p35189024
/
InfiniTensor
forked from jiuyuan/InfiniTensor
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
InfiniTensor/python/infinitensor/operator_timer.py

30 lines
1.4 KiB
Python
Raw Normal View History

Add python interface for CUDA operator evaluation (#42) * Refactor: seperate data generator * Add: python bindings for opTimer * Fix: test_perfengine Co-authored-by: Liyan Zheng <liyan-zheng@outlook.com>
2022-09-27 10:41:12 +08:00
from tokenize import Double
import pyinfinitensor # import getPerfConv, getPerfMatmul
ADD: add mkl runtime for intel cpu , and add mkl kernel for matmul/conv/convtransposed. (#61) * move memory format transformation to TensorObj clang format add MemoryFormat for tensorObj. use post_ops for fused conv/deconv Distinguish mkl op_timer from cuda op timer. add act optype to conv and deconv add operator timer add mkl kernel for convTransposed minor fix for group conv do not use cblas_sgemm_batch CpuRuntimeObj->NativeCpuRuntimeObj add matmul op for mkl * fix: fix bugs when rebasing from master fix: fix bugs when rebasing from master * fix: update api after rebasing * fix: fix format; fix onnx import * fix: fix clang-format * [fix] fix conv_transpose test * [fix] use stronger test case for transposed conv * [fix] remove tensor memory format; fix mkl transpose conv * [fix] add FIXME tag for op_timer python api --------- Co-authored-by: whjthu <haojie0429@gmail.com>
2023-03-27 21:28:49 +08:00
# FIXME: change API from getPerfOpDevice(...) to getPerfOp(device='dev', ...)
def getPerfConvCuda(n, c, h, w, f, r, s, padh, padw, strideh, stridew, dilationh, dilationw, group, name=""):
return pyinfinitensor.getPerfConvCuda(n, c, h, w, f, r, s, padh, padw,
Add: python API for timing ConvTranspose (#46) * Add: python interfaced for timing operators * Fix: CUDA Runtime run Co-authored-by: Liyan Zheng <liyan-zheng@outlook.com>
2022-10-07 16:03:11 +08:00
strideh, stridew, dilationh, dilationw, group, name)
Add python interface for CUDA operator evaluation (#42) * Refactor: seperate data generator * Add: python bindings for opTimer * Fix: test_perfengine Co-authored-by: Liyan Zheng <liyan-zheng@outlook.com>
2022-09-27 10:41:12 +08:00
ADD: add mkl runtime for intel cpu , and add mkl kernel for matmul/conv/convtransposed. (#61) * move memory format transformation to TensorObj clang format add MemoryFormat for tensorObj. use post_ops for fused conv/deconv Distinguish mkl op_timer from cuda op timer. add act optype to conv and deconv add operator timer add mkl kernel for convTransposed minor fix for group conv do not use cblas_sgemm_batch CpuRuntimeObj->NativeCpuRuntimeObj add matmul op for mkl * fix: fix bugs when rebasing from master fix: fix bugs when rebasing from master * fix: update api after rebasing * fix: fix format; fix onnx import * fix: fix clang-format * [fix] fix conv_transpose test * [fix] use stronger test case for transposed conv * [fix] remove tensor memory format; fix mkl transpose conv * [fix] add FIXME tag for op_timer python api --------- Co-authored-by: whjthu <haojie0429@gmail.com>
2023-03-27 21:28:49 +08:00
def getPerfConvTransposed2dCuda(n, c, h, w, f, r, s, padh, padw, strideh, stridew, dilationh, dilationw, oph, opw, group):
return pyinfinitensor.getPerfConvTransposed2dCuda(n, c, h, w, f, r, s, padh, padw, strideh, stridew, dilationh, dilationw, oph, opw, group)
Add python interface for CUDA operator evaluation (#42) * Refactor: seperate data generator * Add: python bindings for opTimer * Fix: test_perfengine Co-authored-by: Liyan Zheng <liyan-zheng@outlook.com>
2022-09-27 10:41:12 +08:00
ADD: add mkl runtime for intel cpu , and add mkl kernel for matmul/conv/convtransposed. (#61) * move memory format transformation to TensorObj clang format add MemoryFormat for tensorObj. use post_ops for fused conv/deconv Distinguish mkl op_timer from cuda op timer. add act optype to conv and deconv add operator timer add mkl kernel for convTransposed minor fix for group conv do not use cblas_sgemm_batch CpuRuntimeObj->NativeCpuRuntimeObj add matmul op for mkl * fix: fix bugs when rebasing from master fix: fix bugs when rebasing from master * fix: update api after rebasing * fix: fix format; fix onnx import * fix: fix clang-format * [fix] fix conv_transpose test * [fix] use stronger test case for transposed conv * [fix] remove tensor memory format; fix mkl transpose conv * [fix] add FIXME tag for op_timer python api --------- Co-authored-by: whjthu <haojie0429@gmail.com>
2023-03-27 21:28:49 +08:00
def getPerfMatmulCuda(b, m, n, k, name=""):
return pyinfinitensor.getPerfMatmulCuda(b, m, n, k, name)
def getPerfConvMkl(n, c, h, w, f, r, s, padh, padw, strideh, stridew, dilationh, dilationw, group, name=""):
return pyinfinitensor.getPerfConvMkl(n, c, h, w, f, r, s, padh, padw,
strideh, stridew, dilationh, dilationw, group)
def getPerfConvTransposed2dMkl(n, c, h, w, f, r, s, padh, padw, strideh, stridew, dilationh, dilationw, oph, opw, group):
return pyinfinitensor.getPerfConvTransposed2dMkl(n, c, h, w, f, r, s, padh, padw, strideh, stridew, dilationh, dilationw, oph, opw, group)
def getPerfMatmulMkl(b, m, n, k, name=""):
return pyinfinitensor.getPerfMatmulMkl(b, m, n, k)