'iree_gpu' Dialectlink

A dialect for common functionality used by GPU focused IREE code generation.

This dialect provides operations and attributes to aid in code generation for GPU targets. The functionality in this dialect can be hardware specific, but is intended to be independent of the lowering target. Late lowerings to SPIR-V/LLVM are handled separately.

'iree_gpu' Dialect

Operationslink

`iree_gpu.barrier_region` (GPU::BarrierRegionOp)link

Synchronizes uses of a shared tensor.

Syntax:

operation ::= `iree_gpu.barrier_region` (`ins` `(` $inputs^ `:` type($inputs) `)` )?
              $region attr-dict `:` type($results)

This op is designed to represent synchronization of workers on the operands and results of the given region. This operation naturally arises when combining the regions of producer-consumer scf.forall operations that share a mapping type.

For example, consider the following pair of parallel loops.

  %0 = scf.forall (%idy, %idx) in (2, 32) shared_outs(%init = %empty) -> (tensor<4x128xf32>) {
    %in = ...
    %2 = affine.apply #affine_map<(d0) -> (d0 * 2)> (%idy)
    %3 = affine.apply #affine_map<(d0) -> (d0 * 4)> (%idx)
    scf.forall.in_parallel {
      tensor.parallel_insert_slice %in into %init[%2, %3] [2, 4] [1, 1]
        : tensor<2x4xf32> into tensor<4x128xf32>
    }
  } {mapping = [#gpu.thread<y>, #gpu.thread<x>]}
  %1 = scf.forall (%idy, %idx) in (8, 8) -> (tensor<128x128xf32>) {
    %4 = affine.apply #affine_map<(d0) -> (d0 * 16)> (%idx)
    %extracted_slice = tensor.extract_slice %0[0, %4] [4, 16] [1, 1]
      : tensor<4x128xf32> to tensor<4x16xf32>
    ...
  } {mapping = [#gpu.thread<y>, #gpu.thread<x>]}

Because these loops share the same worker type and total count, the bodies of these two loops can be merged with a barrier an insert_slice and a shuffle where the boundary of the loops currently is.

  %0 = scf.forall (%idy, %idx) in (8, 8) -> (tensor<4x128xf32>) {
    %alloc = bufferization.alloc_tensor {memory_space = #gpu.address_space<workgroup>}
      : tensor<4x128xf32>
    %barrier = iree_gpu.barrier_region %alloc {
    ^bb0(%shared: tensor<4x128xf32>):
      %ids = affine.delinearize_index %idy * 8 + %idx to (2, 32) : index
      %in = ...
      %2 = affine.apply #affine_map<(d0) -> (d0 * 2)> (%ids#0)
      %3 = affine.apply #affine_map<(d0) -> (d0 * 4)> (%ids#1)
      %inserted_slice = tensor.insert_slice %in into %shared[%2, %3] [2, 4] [1, 1]
        : tensor<2x4xf32> to tensor<4x128xf32>
      iree_gpu.yield %slice : tensor<4x16xf32>
    } : tensor<4x128xf32> -> tensor<4x16xf32>
    %4 = affine.apply #affine_map<(d0) -> (d0 * 16)> (%idx)
    %slice = tensor.extract_slice %barrier[0, %4] [4, 16] [1, 1] : tensor<4x128xf32> to tensor<4x16xf32>
    ...
  } {mapping = [#gpu.thread<y>, #gpu.thread<x>]}

A barrier_region can be lowered to two barriers, one on the input operands and a second one on the results.

Movtivation and Intended Use Cases:

The primary way this op is generated is when fusing parallel loops with tensor results. This operation helps to make lowerings more progressive and flexible. - Lowering directly to an alloc + reads and writes breaks the dependency chain making transformations like barrier placement and pipelining potentially more difficult. - Allows the option of non-vector based lowering paths.

Traits: AlwaysSpeculatableImplTrait, SingleBlockImplicitTerminator<mlir::iree_compiler::IREE::GPU::YieldOp>, SingleBlock

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface)

Effects: MemoryEffects::Effect{}

Operands:link

Operand	Description
`inputs`	variadic of any type

Results:link

Result	Description
`results`	variadic of any type

`iree_gpu.multi_mma` (GPU::MultiMmaOp)link

Models a contraction of multiple mma operations

Syntax:

operation ::= `iree_gpu.multi_mma` $lhs `,` $rhs `,` $acc attr-dict
              `:` type($lhs) `,` type($rhs) `into` type($acc)

Computes the sum of inner MMA operations along a set of outer dimensions. Logically matches closely with a vector.contraction operation, however the combiner type is a specific intrinsic rather than a generic combiner type.

