前言
为了快速了解GPU以及入门CUDA编程,很有必要对GPU硬件有一个基本的认识, 下面通过cuda_runtime提供的API进行GPU设备查询以及相关硬件参数的了解.
测试环境
- OS: Ubuntu 20.04
- CUDA: v11
- GCC-10
- VSCode
测试代码
cudaRuntime提供了相应的API用于设备查询:
cudaGetDeviceCount()cudaGetDeviceProperties()
#include <iostream>
#include <cuda_runtime.h>
#include <stdio.h>
#define CHECK(call) \
do \
{ \
cudaError_t ret = call; \
if (ret != cudaSuccess) \
{ \
printf(" CUDA Error\n"); \
printf(" File:%s\n", __FILE__); \
printf(" Line:%d\n", __LINE__); \
printf(" Error code:%d\n", ret); \
printf(" Error text:%s\n", cudaGetErrorString(ret)); \
} \
} while (0)
int main()
{
int device_count = 0;
CHECK(cudaGetDeviceCount(&device_count));
if(device_count > 0) {
std::cout << "Found " << device_count << " GPUs!" << std::endl;
for(int id=0;id<device_count;id++) {
std::cout << "Device: " << id << std::endl;
cudaDeviceProp prop;
CHECK(cudaGetDeviceProperties(&prop, id));
printf("GPU Name: %s\n", prop.name);
printf("GPU Global Memory(显存容量): %f GB\n", (float)prop.totalGlobalMem/(1024*1024*1024));
printf("GPU Memory 位宽:%d bit\n", prop.memoryBusWidth);
printf("GPU SM个数:%d\n", prop.multiProcessorCount);
printf("GPU 每个SM上最大线程数量:%d\n", prop.maxThreadsPerMultiProcessor);
}
}else {
std::cout << "No NVIDIA GPU Exist !" << std::endl;
}
}
输出结果:
image.png
CUDA提供的GPU 属性介绍:
struct __device_builtin__ cudaDeviceProp
{
char name[256]; /**< ASCII string identifying device */
cudaUUID_t uuid; /**< 16-byte unique identifier */
char luid[8]; /**< 8-byte locally unique identifier. Value is undefined on TCC and non-Windows platforms */
unsigned int luidDeviceNodeMask; /**< LUID device node mask. Value is undefined on TCC and non-Windows platforms */
size_t totalGlobalMem; /**< Global memory available on device in bytes */
size_t sharedMemPerBlock; /**< Shared memory available per block in bytes */
int regsPerBlock; /**< 32-bit registers available per block */
int warpSize; /**< Warp size in threads */
size_t memPitch; /**< Maximum pitch in bytes allowed by memory copies */
int maxThreadsPerBlock; /**< Maximum number of threads per block */
int maxThreadsDim[3]; /**< Maximum size of each dimension of a block */
int maxGridSize[3]; /**< Maximum size of each dimension of a grid */
int clockRate; /**< Clock frequency in kilohertz */
size_t totalConstMem; /**< Constant memory available on device in bytes */
int major; /**< Major compute capability */
int minor; /**< Minor compute capability */
size_t textureAlignment; /**< Alignment requirement for textures */
size_t texturePitchAlignment; /**< Pitch alignment requirement for texture references bound to pitched memory */
int deviceOverlap; /**< Device can concurrently copy memory and execute a kernel. Deprecated. Use instead asyncEngineCount. */
int multiProcessorCount; /**< Number of multiprocessors on device */
int kernelExecTimeoutEnabled; /**< Specified whether there is a run time limit on kernels */
int integrated; /**< Device is integrated as opposed to discrete */
int canMapHostMemory; /**< Device can map host memory with cudaHostAlloc/cudaHostGetDevicePointer */
int computeMode; /**< Compute mode (See ::cudaComputeMode) */
int maxTexture1D; /**< Maximum 1D texture size */
int maxTexture1DMipmap; /**< Maximum 1D mipmapped texture size */
int maxTexture1DLinear; /**< Maximum size for 1D textures bound to linear memory */
int maxTexture2D[2]; /**< Maximum 2D texture dimensions */
int maxTexture2DMipmap[2]; /**< Maximum 2D mipmapped texture dimensions */
int maxTexture2DLinear[3]; /**< Maximum dimensions (width, height, pitch) for 2D textures bound to pitched memory */
int maxTexture2DGather[2]; /**< Maximum 2D texture dimensions if texture gather operations have to be performed */
int maxTexture3D[3]; /**< Maximum 3D texture dimensions */
int maxTexture3DAlt[3]; /**< Maximum alternate 3D texture dimensions */
int maxTextureCubemap; /**< Maximum Cubemap texture dimensions */
int maxTexture1DLayered[2]; /**< Maximum 1D layered texture dimensions */
int maxTexture2DLayered[3]; /**< Maximum 2D layered texture dimensions */
int maxTextureCubemapLayered[2];/**< Maximum Cubemap layered texture dimensions */
