout, "@@@@ dgemm test WAIVED due to lack of DP support\n");
|
||||
exit(EXIT_WAIVED);
|
||||
}
|
||||
|
||||
status =
|
||||
test_gemm_loop<double>(opts, (float)CUBLAS_DGEMM_MAX_ULP_ERR,
|
||||
(double)CUBLAS_DGEMM_MAX_RELATIVE_ERR, handle);
|
||||
nTotalErrors += (status == CUBLASTEST_PASSED ? 0 : 1);
|
||||
opts.N = nTimes;
|
||||
|
||||
// Run with and without streams and then batched. The batched functions are a
|
||||
// feature new feature in 4.1
|
||||
#if CUDART_VERSION >= 4010
|
||||
|
||||
for (int ii = 0; ii < 3; ii++) {
|
||||
#else
|
||||
|
||||
for (int ii = 0; ii < 2; ii++) {
|
||||
#endif
|
||||
|
||||
switch (ii) {
|
||||
case 0:
|
||||
opts.test_method = tmRegular;
|
||||
fprintf(stdout, "\n ==== Running N=%d without streams ==== \n\n",
|
||||
opts.N);
|
||||
break;
|
||||
|
||||
case 1:
|
||||
opts.test_method = tmStream;
|
||||
fprintf(stdout, "\n ==== Running N=%d with streams ==== \n\n", opts.N);
|
||||
break;
|
||||
|
||||
case 2:
|
||||
opts.test_method = tmBatched;
|
||||
fprintf(stdout, "\n ==== Running N=%d batched ==== \n\n", opts.N);
|
||||
break;
|
||||
}
|
||||
|
||||
// Run single version
|
||||
*(opts.elem_type) = 's';
|
||||
status =
|
||||
test_gemm_loop<float>(opts, (float)CUBLAS_SGEMM_MAX_ULP_ERR,
|
||||
(double)CUBLAS_SGEMM_MAX_RELATIVE_ERR, handle);
|
||||
nTotalErrors += (status == CUBLASTEST_PASSED ? 0 : 1);
|
||||
|
||||
// Run Double version
|
||||
*(opts.elem_type) = 'd';
|
||||
|
||||
// Test doesn't meet minSpec, will will wave the DP test
|
||||
if (getDeviceVersion() < DEV_VER_DBL_SUPPORT) {
|
||||
fprintf(stdout, "@@@@ dgemm test WAIVED due to lack of DP support\n");
|
||||
exit(EXIT_WAIVED);
|
||||
} else {
|
||||
status =
|
||||
test_gemm_loop<double>(opts, (float)CUBLAS_DGEMM_MAX_ULP_ERR,
|
||||
(double)CUBLAS_DGEMM_MAX_RELATIVE_ERR, handle);
|
||||
nTotalErrors += (status == CUBLASTEST_PASSED ? 0 : 1);
|
||||
}
|
||||
}
|
||||
|
||||
cublasDestroy(handle);
|
||||
|
||||
printf("\nTest Summary\n");
|
||||
printf("%d error(s)\n", nTotalErrors);
|
||||
exit(nTotalErrors == 0 ? EXIT_SUCCESS : EXIT_FAILURE);
|
||||
}
|
||||
187
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS.h
Normal file
187
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS.h
Normal file
@@ -0,0 +1,187 @@
|
||||
/* Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved.
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
* modification, are permitted provided that the following conditions
|
||||
* are met:
|
||||
* * Redistributions of source code must retain the above copyright
|
||||
* notice, this list of conditions and the following disclaimer.
|
||||
* * Redistributions in binary form must reproduce the above copyright
|
||||
* notice, this list of conditions and the following disclaimer in the
|
||||
* documentation and/or other materials provided with the distribution.
|
||||
* * Neither the name of NVIDIA CORPORATION nor the names of its
|
||||
* contributors may be used to endorse or promote products derived
|
||||
* from this software without specific prior written permission.