Similar to vector.contraction, an iterator type attribute list must be specified, where each element of the list represents an iterator over one of the outer dimensions. Iteration of inner dimensions is defined solely by the intrinsic and may be opaque.

An indexing map attribute list must be specified with an entry for lhs, rhs and acc arguments. An indexing map attribute specifies a mapping from each outer loop iterator in the iterator type list, to each dimension of each operand.

The combiner type is defined by the intrinsic.

Example:

#contraction_accesses = [
 affine_map<(i, j, k) -> (i, k)>,
 affine_map<(i, j, k) -> (k, j)>,
 affine_map<(i, j, k) -> (i, j)>
]
#contraction_trait = {
  indexing_maps = #contraction_accesses,
  iterator_types = ["parallel", "parallel", "reduction"],
  kind = #iree_gpu.mma_layout<MFMA_F32_16x16x16_F16>
}
%3 = iree_gpu.multi_mma %0, %1, %2 #contraction_trait
  : vector<2x3x4xf16>, vector<3x5x4xf16> into vector<2x5x4xf32>

// Takes tensors as well, however the inner dimensions must always be
// static.
%7 = iree_gpu.multi_mma %4, %5, %6 #contraction_trait
  : tensor<?x?x4xf16>, tensor<?x?x4xf16> into tensor<?x?x4xf32>

The example above can be logically lowered directly to loops like this (ignoring type conversions from tensor to vector needed for the mfma).

%outer_m = tensor.dim %6, %c0 : index
%outer_n = tensor.dim %6, %c1 : index
%outer_k = tensor.dim %4, %c1 : index
%7 = scf.for %i = %c0 to %outer_m iter_args(%arg0 = %6) {
  %8 = scf.for %j = %c0 to %outer_n iter_args(%arg1 = %arg0) {
    %9 = scf.for %k = %c0 to %outer_k iter_args(%arg2 = %arg1) {
      %lhs = tensor.extract_slice %4 [%i, %k, 0] [1, 1, 4] [1, 1, 1] : tensor<4xf16>
      %rhs = tensor.extract_slice %5 [%k, %j, 0] [1, 1, 4] [1, 1, 1] : tensor<4xf16>
      %acc = tensor.extract_slice %arg2 [%i, %j, 0] [1, 1, 4] [1, 1, 1] : tensor<4xf32>
      %res = amdgpu.mfma %lhs, %rhs, %acc : tensor<4xf32>
      %ret = tensor.insert_slice %acc into %arg2 [%i, %j, 0] [1, 1, 4] [1, 1, 1] : tensor<?x?x4xf32>
      scf.yield %ret : tensor<?x?x4xf32>
    }
    scf.yield %9 : tensor<?x?x4xf32>
  }
  scf.yield %8 : tensor<?x?x4xf32>
}

Or alternatively unrolled to a single intrinsic when operation on vectors.

#contraction_accesses = [
 affine_map<() -> ()>,
 affine_map<() -> ()>,
 affine_map<() -> ()>
]
#contraction_trait = {
  indexing_maps = #contraction_accesses,
  iterator_types = [],
  kind = #iree_gpu.mma_layout<MFMA_F32_16x16x16_F16>
}
%3 = iree_gpu.multi_mma %0, %1, %2 #contraction_trait
  : vector<4xf16>, vector<4xf16> into vector<4xf32>

This operation can represent an intrinsic both in subgroup/warp and distributed (thread) abstractions through the intrinsic attribute interface. It does so semi-opaquely by including optional permutations of each MMA fragment with respect to the "canonical" MNK row major matrix multiply.

Since the canonical dimensionality of the inner dimensions are somewhat intrinsic specific, verification of this op requires only that element counts of the inner dimensions match the intrinsic.

For example, an MMT product of inner dimensions with warp semantics can be represented with the following. Permutations are only allowed for ops with subgroup semantics and must be resolved before distribution.

#contraction_accesses = [
 affine_map<(i, j, k) -> (i, k)>,
 affine_map<(i, j, k) -> (k, j)>,
 affine_map<(i, j, k) -> (i, j)>
]
#contraction_trait = {
  indexing_maps = #contraction_accesses,
  iterator_types = ["parallel", "parallel", "reduction"],
  kind = #iree_gpu.mma_layout<MFMA_F32_16x16x16_F16>,
  rhs_permutation = [1, 0]
}
%7 = iree_gpu.multi_mma %4, %5, %6 #contraction_trait
  : tensor<?x?x16x16xf16>, tensor<?x?x16x16xf16> into tensor<?x?x16x16xf32>

Motivation, Design Choices, and Pitfallslink

The idea behind this operation is to decouple the layout setting/tiling required to target certain intrinsics from the lowering to them. Because typically tiling of this sort happens on tensor operands, however the target intrinsics operate on vectors, we use this operation to bridge the gap. The choice for a shared operation is intended to ease the lowering process and allow for different transformations at different stages of the pipeline without needing to essentially clone this op.