int maxSurface1D; /**< Maximum 1D surface size */
int maxSurface2D[2]; /**< Maximum 2D surface dimensions */
int maxSurface3D[3]; /**< Maximum 3D surface dimensions */
int maxSurface1DLayered[2]; /**< Maximum 1D layered surface dimensions */
int maxSurface2DLayered[3]; /**< Maximum 2D layered surface dimensions */
int maxSurfaceCubemap; /**< Maximum Cubemap surface dimensions */
int maxSurfaceCubemapLayered[2];/**< Maximum Cubemap layered surface dimensions */
size_t surfaceAlignment; /**< Alignment requirements for surfaces */
int concurrentKernels; /**< Device can possibly execute multiple kernels concurrently */
int ECCEnabled; /**< Device has ECC support enabled */
int pciBusID; /**< PCI bus ID of the device */
int pciDeviceID; /**< PCI device ID of the device */
int pciDomainID; /**< PCI domain ID of the device */
int tccDriver; /**< 1 if device is a Tesla device using TCC driver, 0 otherwise */
int asyncEngineCount; /**< Number of asynchronous engines */
int unifiedAddressing; /**< Device shares a unified address space with the host */
int memoryClockRate; /**< Peak memory clock frequency in kilohertz */
int memoryBusWidth; /**< Global memory bus width in bits */
int l2CacheSize; /**< Size of L2 cache in bytes */
int persistingL2CacheMaxSize; /**< Device's maximum l2 persisting lines capacity setting in bytes */
int maxThreadsPerMultiProcessor;/**< Maximum resident threads per multiprocessor */
int streamPrioritiesSupported; /**< Device supports stream priorities */
int globalL1CacheSupported; /**< Device supports caching globals in L1 */
int localL1CacheSupported; /**< Device supports caching locals in L1 */
size_t sharedMemPerMultiprocessor; /**< Shared memory available per multiprocessor in bytes */
int regsPerMultiprocessor; /**< 32-bit registers available per multiprocessor */
int managedMemory; /**< Device supports allocating managed memory on this system */
int isMultiGpuBoard; /**< Device is on a multi-GPU board */
int multiGpuBoardGroupID; /**< Unique identifier for a group of devices on the same multi-GPU board */
int hostNativeAtomicSupported; /**< Link between the device and the host supports native atomic operations */
int singleToDoublePrecisionPerfRatio; /**< Ratio of single precision performance (in floating-point operations per second) to double precision performance */
int pageableMemoryAccess; /**< Device supports coherently accessing pageable memory without calling cudaHostRegister on it */
int concurrentManagedAccess; /**< Device can coherently access managed memory concurrently with the CPU */
int computePreemptionSupported; /**< Device supports Compute Preemption */
int canUseHostPointerForRegisteredMem; /**< Device can access host registered memory at the same virtual address as the CPU */
int cooperativeLaunch; /**< Device supports launching cooperative kernels via ::cudaLaunchCooperativeKernel */
int cooperativeMultiDeviceLaunch; /**< Device can participate in cooperative kernels launched via ::cudaLaunchCooperativeKernelMultiDevice */
size_t sharedMemPerBlockOptin; /**< Per device maximum shared memory per block usable by special opt in */
int pageableMemoryAccessUsesHostPageTables; /**< Device accesses pageable memory via the host's page tables */
int directManagedMemAccessFromHost; /**< Host can directly access managed memory on the device without migration. */
int maxBlocksPerMultiProcessor; /**< Maximum number of resident blocks per multiprocessor */
int accessPolicyMaxWindowSize; /**< The maximum value of ::cudaAccessPolicyWindow::num_bytes. */
size_t reservedSharedMemPerBlock; /**< Shared memory reserved by CUDA driver per block in bytes */
};
deviceQuery App
在安装好CUDA之后, CUDA提供了一些sample, 其他包含deviceQuery 可执行的app, 用法查看本机上的GPU硬件, sample的位置: /usr/local/cuda/samples
├── 0_Simple
├── 1_Utilities
├── 2_Graphics
├── 3_Imaging
├── 4_Finance
├── 5_Simulations
├── 6_Advanced
├── 7_CUDALibraries
├── bin
├── common
├── EULA.txt
└── Makefile
samples目录包含了若干例子, samples的根目录提供的makefile, 因此只需进行make即可生成app:
make -j8
查找 deviceQuery 可执行程序:
cd bin/x84_x64/linux/release
ls | grep deviceQuery
# Run
./deviceQuery
输出结果:
/deviceQuery Starting...
