leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 30 Packages Projects Releases Wiki Activity

Make Python interface work for non-SM80 targets (#726)

Browse Source 
* Make Python interface work for non-SM80 targets

* Remove line in README
This commit is contained in:
Jack Kosaian
2022-12-07 21:53:33 -05:00
committed by GitHub
parent d6117ca362
commit df81d847d7
33 changed files with 149 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/library/scripts/pycutlass/test/gemm/gemm_bf16_sm80.py
+3
View File
@@ -4,7 +4,10 @@ from pycutlass.test import *
import unittest
from pycutlass.test.gemm_testbed import test_all_gemm
from pycutlass.utils.device import device_cc
@unittest.skipIf(device_cc() < 80, "Device compute capability is insufficient for SM80 tests.")
class GemmBF16TensorOpSm80(unittest.TestCase):
def SM80_Device_Gemm_bf16n_bf16n_f32t_tensor_op_f32_64x128x64_32x64x64(self):
math_inst = MathInstruction(
tools/library/scripts/pycutlass/test/gemm/gemm_f16_sm80.py
+2
View File
@@ -4,8 +4,10 @@ from pycutlass.test import *
import unittest
from pycutlass.test.gemm_testbed import test_all_gemm
from pycutlass.utils.device import device_cc
@unittest.skipIf(device_cc() < 80, "Device compute capability is insufficient for SM80 tests.")
class GemmF16Sm80(unittest.TestCase):
def test_SM80_Device_Gemm_f32t_f32n_f32t_tensor_op_bf16_f32_128x128x32_64x64x32(self):
math_inst = MathInstruction(
tools/library/scripts/pycutlass/test/gemm/gemm_f32_sm80.py
+2
View File
@@ -5,8 +5,10 @@ from pycutlass.test import *
import unittest
from pycutlass.test.gemm_testbed import test_all_gemm
from pycutlass.utils.device import device_cc
@unittest.skipIf(device_cc() < 80, "Device compute capability is insufficient for SM80 tests.")
class GemmF32nF32nF32nTensorOpF32Sm80(unittest.TestCase):
def test_SM80_Device_Gemm_f32t_f32n_f32t_tensor_op_bf16_f32_128x128x32_64x64x32(self):
math_inst = MathInstruction(
tools/library/scripts/pycutlass/test/gemm/gemm_f64_sm80.py
+3
View File
@@ -4,7 +4,10 @@ from pycutlass.test import *
import unittest
from pycutlass.test.gemm_testbed import test_all_gemm
from pycutlass.utils.device import device_cc
@unittest.skipIf(device_cc() < 80, "Device compute capability is insufficient for SM80 tests.")
class GemmF64TensorOpSm80(unittest.TestCase):
def test_SM80_Device_Gemm_f64n_f64t_f64t_tensor_op_f64_32x32x16_16x16x16(self):
math_inst = MathInstruction(
tools/library/scripts/pycutlass/test/gemm/gemm_grouped_sm80.py
+2
View File
@@ -4,8 +4,10 @@ from pycutlass.test import *
import unittest
from pycutlass.test.gemm_grouped_testbed import TestbedGrouped
from pycutlass.utils.device import device_cc
@unittest.skipIf(device_cc() < 80, "Device compute capability is insufficient for SM80 tests.")
class GemmGroupedSm80(unittest.TestCase):
def test_SM80_Device_GemmGrouped_f16n_f16t_f32n_tensor_op_f32_128x128x32_64x64x32(self):
math_inst = MathInstruction(
tools/library/scripts/pycutlass/test/gemm/gemm_s8_sm80.py
+3
View File
@@ -5,7 +5,10 @@ from pycutlass.test import *
import unittest
from pycutlass.test.gemm_testbed import test_all_gemm
from pycutlass.utils.device import device_cc
@unittest.skipIf(device_cc() < 80, "Device compute capability is insufficient for SM80 tests.")
class GemmS8TensorOpF32Sm80(unittest.TestCase):
def test_SM80_Device_Gemm_s8t_s8n_s8t_tensor_op_s32_64x64x64_32x32x64(self):
math_inst = MathInstruction(