The choice to let the inner dimensions required to compute the intrinsic be implicit based on the indexing maps was made to make this operation easier to generate and to skip the need for type conversion ops. However this comes at the expense of ease of verification for the operation. It is also implicitly linked to a lane-level parent scf.forall operation.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, DestinationStyleOpInterface, NoMemoryEffect (MemoryEffectOpInterface), TilingInterface, VectorUnrollOpInterface

Effects: MemoryEffects::Effect{}

Attributes:link

Attribute	MLIR Type	Description
`indexing_maps`	::mlir::ArrayAttr	array attribute
`iterator_types`	::mlir::ArrayAttr	Iterator type should be an enum.
`kind`	IREE::GPU::MmaInterfaceAttr	buffer-like constant attribute values
`lhs_permutation`	::mlir::DenseI64ArrayAttr	i64 dense array attribute
`rhs_permutation`	::mlir::DenseI64ArrayAttr	i64 dense array attribute
`acc_permutation`	::mlir::DenseI64ArrayAttr	i64 dense array attribute

Operands:link

Operand	Description
`lhs`	ranked tensor or vector of any type values
`rhs`	ranked tensor or vector of any type values
`acc`	ranked tensor or vector of any type values

Results:link

Result	Description
`result`	ranked tensor or vector of any type values

`iree_gpu.value_barrier` (GPU::ValueBarrierOp)link

Synchronizes workers on a value semantic tensor or vector.

Syntax:

operation ::= `iree_gpu.value_barrier` $inputs attr-dict `:` type($inputs)

This operation acts as a barrier on a value semantic SSA values (tensor or vector). It takes multiple operands and produces a value equivalent to each input. This does not have copy and/or data movement semantics and simply represents a barrier on all writes in the tensor case, and a barrier until all threads acquire the input vector in the vector case.

The inputs must be either all tensors, or all vectors.

This operation is a no-op when not present in a parallel context. This operation is pure as it only requires synchronization for the value it produces.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface)

Effects: MemoryEffects::Effect{}

Operands:link

Operand	Description
`inputs`	variadic of ranked tensor or vector of any type values

Results:link

Result	Description
`results`	variadic of ranked tensor or vector of any type values

`iree_gpu.yield` (GPU::YieldOp)link

Yield values from a region

Syntax:

operation ::= `iree_gpu.yield` attr-dict ($values^ `:` type($values))?

This operation is used to yield values from a within a region.

Traits: AlwaysSpeculatableImplTrait, HasParent<::mlir::iree_compiler::IREE::GPU::BarrierRegionOp>, ReturnLike, Terminator

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), RegionBranchTerminatorOpInterface

Effects: MemoryEffects::Effect{}

Operands:link

Operand	Description
`values`	variadic of any type

Attributeslink

ComputeBitwidthsAttrlink

Supported bitwidths for compute

Syntax:

#iree_gpu.compute_bitwidths<
  ::mlir::iree_compiler::IREE::GPU::ComputeBitwidths   # value
>

Enum cases: * fp64 (FP64) * fp32 (FP32) * fp16 (FP16) * int64 (Int64) * int32 (Int32) * int16 (Int16) * int8 (Int8)

Parameters:link

Parameter	C++ type	Description
value	`::mlir::iree_compiler::IREE::GPU::ComputeBitwidths`	an enum of type ComputeBitwidths

DataTiledMMAAttrlink

Syntax:

#iree_gpu.data_tiled_mma_layout<
  ::mlir::iree_compiler::IREE::GPU::MMAIntrinsicAttr,   # intrinsic
  int64_t,   # unroll_m
  int64_t,   # subgroups_m
  int64_t,   # unroll_n
  int64_t,   # subgroups_n
  int64_t   # unroll_k
>

This mma variant represents MMA ops with data-tiling details. The |intrinsic| field specifies which particular MMA intrinsic is targeted by the data-tiling.

The other fields default to one, and that default results in a single intrinsic equivalent to MMAAttr, while values greater than one result in wider "kernels" consisting of multiple intrinsics, with the data layout already swizzled into a tile layout that allows each intrinsic to access data at an offset that's as simple as possible a mapping from the thread ID.

Parameters:link

Parameter	C++ type	Description
intrinsic	`::mlir::iree_compiler::IREE::GPU::MMAIntrinsicAttr`
unroll_m	`int64_t`	Unrolling along the M dimension, on the same thread.
subgroups_m	`int64_t`	Unrolling along the M dimension, distributed across this many more threads.
unroll_n	`int64_t`	Unrolling along the N dimension, on the same thread.
subgroups_n	`int64_t`	Unrolling along the N dimension, distributed across this many more threads.
unroll_k	`int64_t`	Unrolling along the K dimension, on the same thread, with interleaved layout.

