Practice cuda by matrix multiplication.
Author: Sunqianqi
- version 1: use single loop(1.10ms)
- version 2: use cuBLAS(0.77ms)
- version 3: utilize shared memory based on v1(0.98ms)
- version 4: utilize prefetch based on v3(0.59ms)
| Device | NVIDIA A100-PCIE-40GB |
|---|---|
| CUDA Driver Version / Runtime Version | 11.4 / 11.2 |
| CUDA Capability Major/Minor version number | 8.0 |
| Total amount of global memory | 39.59 GBytes (42505273344 bytes) |
| GPU Clock rate | 1410 MHz (1.41 GHz) |
| Memory Clock rate | 1215 Mhz |
| Memory Bus Width | 5120-bit |
| L2 Cache Size | 41943040 bytes |
| Max Texture Dimension Size (x,y,z) | 1D=(131072), 2D=(131072,65536), 3D=(16384,16384,16384) |
| Max Layered Texture Size (dim) x layers | 1D=(32768) x 2048, 2D=(32768,32768) x 2048 |
| Total amount of constant memory | 65536 bytes |
| Total amount of shared memory per block | 49152 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 |
| Maximum sizes of each dimension of a block | 1024 x 1024 x 64 |
| Maximum sizes of each dimension of a grid | 2147483647 x 65535 x 65535 |
| Maximum memory pitch | 2147483647 bytes |