mlir.dialects.transform.x86¶

Classes¶

`ApplyMathRsqrtApproximationPatternsOp`	Collect patterns that approximate vectorized f32 `math.rsqrt` operations
`ApplyMathRsqrtApproximationPatternsOpAdaptor`
`ApplyMoveAccumulatorForContractLoopPatternsOp`	Rewrites an additive vector contraction by zeroing the accumulator input
`ApplyMoveAccumulatorForContractLoopPatternsOpAdaptor`
`ApplyPrepareVectorContractLoopsPatternsOp`	Collect patterns that prepare vector contraction loops for x86 lowering.
`ApplyPrepareVectorContractLoopsPatternsOpAdaptor`
`ApplyShuffleVectorFMAOpsPatternsOp`	Collect patterns to shuffle FMAs with x86 operations as operands
`ApplyShuffleVectorFMAOpsPatternsOpAdaptor`
`ApplySinkVectorProducerOpsPatternsOp`	Collect patterns to sink vector producer operations forward in a block to
`ApplySinkVectorProducerOpsPatternsOpAdaptor`
`ApplyVectorContractBF16ToFMAPatternsOp`	Collect patterns to lower a BF16 type vector.contract operation
`ApplyVectorContractBF16ToFMAPatternsOpAdaptor`
`ApplyVectorContractMultiLevelUnrollForAMXPatternsOp`	Collect patterns that tile and unroll contractions in order to prepare them
`ApplyVectorContractMultiLevelUnrollForAMXPatternsOpAdaptor`
`ApplyVectorContractToAMXDotProductPatternsOp`	Collect patterns to lower a BF16/Int8 type vector.contract operation
`ApplyVectorContractToAMXDotProductPatternsOpAdaptor`
`ApplyVectorContractToFMAPatternsOp`	Collect patterns to lower a F32 type vector.contract operation to a FMA.
`ApplyVectorContractToFMAPatternsOpAdaptor`
`ApplyVectorContractToPackedTypeDotProductPatternsOp`	Collect patterns to lower a BF16/Int8 type vector.contract operation
`ApplyVectorContractToPackedTypeDotProductPatternsOpAdaptor`

Functions¶

`apply_patterns_x86_math_rsqrt_approximation`(...)
`apply_patterns_x86_move_accumulator_for_contract_loop`(...)
`apply_patterns_x86_prepare_vector_contract_loops`(...)
`apply_patterns_x86_shuffle_vector_fma_ops`(...)
`apply_patterns_x86_sink_vector_producer_ops`(...)
`apply_patterns_x86_vector_contract_bf16_to_fma`(...)
`apply_patterns_x86_vector_contract_multi_level_unroll_for_amx`(...)
`apply_patterns_x86_vector_contract_to_amx_dot_product`(...)
`apply_patterns_x86_vector_contract_to_fma`(...)
`apply_patterns_x86_vector_contract_to_packed_type_dot_product`(...)

Module Contents¶

class mlir.dialects.transform.x86.ApplyMathRsqrtApproximationPatternsOp(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None)¶

Bases: _ods_ir

Collect patterns that approximate vectorized f32 math.rsqrt operations using the AVX vrsqrtps instruction (x86.avx.rsqrt) followed by one Newton-Raphson step. The inner vector dimension must be a multiple of 8.

OPERATION_NAME = 'transform.apply_patterns.x86.math_rsqrt_approximation'¶

_ODS_REGIONS = (0, True)¶

class mlir.dialects.transform.x86.ApplyMathRsqrtApproximationPatternsOpAdaptor¶

Bases: _ods_ir

OPERATION_NAME = 'transform.apply_patterns.x86.math_rsqrt_approximation'¶

mlir.dialects.transform.x86.apply_patterns_x86_math_rsqrt_approximation(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None) → ApplyMathRsqrtApproximationPatternsOp¶

class mlir.dialects.transform.x86.ApplyMoveAccumulatorForContractLoopPatternsOp(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None)¶

Bases: _ods_ir

Rewrites an additive vector contraction by zeroing the accumulator input and materializing the accumulation as a separate add after the contraction.

OPERATION_NAME = 'transform.apply_patterns.x86.move_accumulator_for_contract_loop'¶

_ODS_REGIONS = (0, True)¶

class mlir.dialects.transform.x86.ApplyMoveAccumulatorForContractLoopPatternsOpAdaptor¶

Bases: _ods_ir

OPERATION_NAME = 'transform.apply_patterns.x86.move_accumulator_for_contract_loop'¶

mlir.dialects.transform.x86.apply_patterns_x86_move_accumulator_for_contract_loop(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None) → ApplyMoveAccumulatorForContractLoopPatternsOp¶

class mlir.dialects.transform.x86.ApplyPrepareVectorContractLoopsPatternsOp(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None)¶

Bases: _ods_ir

Collect patterns that prepare vector contraction loops for x86 lowering.

OPERATION_NAME = 'transform.apply_patterns.x86.prepare_vector_contract_loops'¶

_ODS_REGIONS = (0, True)¶

class mlir.dialects.transform.x86.ApplyPrepareVectorContractLoopsPatternsOpAdaptor¶

Bases: _ods_ir

OPERATION_NAME = 'transform.apply_patterns.x86.prepare_vector_contract_loops'¶

mlir.dialects.transform.x86.apply_patterns_x86_prepare_vector_contract_loops(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None) → ApplyPrepareVectorContractLoopsPatternsOp¶

class mlir.dialects.transform.x86.ApplyShuffleVectorFMAOpsPatternsOp(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None)¶

Bases: _ods_ir

Collect patterns to shuffle FMAs with x86 operations as operands such that FMAs are grouped with respect to odd/even packed index.