DerivedThreadConfigAttrlink

drive lowering of an operation by deriving thread distribution when needed.

Syntax: #iree_gpu.derived_thread_config

Lowering config for a single thread tiling level that is inferred after previous (often reduction) levels of tile + fuse. This is intended for fused operations where it is much easier to compute the tile sizes to use after previous levels of tile + fuse, rather than trying to pre-propagate tiling configs.

DotProductOpsAttrlink

Supported dot product ops

Syntax:

#iree_gpu.dotproduct_ops<
  ::mlir::iree_compiler::IREE::GPU::DotProductOps   # value
>

Enum cases: * none (None) * dp4xi8toi32 (DP4xI8ToI32)

Parameters:link

Parameter	C++ type	Description
value	`::mlir::iree_compiler::IREE::GPU::DotProductOps`	an enum of type DotProductOps

GPUPipelineOptionsAttrlink

GPU pipeline options attribute.

Syntax:

#iree_gpu.pipeline_options<
  BoolAttr,   # prefetch_shared_memory
  BoolAttr,   # no_reduce_shared_memory_bank_conflicts
  BoolAttr,   # use_igemm_convolution
  ReorderWorkgroupsStrategyAttr   # reorder_workgroups_strategy
>

This attributes describes lowering pipeline specific configuration options: * prefetch_shared_memory: Boolean option indicating whether or not to run the loop prefetching pass in the lowering pipeline. * no_reduce_shared_memory_bank_conflicts: Boolean option indicating whether or not to skip the bank conflict reduction pass in the lowering pipeline. * reorder_workgroups_strategy: Enum attribute indicating which strategy to choose for the workgroup reordering pass. Options are None, Swizzle, and Transpose.

Parameters:link

Parameter	C++ type	Description
prefetch_shared_memory	`BoolAttr`
no_reduce_shared_memory_bank_conflicts	`BoolAttr`
use_igemm_convolution	`BoolAttr`
reorder_workgroups_strategy	`ReorderWorkgroupsStrategyAttr`

IteratorTypeAttrlink

Iterator type

Syntax:

#iree_gpu.iterator_type<
  ::mlir::utils::IteratorType   # value
>

Enum cases: * parallel (parallel) * reduction (reduction)

Parameters:link

Parameter	C++ type	Description
value	`::mlir::utils::IteratorType`	an enum of type IteratorType

LaneIdAttrlink

Syntax:

#iree_gpu.lane_id<
  int64_t   # dim
>

An attribute for mapping scf.forall ops to subgroup lanes.

Parameters:link

Parameter	C++ type	Description
dim	`int64_t`

LoweringConfigAttrlink

drive lowering of an operation for gpu compilation.

Syntax:

#iree_gpu.lowering_config<
  DictionaryAttr   # attributes
>

GPU specific implementation of a lowering config. This carries just a dictionary attribute to store any relevant fields. This is the simplest form of a lowering config, offering flexibility at the cost of structure.

Parameters:link

Parameter	C++ type	Description
attributes	`DictionaryAttr`	The configured fields, including tiling levels

MMAAttrlink

Attribute describing a particular shape of matrix-multiply and accumulate instruction. Abstractly, all attributes of this type represent the following unit of arithmetic for matrices A, B, and C.

  C += A x B

Where the shape of matrix A is [m, k], B is [k, n], and C is [m, n]. This intentionally leaves the layout information abstract and uses interface methods to materialize layout information only when needed. The shape of the mma intrinsic is stored explicitly here as that information is queried frequently.

link

This mma variant describes configurations for MMA ops. The |intrinsic| field specifies which particular MMA intrinsic this refers to, with each intrinsic implicating a specific MNK shape and operand types. See IREEGPUEnums.td for the definition of the intrinsics.