|
||||
*
|
||||
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
|
||||
* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
||||
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
|
||||
* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
|
||||
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
|
||||
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
|
||||
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
|
||||
* OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
||||
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
||||
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||||
*/
|
||||
|
||||
/*
|
||||
* This example demonstrates how to get better performance by
|
||||
* batching CUBLAS calls with the use of using streams
|
||||
*/
|
||||
|
||||
#include <cublas_v2.h>
|
||||
#include <cuda_runtime_api.h>
|
||||
#include <math.h>
|
||||
#include <stdlib.h>
|
||||
|
||||
#define SWITCH_CHAR '-'
|
||||
|
||||
#define REFFUNC(funcname) ref_##funcname
|
||||
#define TESTGEN(funcname) get_##funcname##_params
|
||||
#define TESTPARAMS(funcname) funcname##TestParams
|
||||
|
||||
#define DEV_VER_DBL_SUPPORT (130)
|
||||
#define DEV_VER_ALL_SUPPORT (999)
|
||||
|
||||
/* Errors Tests to be returned by all the Cublas test */
|
||||
#define CUBLASTEST_PASSED 0
|
||||
#define CUBLASTEST_FAILED 1
|
||||
#define CUBLASTEST_WAIVED 2
|
||||
|
||||
//==============================================================================
|
||||
// Device information utilities
|
||||
//==============================================================================
|
||||
|
||||
#if defined(__cplusplus)
|
||||
extern "C" {
|
||||
#endif /* __cplusplus */
|
||||
int getDeviceVersion(void);
|
||||
size_t getDeviceMemory(void);
|
||||
#if defined(__cplusplus)
|
||||
}
|
||||
#endif /* __cplusplus */
|
||||
|
||||
//==============================================================================
|
||||
// Math utilities
|
||||
//==============================================================================
|
||||
|
||||
static __inline__ int imax(int x, int y) { return (x > y) ? x : y; }
|
||||
|
||||
static __inline__ unsigned floatAsUInt(float x) {
|
||||
volatile union {
|
||||
float f;
|
||||
unsigned i;
|
||||
} xx;
|
||||
xx.f = x;
|
||||
return xx.i;
|
||||
}
|
||||
|
||||
static __inline__ unsigned long long doubleAsULL(double x) {
|
||||
volatile union {
|
||||
double f;
|
||||
unsigned long long i;
|
||||
} xx;
|
||||
xx.f = x;
|
||||
return xx.i;
|
||||
}
|
||||
|
||||
static __inline__ unsigned cuRand(void) {
|
||||
/* George Marsaglia's fast inline random number generator */
|
||||
#define CUDA_ZNEW (cuda_z = 36969 * (cuda_z & 65535) + (cuda_z >> 16))
|
||||
#define CUDA_WNEW (cuda_w = 18000 * (cuda_w & 65535) + (cuda_w >> 16))
|
||||
#define CUDA_MWC ((CUDA_ZNEW << 16) + CUDA_WNEW)
|
||||
#define CUDA_SHR3 \
|
||||
(cuda_jsr = cuda_jsr ^ (cuda_jsr << 17), \
|
||||
cuda_jsr = cuda_jsr ^ (cuda_jsr >> 13), \
|
||||
cuda_jsr = cuda_jsr ^ (cuda_jsr << 5))
|
||||
#define CUDA_CONG (cuda_jcong = 69069 * cuda_jcong + 1234567)
|
||||
#define KISS ((CUDA_MWC ^ CUDA_CONG) + CUDA_SHR3)
|
||||
static unsigned int cuda_z = 362436069, cuda_w = 521288629;
|
||||
static unsigned int cuda_jsr = 123456789, cuda_jcong = 380116160;
|
||||
return KISS;
|
||||
}
|
||||
|
||||
//==============================================================================
|
||||
// cuGet and cuEqual versions
|
||||
//==============================================================================
|
||||
|
||||
template <typename T_ELEM>
|
||||
__inline__ __device__ __host__ T_ELEM cuGet(double);
|
||||
template <>
|
||||
__inline__ __device__ __host__ float cuGet<float>(double x) {
|
||||
return float(x);
|
||||
}
|
||||
template <>
|
||||
__inline__ __device__ __host__ double cuGet<double>(double x) {
|
||||
return double(x);
|
||||
}
|
||||
|
||||
template <typename T_ELEM>
|
||||
__inline__ __device__ __host__ T_ELEM cuGet(double, double);
|
||||
template <>
|
||||
__inline__ __device__ __host__ float cuGet<float>(double x, double y) {
|
||||
