MatmulAlgoCapAttribute#
-
enum nvmath.
bindings. cublasLt. MatmulAlgoCapAttribute(value)[source]# See
cublasLtMatmulAlgoCapAttributes_t
.- Member Type:
Valid values are as follows:
- SPLITK_SUPPORT = <MatmulAlgoCapAttribute.SPLITK_SUPPORT: 0>#
- REDUCTION_SCHEME_MASK = <MatmulAlgoCapAttribute.REDUCTION_SCHEME_MASK: 1>#
- CTA_SWIZZLING_SUPPORT = <MatmulAlgoCapAttribute.CTA_SWIZZLING_SUPPORT: 2>#
- STRIDED_BATCH_SUPPORT = <MatmulAlgoCapAttribute.STRIDED_BATCH_SUPPORT: 3>#
- OUT_OF_PLACE_RESULT_SUPPORT = <MatmulAlgoCapAttribute.OUT_OF_PLACE_RESULT_SUPPORT: 4>#
- UPLO_SUPPORT = <MatmulAlgoCapAttribute.UPLO_SUPPORT: 5>#
- TILE_IDS = <MatmulAlgoCapAttribute.TILE_IDS: 6>#
- CUSTOM_OPTION_MAX = <MatmulAlgoCapAttribute.CUSTOM_OPTION_MAX: 7>#
- CUSTOM_MEMORY_ORDER = <MatmulAlgoCapAttribute.CUSTOM_MEMORY_ORDER: 10>#
- POINTER_MODE_MASK = <MatmulAlgoCapAttribute.POINTER_MODE_MASK: 11>#
- EPILOGUE_MASK = <MatmulAlgoCapAttribute.EPILOGUE_MASK: 12>#
- STAGES_IDS = <MatmulAlgoCapAttribute.STAGES_IDS: 13>#
- LD_NEGATIVE = <MatmulAlgoCapAttribute.LD_NEGATIVE: 14>#
- NUMERICAL_IMPL_FLAGS = <MatmulAlgoCapAttribute.NUMERICAL_IMPL_FLAGS: 15>#
- MIN_ALIGNMENT_A_BYTES = <MatmulAlgoCapAttribute.MIN_ALIGNMENT_A_BYTES: 16>#
- MIN_ALIGNMENT_B_BYTES = <MatmulAlgoCapAttribute.MIN_ALIGNMENT_B_BYTES: 17>#
- MIN_ALIGNMENT_C_BYTES = <MatmulAlgoCapAttribute.MIN_ALIGNMENT_C_BYTES: 18>#
- MIN_ALIGNMENT_D_BYTES = <MatmulAlgoCapAttribute.MIN_ALIGNMENT_D_BYTES: 19>#
- ATOMIC_SYNC = <MatmulAlgoCapAttribute.ATOMIC_SYNC: 20>#
- MATHMODE_IMPL = <MatmulAlgoCapAttribute.MATHMODE_IMPL: 8>#
- GAUSSIAN_IMPL = <MatmulAlgoCapAttribute.GAUSSIAN_IMPL: 9>#