Parameters:link

Parameter	C++ type	Description
intrinsic	`MMAIntrinsicAttr`
mSize	`int64_t`
nSize	`int64_t`
kSize	`int64_t`
aType	`::mlir::Type`
bType	`::mlir::Type`
cType	`::mlir::Type`

MMAIntrinsicAttrlink

Descriptor for different MMA intrinsics

Syntax:

#iree_gpu.mma_intrinsic<
  ::mlir::iree_compiler::IREE::GPU::MMAIntrinsic   # value
>

Enum cases: * MFMA_F32_16x16x4_F32 (MFMA_F32_16x16x4_F32) * MFMA_F32_16x16x16_F16 (MFMA_F32_16x16x16_F16) * MFMA_F32_32x32x8_F16 (MFMA_F32_32x32x8_F16) * MFMA_I32_16x16x16_I8 (MFMA_I32_16x16x16_I8) * MFMA_I32_32x32x8_I8 (MFMA_I32_32x32x8_I8) * MFMA_F32_16x16x8_BF16 (MFMA_F32_16x16x8_BF16) * MFMA_F32_32x32x4_BF16 (MFMA_F32_32x32x4_BF16) * MFMA_F64_16x16x4_F64 (MFMA_F64_16x16x4_F64) * MFMA_F32_16x16x16_BF16 (MFMA_F32_16x16x16_BF16) * MFMA_F32_32x32x8_BF16 (MFMA_F32_32x32x8_BF16) * MFMA_F32_16x16x32_F8E5M2FNUZ (MFMA_F32_16x16x32_F8E5M2FNUZ) * MFMA_F32_16x16x32_F8E5M2FNUZ_F8E4M3FNUZ (MFMA_F32_16x16x32_F8E5M2FNUZ_F8E4M3FNUZ) * MFMA_F32_16x16x32_F8E4M3FNUZ (MFMA_F32_16x16x32_F8E4M3FNUZ) * MFMA_F32_16x16x32_F8E4M3FNUZ_F8E5M2FNUZ (MFMA_F32_16x16x32_F8E4M3FNUZ_F8E5M2FNUZ) * MFMA_F32_32x32x16_F8E5M2FNUZ (MFMA_F32_32x32x16_F8E5M2FNUZ) * MFMA_F32_32x32x16_F8E5M2FNUZ_F8E4M3FNUZ (MFMA_F32_32x32x16_F8E5M2FNUZ_F8E4M3FNUZ) * MFMA_F32_32x32x16_F8E4M3FNUZ (MFMA_F32_32x32x16_F8E4M3FNUZ) * MFMA_F32_32x32x16_F8E4M3FNUZ_F8E5M2FNUZ (MFMA_F32_32x32x16_F8E4M3FNUZ_F8E5M2FNUZ) * MFMA_I32_16x16x32_I8 (MFMA_I32_16x16x32_I8) * MFMA_I32_32x32x16_I8 (MFMA_I32_32x32x16_I8) * WMMA_F32_16x16x16_F16 (WMMA_F32_16x16x16_F16) * WMMA_F16_16x16x16_F16 (WMMA_F16_16x16x16_F16) * WMMA_F32_16x16x16_BF16 (WMMA_F32_16x16x16_BF16) * WMMA_BF16_16x16x16_BF16 (WMMA_BF16_16x16x16_BF16) * WMMA_I32_16x16x16_I8 (WMMA_I32_16x16x16_I8) * NV_WMMA_F32_16x16x16_F16 (NV_WMMA_F32_16x16x16_F16) * NV_WMMA_F16_16x16x16_F16 (NV_WMMA_F16_16x16x16_F16)

Parameters:link

Parameter	C++ type	Description
value	`::mlir::iree_compiler::IREE::GPU::MMAIntrinsic`	an enum of type MMAIntrinsic

MMAOpsArrayAttrlink

Syntax:

#iree_gpu.mma_ops<
  ::llvm::ArrayRef<MMAAttr>   # value
>

Parameters:link

Parameter	C++ type	Description
value	`::llvm::ArrayRef<MMAAttr>`

MMAScheduleAttrlink

Syntax:

#iree_gpu.mma_schedule<
  ::mlir::iree_compiler::IREE::GPU::MmaInterfaceAttr,   # intrinsic
  int64_t,   # subgroup_m_count
  int64_t   # subgroup_n_count
>

A schedule of MMA intrinsic instruction and various levels of tile sizes to solve a specific contraction problem.

Parameters:link

Parameter	C++ type	Description
intrinsic	`::mlir::iree_compiler::IREE::GPU::MmaInterfaceAttr`
subgroup_m_count	`int64_t`
subgroup_n_count	`int64_t`

ReorderWorkgroupsStrategyAttrlink

Strategy for workgroup reordering

Syntax:

#iree_gpu.reorder_workgroups_strategy<
  ::mlir::iree_compiler::IREE::GPU::ReorderWorkgroupsStrategy   # value
>

Enum cases: * None (None) * Transpose (Transpose)

Parameters:link

Parameter	C++ type	Description
value	`::mlir::iree_compiler::IREE::GPU::ReorderWorkgroupsStrategy`	an enum of type ReorderWorkgroupsStrategy

StorageBitwidthsAttrlink

Supported bitwidths for storage

Syntax:

#iree_gpu.storage_bitwidths<
  ::mlir::iree_compiler::IREE::GPU::StorageBitwidths   # value
>