return float(x);
|
||||
}
|
||||
template <>
|
||||
__inline__ __device__ __host__ double cuGet<double>(double x, double y) {
|
||||
return double(x);
|
||||
}
|
||||
static __inline__ __device__ __host__ bool cuEqual(float x, float y) {
|
||||
return (x == y);
|
||||
}
|
||||
static __inline__ __device__ __host__ bool cuEqual(double x, double y) {
|
||||
return (x == y);
|
||||
}
|
||||
|
||||
//==============================================================================
|
||||
// Platform dependent timing utility
|
||||
//==============================================================================
|
||||
|
||||
#if defined(WIN32) || defined(_WIN32) || defined(WIN64) || defined(_WIN64)
|
||||
#if !defined(WIN32_LEAN_AND_MEAN)
|
||||
#define WIN32_LEAN_AND_MEAN
|
||||
#endif
|
||||
#include <windows.h>
|
||||
static __inline__ double second(void) {
|
||||
LARGE_INTEGER t;
|
||||
static double oofreq;
|
||||
static int checkedForHighResTimer;
|
||||
static BOOL hasHighResTimer;
|
||||
|
||||
if (!checkedForHighResTimer) {
|
||||
hasHighResTimer = QueryPerformanceFrequency(&t);
|
||||
oofreq = 1.0 / (double)t.QuadPart;
|
||||
checkedForHighResTimer = 1;
|
||||
}
|
||||
|
||||
if (hasHighResTimer) {
|
||||
QueryPerformanceCounter(&t);
|
||||
return (double)t.QuadPart * oofreq;
|
||||
} else {
|
||||
return (double)GetTickCount() / 1000.0;
|
||||
}
|
||||
}
|
||||
#elif defined(__linux__) || defined(__QNX__)
|
||||
#include <stddef.h>
|
||||
#include <sys/resource.h>
|
||||
#include <sys/time.h>
|
||||
static double second(void) {
|
||||
struct timeval tv;
|
||||
gettimeofday(&tv, NULL);
|
||||
return (double)tv.tv_sec + (double)tv.tv_usec / 1000000.0;
|
||||
}
|
||||
#elif defined(__APPLE__)
|
||||
#include <stddef.h>
|
||||
#include <sys/resource.h>
|
||||
#include <sys/sysctl.h>
|
||||
#include <sys/time.h>
|
||||
#include <sys/types.h>
|
||||
static double second(void) {
|
||||
struct timeval tv;
|
||||
gettimeofday(&tv, NULL);
|
||||
return (double)tv.tv_sec + (double)tv.tv_usec / 1000000.0;
|
||||
}
|
||||
#else
|
||||
#error unsupported platform
|
||||
#endif
|
||||
20
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS_vs2017.sln
Normal file
20
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS_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}") = "batchCUBLAS", "batchCUBLAS_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
|
||||
112
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS_vs2017.vcxproj
Normal file
112
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS_vs2017.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>batchCUBLAS_vs2017</RootNamespace>
|
||||
<ProjectName>batchCUBLAS</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;$(CudaToolkitIncludeDir);</AdditionalIncludeDirectories>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<SubSystem>Console</SubSystem>
|
||||
<AdditionalDependencies>cublas.lib;cudart_static.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
|
||||
<AdditionalLibraryDirectories>$(CudaToolkitLibDir);</AdditionalLibraryDirectories>
|
||||
<OutputFile>$(OutDir)/batchCUBLAS.exe</OutputFile>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<CodeGeneration>compute_35,compute_35;</CodeGeneration>
|
||||
<AdditionalOptions>-Xcompiler "/wd 4819" --threads 0 </AdditionalOptions>
|
||||
<Include>./;../../../Common</Include>
|
||||
<Defines>WIN32</Defines>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(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="'$(Configuration)'=='Release'">
|
||||
<ClCompile>
|
||||
<Optimization>MaxSpeed</Optimization>
|
||||
<RuntimeLibrary>MultiThreaded</RuntimeLibrary>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<GenerateDebugInformation>false</GenerateDebugInformation>
|
||||
<LinkTimeCodeGeneration>UseLinkTimeCodeGeneration</LinkTimeCodeGeneration>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<Runtime>MT</Runtime>
|
||||
<TargetMachinePlatform>64</TargetMachinePlatform>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemGroup>
|
||||
<ClCompile Include="batchCUBLAS.cpp" />
|
||||
