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

Support for Mixed Input TensorOp (#1084)

Browse Source 
* Passing warp-level mixed input F16*(S8/U8) tests

* passing device-level mixed input F16*(S8/U8) tests

* add to profiler - I8 (111 TFLOPs), U (123 TFLOPs)

* fast numeric conversions (I8 = 132 TFLOPs, U8 = 148 TFLOPs)

* Speedup reference compilation (REVERT THIS COMMIT)

* wider_add.u32_packed_sub.f16x2 (I8 = 132TFLOP/s, U8 = 170 TFLOP/s)

* Improve s8->f16 cvt and support bf16*u8 @158 TFLOPs

* BF16 * S8 (142 TFLOPs)

* Handle mixed-input upcast on OperandA (Support [S8|U8]*[F16|BF16]

* rename OpMultiplyAddMixedInput to OpMultiplyAddMixedInputUpcast

* Add device-level test and profiler support for upcast on operand A

* Move shfl before the cvt and reduce #shfls by 1/2

* fix smem_usage calculation for mixed_input types

* uncomment the stuff (getting ready for merge)

* profiler changes and mixed-input reference

* mixed input reference are in a new file

* use platform instead of std

* comments and typo only

* Use CreateGemmOperator and delete CreateMixedInputGemmOperator

* copyright for new files

* rebase follow-up
This commit is contained in:
Manish Gupta
2023-09-27 08:18:30 -07:00
committed by GitHub
parent 5cd735c48e
commit 7d8317a63e
26 changed files with 2064 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
python/cutlass_library/gemm_operation.py
View File

@@ -88,6 +88,10 @@ class GemmOperation:
]
return self.tile_description.math_instruction.math_operation in complex_operators
#
def is_mixed_input(self):
return self.A.element != self.B.element
#
def is_planar_complex(self):
return self.gemm_kind in (GemmKind.PlanarComplex, GemmKind.PlanarComplexArray)
@@ -149,14 +153,19 @@ class GemmOperation:
if self.C.element != self.tile_description.math_instruction.element_accumulator and \
self.A.element != self.tile_description.math_instruction.element_accumulator:
extended_name = "${element_c}_${core_name}_${element_a}"
if self.is_mixed_input():
extended_name += "_${element_b}"
elif self.C.element == self.tile_description.math_instruction.element_accumulator and \
self.A.element != self.tile_description.math_instruction.element_accumulator:
extended_name = "${core_name}_${element_a}"
if self.is_mixed_input():
extended_name += "_${element_b}"
else:
extended_name = "${core_name}"
extended_name = SubstituteTemplate(extended_name, {
'element_a': DataTypeNames[self.A.element],
'element_b': DataTypeNames[self.B.element],
'element_c': DataTypeNames[self.C.element],
'core_name': self.core_name()
})
@@ -235,7 +244,7 @@ class GemmOperation:
ex = self.extended_name(),
tb = threadblock,
l = self.layout_name(),
a = str(self.A.alignment))
a = str(max(self.A.alignment, self.B.alignment)))
#
def configuration_name(self):