Enum cases: * b64 (B64) * b32 (B32) * b16 (B16) * b8 (B8)

Parameters:link

Parameter	C++ type	Description
value	`::mlir::iree_compiler::IREE::GPU::StorageBitwidths`	an enum of type StorageBitwidths

SubgroupOpsAttrlink

Supported subgroup ops

Syntax:

#iree_gpu.subgroup_ops<
  ::mlir::iree_compiler::IREE::GPU::SubgroupOps   # value
>

Enum cases: * none (None) * shuffle (Shuffle) * arithmetic (Arithmetic)

Parameters:link

Parameter	C++ type	Description
value	`::mlir::iree_compiler::IREE::GPU::SubgroupOps`	an enum of type SubgroupOps

TargetAttrlink

Full GPU target attribute

Syntax:

#iree_gpu.target<
  ::llvm::StringRef,   # arch
  ::llvm::StringRef,   # features
  TargetWgpAttr,   # wgp
  TargetChipAttr   # chip
>

This attributes describes a full GPU target. It contains a few fields: * The canonical target architecture for compilation, e.g., sm_80 for cuda, gfx942 for hip * A TargetWgpAttr describing the GPU features and limits in a single GPU workgroup processor (WGP), that is, AMD compute unit or NVIDIA streaming multiprocessor * An optional TargetChipAttr describing GPU features for the final chip or product, e.g., wgp count

Parameters:link

Parameter	C++ type	Description
arch	`::llvm::StringRef`	target architecture
features	`::llvm::StringRef`	target features
wgp	`TargetWgpAttr`
chip	`TargetChipAttr`

TargetChipAttrlink

Chip level target description

Syntax:

#iree_gpu.target_chip<
  uint32_t,   # wgp_count
  DictionaryAttr   # extra
>

This attribute contains hardware features/limits at a single GPU chip level. Here a GPU chip means the hardware functionality scope where the whole software compute grid is scheduled onto. A chip typically contains many AMD compute units or NVIDIA streaming multiprocessors; it's the final SKU.

Parameters:link

Parameter	C++ type	Description
wgp_count	`uint32_t`
extra	`DictionaryAttr`

TargetWgpAttrlink

Workgroup processor level target description

Syntax:

#iree_gpu.target_wgp<
  ComputeBitwidthsAttr,   # compute
  StorageBitwidthsAttr,   # storage
  SubgroupOpsAttr,   # subgroup
  DotProductOpsAttr,   # dot
  MMAOpsArrayAttr,   # mma
  DenseI32ArrayAttr,   # subgroup_size_choices
  DenseI32ArrayAttr,   # max_workgroup_sizes
  int32_t,   # max_thread_count_per_workgroup
  int32_t,   # max_workgroup_memory_bytes
  DenseI32ArrayAttr,   # max_workgroup_counts
  std::optional<int32_t>,   # max_load_instruction_bits
  std::optional<int32_t>,   # simds_per_wgp
  std::optional<int32_t>,   # vgpr_space_bits
  DictionaryAttr   # extra
>

This attribute contains hardware features/limits at a single GPU workgroup processor (WGP) level. Here a GPU workgroup processor means the basic hardware functionality unit where a software workgroup is scheduled onto; that is, a compute unit for AMD GPUs or a streaming multiprocessor for NVIDIA GPUs.

Parameters:link

Parameter	C++ type	Description
compute	`ComputeBitwidthsAttr`
storage	`StorageBitwidthsAttr`
subgroup	`SubgroupOpsAttr`
dot	`DotProductOpsAttr`
mma	`MMAOpsArrayAttr`
subgroup_size_choices	`DenseI32ArrayAttr`
max_workgroup_sizes	`DenseI32ArrayAttr`
max_thread_count_per_workgroup	`int32_t`
max_workgroup_memory_bytes	`int32_t`
max_workgroup_counts	`DenseI32ArrayAttr`
max_load_instruction_bits	`std::optional<int32_t>`
simds_per_wgp	`std::optional<int32_t>`
vgpr_space_bits	`std::optional<int32_t>`
extra	`DictionaryAttr`

VirtualMMAAttrlink

Syntax:

#iree_gpu.virtual_mma_layout<
  ::mlir::iree_compiler::IREE::GPU::VirtualMMAIntrinsicAttr   # intrinsic
>

This mma variant represents "virtual" MMA ops that has modification to its native layouts by unrollK and/or interleave reads. The |intrinsic| field represents different kinds of "Virtual" MMA Ops we found helpful.

These interleaving and/or unrolling changes in the layout is especially useful to coalesce reads from shared memory to register or align layouts in a chained-matmul operation.