<ClInclude Include="batchCUBLAS.h" />
|
||||
</ItemGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
|
||||
<ImportGroup Label="ExtensionTargets">
|
||||
<Import Project="$(CUDAPropsPath)\CUDA 11.6.targets" />
|
||||
</ImportGroup>
|
||||
</Project>
|
||||
20
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS_vs2019.sln
Normal file
20
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS_vs2019.sln
Normal file
@@ -0,0 +1,20 @@
|
||||
|
||||
Microsoft Visual Studio Solution File, Format Version 12.00
|
||||
# Visual Studio 2019
|
||||
Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "batchCUBLAS", "batchCUBLAS_vs2019.vcxproj", "{997E0757-EA74-4A4E-A0FC-47D8C8831A15}"
|
||||
EndProject
|
||||
Global
|
||||
GlobalSection(SolutionConfigurationPlatforms) = preSolution
|
||||
Debug|x64 = Debug|x64
|
||||
Release|x64 = Release|x64
|
||||
EndGlobalSection
|
||||
GlobalSection(ProjectConfigurationPlatforms) = postSolution
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Debug|x64.ActiveCfg = Debug|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Debug|x64.Build.0 = Debug|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Release|x64.ActiveCfg = Release|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Release|x64.Build.0 = Release|x64
|
||||
EndGlobalSection
|
||||
GlobalSection(SolutionProperties) = preSolution
|
||||
HideSolutionNode = FALSE
|
||||
EndGlobalSection
|
||||
EndGlobal
|
||||
108
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS_vs2019.vcxproj
Normal file
108
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS_vs2019.vcxproj
Normal file
@@ -0,0 +1,108 @@
|
||||
<?xml version="1.0" encoding="utf-8"?>
|
||||
<Project DefaultTargets="Build" ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
|
||||
<PropertyGroup>
|
||||
<CUDAPropsPath Condition="'$(CUDAPropsPath)'==''">$(VCTargetsPath)\BuildCustomizations</CUDAPropsPath>
|
||||
</PropertyGroup>
|
||||
<ItemGroup Label="ProjectConfigurations">
|
||||
<ProjectConfiguration Include="Debug|x64">
|
||||
<Configuration>Debug</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
<ProjectConfiguration Include="Release|x64">
|
||||
<Configuration>Release</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
</ItemGroup>
|
||||
<PropertyGroup Label="Globals">
|
||||
<ProjectGuid>{997E0757-EA74-4A4E-A0FC-47D8C8831A15}</ProjectGuid>
|
||||
<RootNamespace>batchCUBLAS_vs2019</RootNamespace>
|
||||
<ProjectName>batchCUBLAS</ProjectName>
|
||||
<CudaToolkitCustomDir />
|
||||
</PropertyGroup>
|
||||
<Import Project="$(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;$(CudaToolkitIncludeDir);</AdditionalIncludeDirectories>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<SubSystem>Console</SubSystem>
|
||||
<AdditionalDependencies>cublas.lib;cudart_static.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
|
||||
<AdditionalLibraryDirectories>$(CudaToolkitLibDir);</AdditionalLibraryDirectories>
|
||||
<OutputFile>$(OutDir)/batchCUBLAS.exe</OutputFile>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<CodeGeneration>compute_35,compute_35;</CodeGeneration>
|
||||
<AdditionalOptions>-Xcompiler "/wd 4819" --threads 0 </AdditionalOptions>
|
||||
<Include>./;../../../Common</Include>
|
||||
<Defines>WIN32</Defines>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(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="'$(Configuration)'=='Release'">
|
||||
<ClCompile>
|
||||
<Optimization>MaxSpeed</Optimization>
|
||||
<RuntimeLibrary>MultiThreaded</RuntimeLibrary>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<GenerateDebugInformation>false</GenerateDebugInformation>
|
||||
<LinkTimeCodeGeneration>UseLinkTimeCodeGeneration</LinkTimeCodeGeneration>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<Runtime>MT</Runtime>
|
||||
<TargetMachinePlatform>64</TargetMachinePlatform>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemGroup>
|
||||
<ClCompile Include="batchCUBLAS.cpp" />
|
||||
<ClInclude Include="batchCUBLAS.h" />
|
||||
</ItemGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
|
||||
<ImportGroup Label="ExtensionTargets">
