Network Code: test/benchmark.py
| F32/F16 | Spconv 1.x F32 (1080Ti) | Native | Implicit Gemm | Implicit Gemm Split Mask |
|---|---|---|---|---|
| Forward | 43ms | 21.7ms/13.7ms | 23.5ms/11.2ms | 22ms/12.2ms |
| Backward | 80ms | 41.9ms/25.2ms | 51.0ms/13.8ms | 41.1ms/12.2ms |
| F16 Forward | Native | Implicit Gemm | Implicit Gemm Split Mask |
|---|---|---|---|
| RTX 3080 Laptop 150W | 13.7ms | 11.2ms | 12.2ms |
| RTX A6000 | 19.1ms | 11.7ms | 14.0ms |
| TESLA V100 | 17.9ms | 11.4ms | 13.4ms |
| F16 Backward | Native | Implicit Gemm | Implicit Gemm Split Mask |
|---|---|---|---|
| RTX 3080 Laptop 150W | 25.2ms | 13.8ms | 12.2ms |
| RTX A6000 | 28.1ms | 9.2ms | 8.9ms |
| TESLA V100 | 33.9ms | 12.2ms | 12.9ms |
Network Code: test/benchmark.py
The network/input/profile code is same as above table.
This table only profile fp16 gemm kernels without output tensor create/clear overhead. this table show the performance upper bound of our algorithm.
| F16 | Native | Implicit Gemm | Implicit Gemm Split Mask |
|---|---|---|---|
| Forward | 8.0ms | 4.3ms | 4.0ms |
We can see that the implicit gemm is very fast, gemm only use 4.3ms/11.2ms in network forward. we can achieve better performance in TensorRT + Pure C++.
NOTE When you want to benchmark network in your laptop, don't forget to close all apps except terminals! Other apps will consume GPU resource and make kernels run slower.
Comparsion with MinkowskiEngine and torchsparse
TODO