Parameters:link

Parameter	C++ type	Description
intrinsic	`::mlir::iree_compiler::IREE::GPU::VirtualMMAIntrinsicAttr`

VirtualMMAIntrinsicAttrlink

Descriptor for different Virtual MMA intrinsics

Syntax:

#iree_gpu.virtual_mma_intrinsic<
  ::mlir::iree_compiler::IREE::GPU::VirtualMMAIntrinsic   # value
>

Enum cases: * VMFMA_F32_16x16x32_F16 (VMFMA_F32_16x16x32_F16) * VMFMA_F32_32x32x16_F16 (VMFMA_F32_32x32x16_F16) * VMFMA_F32_16x16x32_F8E4M3FNUZ (VMFMA_F32_16x16x32_F8E4M3FNUZ) * VMFMA_F32_32x32x16_F8E4M3FNUZ (VMFMA_F32_32x32x16_F8E4M3FNUZ)

Parameters:link

Parameter	C++ type	Description
value	`::mlir::iree_compiler::IREE::GPU::VirtualMMAIntrinsic`	an enum of type VirtualMMAIntrinsic

Enumslink

ComputeBitwidthslink

Supported bitwidths for compute

Cases:link

Symbol	Value	String
FP64	`1`	fp64
FP32	`2`	fp32
FP16	`4`	fp16
Int64	`8`	int64
Int32	`16`	int32
Int16	`32`	int16
Int8	`64`	int8

DotProductOpslink

Supported dot product ops

Cases:link

Symbol	Value	String
None	`0`	none
DP4xI8ToI32	`1`	dp4xi8toi32

MMAFragmentlink

Descriptor for a particular fragment of an MMA operation

Cases:link

Symbol	Value	String
Lhs	`0`	Lhs
Rhs	`1`	Rhs
Acc	`2`	Acc

MMAIntrinsiclink

Descriptor for different MMA intrinsics

Cases:link

Symbol	Value	String
MFMA_F32_16x16x4_F32	`4112`	MFMA_F32_16x16x4_F32
MFMA_F32_16x16x16_F16	`4128`	MFMA_F32_16x16x16_F16
MFMA_F32_32x32x8_F16	`4129`	MFMA_F32_32x32x8_F16
MFMA_I32_16x16x16_I8	`4288`	MFMA_I32_16x16x16_I8
MFMA_I32_32x32x8_I8	`4289`	MFMA_I32_32x32x8_I8
MFMA_F32_16x16x8_BF16	`4384`	MFMA_F32_16x16x8_BF16
MFMA_F32_32x32x4_BF16	`4385`	MFMA_F32_32x32x4_BF16
MFMA_F64_16x16x4_F64	`4352`	MFMA_F64_16x16x4_F64
MFMA_F32_16x16x16_BF16	`4640`	MFMA_F32_16x16x16_BF16
MFMA_F32_32x32x8_BF16	`4641`	MFMA_F32_32x32x8_BF16
MFMA_F32_16x16x32_F8E5M2FNUZ	`4656`	MFMA_F32_16x16x32_F8E5M2FNUZ
MFMA_F32_16x16x32_F8E5M2FNUZ_F8E4M3FNUZ	`4657`	MFMA_F32_16x16x32_F8E5M2FNUZ_F8E4M3FNUZ
MFMA_F32_16x16x32_F8E4M3FNUZ	`4658`	MFMA_F32_16x16x32_F8E4M3FNUZ
MFMA_F32_16x16x32_F8E4M3FNUZ_F8E5M2FNUZ	`4659`	MFMA_F32_16x16x32_F8E4M3FNUZ_F8E5M2FNUZ
MFMA_F32_32x32x16_F8E5M2FNUZ	`4660`	MFMA_F32_32x32x16_F8E5M2FNUZ
MFMA_F32_32x32x16_F8E5M2FNUZ_F8E4M3FNUZ	`4661`	MFMA_F32_32x32x16_F8E5M2FNUZ_F8E4M3FNUZ
MFMA_F32_32x32x16_F8E4M3FNUZ	`4662`	MFMA_F32_32x32x16_F8E4M3FNUZ
MFMA_F32_32x32x16_F8E4M3FNUZ_F8E5M2FNUZ	`4663`	MFMA_F32_32x32x16_F8E4M3FNUZ_F8E5M2FNUZ
MFMA_I32_16x16x32_I8	`4800`	MFMA_I32_16x16x32_I8
MFMA_I32_32x32x16_I8	`4801`	MFMA_I32_32x32x16_I8
WMMA_F32_16x16x16_F16	`6176`	WMMA_F32_16x16x16_F16
WMMA_F16_16x16x16_F16	`6177`	WMMA_F16_16x16x16_F16
WMMA_F32_16x16x16_BF16	`6178`	WMMA_F32_16x16x16_BF16
WMMA_BF16_16x16x16_BF16	`6179`	WMMA_BF16_16x16x16_BF16
WMMA_I32_16x16x16_I8	`6336`	WMMA_I32_16x16x16_I8
NV_WMMA_F32_16x16x16_F16	`8224`	NV_WMMA_F32_16x16x16_F16
NV_WMMA_F16_16x16x16_F16	`8225`	NV_WMMA_F16_16x16x16_F16