CUDA Device Query (Runtime API) version (CUDART static linking)
Detected 1 CUDA Capable device(s)
Device 0: "NVIDIA GeForce MX250"
CUDA Driver Version / Runtime Version 11.5 / 11.1
CUDA Capability Major/Minor version number: 6.1
Total amount of global memory: 2003 MBytes (2099904512 bytes)
( 3) Multiprocessors, (128) CUDA Cores/MP: 384 CUDA Cores
GPU Max Clock rate: 1582 MHz (1.58 GHz)
Memory Clock rate: 3004 Mhz
Memory Bus Width: 64-bit
L2 Cache Size: 524288 bytes
Maximum Texture Dimension Size (x,y,z) 1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
Maximum Layered 1D Texture Size, (num) layers 1D=(32768), 2048 layers
Maximum Layered 2D Texture Size, (num) layers 2D=(32768, 32768), 2048 layers
Total amount of constant memory: 65536 bytes
Total amount of shared memory per block: 49152 bytes
Total shared memory per multiprocessor: 98304 bytes
Total number of registers available per block: 65536
Warp size: 32
Maximum number of threads per multiprocessor: 2048
Maximum number of threads per block: 1024
Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
Max dimension size of a grid size (x,y,z): (2147483647, 65535, 65535)
Maximum memory pitch: 2147483647 bytes
Texture alignment: 512 bytes
Concurrent copy and kernel execution: Yes with 2 copy engine(s)
Run time limit on kernels: Yes
Integrated GPU sharing Host Memory: No
Support host page-locked memory mapping: Yes
Alignment requirement for Surfaces: Yes
Device has ECC support: Disabled
Device supports Unified Addressing (UVA): Yes
Device supports Managed Memory: Yes
Device supports Compute Preemption: Yes
Supports Cooperative Kernel Launch: Yes
Supports MultiDevice Co-op Kernel Launch: Yes
Device PCI Domain ID / Bus ID / location ID: 0 / 1 / 0
Compute Mode:
< Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >
deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 11.5, CUDA Runtime Version = 11.1, NumDevs = 1
Result = PASS
BandWidthTest App
用途: 测试GPU显存的传输带宽
./bandwidthTest
[CUDA Bandwidth Test] - Starting...
Running on...
Device 0: NVIDIA GeForce MX250
Quick Mode
Host to Device Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(GB/s)
32000000 3.1
Device to Host Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(GB/s)
32000000 3.3
Device to Device Bandwidth, 1 Device(s)
PINNED Memory Transfers
Transfer Size (Bytes) Bandwidth(GB/s)
32000000 41.2
Result = PASS
NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.
- Device To Host bandwidth 和 Host to Device bandwidth基本上都是 3GB/s, GPU和CPU的通信是通过PCIe总线, PCIe的带宽是CPU-GPU的通信瓶颈.
- Device To Device bandwidth: 41GB/s, GPU显存自身的带宽比PCIe通信的带宽大的多, 因此大多数GPU 程序都是提前把输入数据准备好一次性复制到GPU显存, 这样可以减少CPU-GPU通信带来的访存overhead.
网友评论