|
||||
<Import Project="$(CUDAPropsPath)\CUDA 11.6.targets" />
|
||||
</ImportGroup>
|
||||
</Project>
|
||||
20
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS_vs2022.sln
Normal file
20
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS_vs2022.sln
Normal file
@@ -0,0 +1,20 @@
|
||||
|
||||
Microsoft Visual Studio Solution File, Format Version 12.00
|
||||
# Visual Studio 2022
|
||||
Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "batchCUBLAS", "batchCUBLAS_vs2022.vcxproj", "{997E0757-EA74-4A4E-A0FC-47D8C8831A15}"
|
||||
EndProject
|
||||
Global
|
||||
GlobalSection(SolutionConfigurationPlatforms) = preSolution
|
||||
Debug|x64 = Debug|x64
|
||||
Release|x64 = Release|x64
|
||||
EndGlobalSection
|
||||
GlobalSection(ProjectConfigurationPlatforms) = postSolution
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Debug|x64.ActiveCfg = Debug|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Debug|x64.Build.0 = Debug|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Release|x64.ActiveCfg = Release|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Release|x64.Build.0 = Release|x64
|
||||
EndGlobalSection
|
||||
GlobalSection(SolutionProperties) = preSolution
|
||||
HideSolutionNode = FALSE
|
||||
EndGlobalSection
|
||||
EndGlobal
|
||||
108
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS_vs2022.vcxproj
Normal file
108
Samples/4_CUDA_Libraries/batchCUBLAS/batchCUBLAS_vs2022.vcxproj
Normal file
@@ -0,0 +1,108 @@
|
||||
<?xml version="1.0" encoding="utf-8"?>
|
||||
<Project DefaultTargets="Build" ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
|
||||
<PropertyGroup>
|
||||
<CUDAPropsPath Condition="'$(CUDAPropsPath)'==''">$(VCTargetsPath)\BuildCustomizations</CUDAPropsPath>
|
||||
</PropertyGroup>
|
||||
<ItemGroup Label="ProjectConfigurations">
|
||||
<ProjectConfiguration Include="Debug|x64">
|
||||
<Configuration>Debug</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
<ProjectConfiguration Include="Release|x64">
|
||||
<Configuration>Release</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
</ItemGroup>
|
||||
<PropertyGroup Label="Globals">
|
||||
<ProjectGuid>{997E0757-EA74-4A4E-A0FC-47D8C8831A15}</ProjectGuid>
|
||||
<RootNamespace>batchCUBLAS_vs2022</RootNamespace>
|
||||
<ProjectName>batchCUBLAS</ProjectName>
|
||||
<CudaToolkitCustomDir />
|
||||
</PropertyGroup>
|
||||
<Import Project="$(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;$(CudaToolkitIncludeDir);</AdditionalIncludeDirectories>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<SubSystem>Console</SubSystem>
|
||||
<AdditionalDependencies>cublas.lib;cudart_static.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
|
||||
<AdditionalLibraryDirectories>$(CudaToolkitLibDir);</AdditionalLibraryDirectories>
|
||||
<OutputFile>$(OutDir)/batchCUBLAS.exe</OutputFile>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<CodeGeneration>compute_35,compute_35;</CodeGeneration>
|
||||
<AdditionalOptions>-Xcompiler "/wd 4819" --threads 0 </AdditionalOptions>
|
||||
<Include>./;../../../Common</Include>
|
||||
<Defines>WIN32</Defines>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(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="'$(Configuration)'=='Release'">
|
||||
<ClCompile>
|
||||
<Optimization>MaxSpeed</Optimization>
|
||||
<RuntimeLibrary>MultiThreaded</RuntimeLibrary>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<GenerateDebugInformation>false</GenerateDebugInformation>
|
||||
<LinkTimeCodeGeneration>UseLinkTimeCodeGeneration</LinkTimeCodeGeneration>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<Runtime>MT</Runtime>
|
||||
<TargetMachinePlatform>64</TargetMachinePlatform>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemGroup>
|
||||
<ClCompile Include="batchCUBLAS.cpp" />
|
||||
<ClInclude Include="batchCUBLAS.h" />
|
||||
</ItemGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
|
||||
<ImportGroup Label="ExtensionTargets">
|
||||
<Import Project="$(CUDAPropsPath)\CUDA 11.6.targets" />
|
||||
</ImportGroup>
|
||||
</Project>
|
||||
Reference in New Issue
Block a user