MMAScopelink

Descriptor for a particular scope of an MMA operation

Cases:link

Symbol	Value	String
Workgroup	`0`	Workgroup
Subgroup	`1`	Subgroup

ReorderWorkgroupsStrategylink

Strategy for workgroup reordering

Cases:link

Symbol	Value	String
None	`0`	None
Transpose	`1`	Transpose

StorageBitwidthslink

Supported bitwidths for storage

Cases:link

Symbol	Value	String
B64	`1`	b64
B32	`2`	b32
B16	`4`	b16
B8	`8`	b8

SubgroupOpslink

Supported subgroup ops

Cases:link

Symbol	Value	String
None	`0`	none
Shuffle	`1`	shuffle
Arithmetic	`2`	arithmetic

TilingLevellink

Descriptor for tiling levels for GPU lowering configs

Cases:link

Symbol	Value	String
Workgroup	`0`	Workgroup
Reduction	`1`	Reduction
PartialReduction	`2`	PartialReduction
Thread	`3`	Thread
Subgroup	`4`	Subgroup
Lane	`5`	Lane

VirtualMMAIntrinsiclink

Descriptor for different Virtual MMA intrinsics

Cases:link

Symbol	Value	String
VMFMA_F32_16x16x32_F16	`0`	VMFMA_F32_16x16x32_F16
VMFMA_F32_32x32x16_F16	`1`	VMFMA_F32_32x32x16_F16
VMFMA_F32_16x16x32_F8E4M3FNUZ	`2`	VMFMA_F32_16x16x32_F8E4M3FNUZ
VMFMA_F32_32x32x16_F8E4M3FNUZ	`3`	VMFMA_F32_32x32x16_F8E4M3FNUZ

IteratorTypelink

Iterator type

Cases:link

Symbol	Value	String
parallel	`0`	parallel
reduction	`1`	reduction

'iree_gpu' Dialectlink

Operationslink

iree_gpu.barrier_region (GPU::BarrierRegionOp)link

Operands:link

Results:link

iree_gpu.multi_mma (GPU::MultiMmaOp)link

Motivation, Design Choices, and Pitfallslink

Attributes:link

Operands:link

Results:link

iree_gpu.value_barrier (GPU::ValueBarrierOp)link

Operands:link

Results:link

iree_gpu.yield (GPU::YieldOp)link

Operands:link

Attributeslink

ComputeBitwidthsAttrlink

Parameters:link

DataTiledMMAAttrlink

Parameters:link

DerivedThreadConfigAttrlink

DotProductOpsAttrlink

Parameters:link

GPUPipelineOptionsAttrlink

Parameters:link

IteratorTypeAttrlink

Parameters:link

LaneIdAttrlink

Parameters:link

LoweringConfigAttrlink

Parameters:link

MMAAttrlink

link

Parameters:link

MMAIntrinsicAttrlink

Parameters:link

MMAOpsArrayAttrlink

Parameters:link

MMAScheduleAttrlink

Parameters:link

ReorderWorkgroupsStrategyAttrlink

Parameters:link

StorageBitwidthsAttrlink

Parameters:link

SubgroupOpsAttrlink

Parameters:link

TargetAttrlink

Parameters:link

TargetChipAttrlink

Parameters:link

TargetWgpAttrlink

Parameters:link

VirtualMMAAttrlink

Parameters:link

VirtualMMAIntrinsicAttrlink

Parameters:link

Enumslink

ComputeBitwidthslink

Cases:link

DotProductOpslink

Cases:link

MMAFragmentlink

Cases:link

MMAIntrinsiclink

Cases:link

MMAScopelink

Cases:link

ReorderWorkgroupsStrategylink

Cases:link

StorageBitwidthslink

Cases:link

SubgroupOpslink

Cases:link

TilingLevellink

Cases:link

VirtualMMAIntrinsiclink

Cases:link

IteratorTypelink

Cases:link

`iree_gpu.barrier_region` (GPU::BarrierRegionOp)link

`iree_gpu.multi_mma` (GPU::MultiMmaOp)link

`iree_gpu.value_barrier` (GPU::ValueBarrierOp)link

`iree_gpu.yield` (GPU::YieldOp)link