OPERATION_NAME = 'transform.apply_patterns.x86.shuffle_vector_fma_ops'¶

_ODS_REGIONS = (0, True)¶

class mlir.dialects.transform.x86.ApplyShuffleVectorFMAOpsPatternsOpAdaptor¶

Bases: _ods_ir

OPERATION_NAME = 'transform.apply_patterns.x86.shuffle_vector_fma_ops'¶

mlir.dialects.transform.x86.apply_patterns_x86_shuffle_vector_fma_ops(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None) → ApplyShuffleVectorFMAOpsPatternsOp¶

class mlir.dialects.transform.x86.ApplySinkVectorProducerOpsPatternsOp(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None)¶

Bases: _ods_ir

Collect patterns to sink vector producer operations forward in a block to place them immediately before their first use.

OPERATION_NAME = 'transform.apply_patterns.x86.sink_vector_producer_ops'¶

_ODS_REGIONS = (0, True)¶

class mlir.dialects.transform.x86.ApplySinkVectorProducerOpsPatternsOpAdaptor¶

Bases: _ods_ir

OPERATION_NAME = 'transform.apply_patterns.x86.sink_vector_producer_ops'¶

mlir.dialects.transform.x86.apply_patterns_x86_sink_vector_producer_ops(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None) → ApplySinkVectorProducerOpsPatternsOp¶

class mlir.dialects.transform.x86.ApplyVectorContractBF16ToFMAPatternsOp(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None)¶

Bases: _ods_ir

Collect patterns to lower a BF16 type vector.contract operation to a FMA via emulation lowering using BF16 packed operations.

OPERATION_NAME = 'transform.apply_patterns.x86.vector_contract_bf16_to_fma'¶

_ODS_REGIONS = (0, True)¶

class mlir.dialects.transform.x86.ApplyVectorContractBF16ToFMAPatternsOpAdaptor¶

Bases: _ods_ir

OPERATION_NAME = 'transform.apply_patterns.x86.vector_contract_bf16_to_fma'¶

mlir.dialects.transform.x86.apply_patterns_x86_vector_contract_bf16_to_fma(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None) → ApplyVectorContractBF16ToFMAPatternsOp¶

class mlir.dialects.transform.x86.ApplyVectorContractMultiLevelUnrollForAMXPatternsOp(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None)¶

Bases: _ods_ir

Collect patterns that tile and unroll contractions in order to prepare them for the AMX nanokernel lowerings.

OPERATION_NAME = 'transform.apply_patterns.x86.vector_contract_multi_level_unroll_for_amx'¶

_ODS_REGIONS = (0, True)¶

class mlir.dialects.transform.x86.ApplyVectorContractMultiLevelUnrollForAMXPatternsOpAdaptor¶

Bases: _ods_ir

OPERATION_NAME = 'transform.apply_patterns.x86.vector_contract_multi_level_unroll_for_amx'¶

mlir.dialects.transform.x86.apply_patterns_x86_vector_contract_multi_level_unroll_for_amx(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None) → ApplyVectorContractMultiLevelUnrollForAMXPatternsOp¶

class mlir.dialects.transform.x86.ApplyVectorContractToAMXDotProductPatternsOp(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None)¶

Bases: _ods_ir

Collect patterns to lower a BF16/Int8 type vector.contract operation to a BF16/Int8 AMX tiled dot-product.

OPERATION_NAME = 'transform.apply_patterns.x86.vector_contract_to_amx_dot_product'¶

_ODS_REGIONS = (0, True)¶

class mlir.dialects.transform.x86.ApplyVectorContractToAMXDotProductPatternsOpAdaptor¶

Bases: _ods_ir

OPERATION_NAME = 'transform.apply_patterns.x86.vector_contract_to_amx_dot_product'¶

mlir.dialects.transform.x86.apply_patterns_x86_vector_contract_to_amx_dot_product(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None) → ApplyVectorContractToAMXDotProductPatternsOp¶

class mlir.dialects.transform.x86.ApplyVectorContractToFMAPatternsOp(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None)¶

Bases: _ods_ir

Collect patterns to lower a F32 type vector.contract operation to a FMA.

OPERATION_NAME = 'transform.apply_patterns.x86.vector_contract_to_fma'¶

_ODS_REGIONS = (0, True)¶

class mlir.dialects.transform.x86.ApplyVectorContractToFMAPatternsOpAdaptor¶

Bases: _ods_ir

OPERATION_NAME = 'transform.apply_patterns.x86.vector_contract_to_fma'¶

mlir.dialects.transform.x86.apply_patterns_x86_vector_contract_to_fma(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None) → ApplyVectorContractToFMAPatternsOp¶

class mlir.dialects.transform.x86.ApplyVectorContractToPackedTypeDotProductPatternsOp(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None)¶

Bases: _ods_ir

Collect patterns to lower a BF16/Int8 type vector.contract operation to a BF16/Int8 dot-product.

OPERATION_NAME = 'transform.apply_patterns.x86.vector_contract_to_packed_type_dot_product'¶

_ODS_REGIONS = (0, True)¶

class mlir.dialects.transform.x86.ApplyVectorContractToPackedTypeDotProductPatternsOpAdaptor¶

Bases: _ods_ir

OPERATION_NAME = 'transform.apply_patterns.x86.vector_contract_to_packed_type_dot_product'¶

mlir.dialects.transform.x86.apply_patterns_x86_vector_contract_to_packed_type_dot_product(*, loc: _ods_ir | None = None, ip: _ods_ir | None = None) → ApplyVectorContractToPackedTypeDotProductPatternsOp¶