Files
TheRock/rocm-libraries/shared/rocroller/lib/source/KernelGraph/Transformations/AddF6LDSPadding.cpp
T
derek.holloway 7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
Flatten subprojects into a single base monorepo
2026-09-02 16:34:47 -07:00

133 lines
5.2 KiB
C++

// Copyright Advanced Micro Devices, Inc., or its affiliates.
// SPDX-License-Identifier: MIT
#include <rocRoller/CodeGen/Utils.hpp>
#include <rocRoller/KernelGraph/Transforms/AddF6LDSPadding.hpp>
#include <rocRoller/KernelGraph/Visitors.hpp>
namespace rocRoller
{
namespace KernelGraph
{
inline bool isMatrixA(CoordinateGraph::MacroTile const& macTile)
{
return macTile.layoutType == LayoutType::MATRIX_A;
}
inline bool isMatrixB(CoordinateGraph::MacroTile const& macTile)
{
return macTile.layoutType == LayoutType::MATRIX_B;
}
CoordinateGraph::MacroTile paddedMacroTile(CoordinateGraph::MacroTile& macroTile,
uint paddingBytes)
{
uint const elementBits = 6;
uint const packing = 16;
auto fastMovingDim = isMatrixA(macroTile) ? 0 : 1;
auto padElements = macroTile.sizes[fastMovingDim] / packing;
std::vector<uint> padBytesA = {padElements * paddingBytes, 0};
std::vector<uint> padBytesB = {0, padElements * paddingBytes};
return CoordinateGraph::MacroTile(macroTile,
isMatrixA(macroTile) ? padBytesA : padBytesB);
}
auto findCandidates(KernelGraph const& graph)
{
auto isLoadLDSTileOfTransposedTile = [&](int tag) {
auto load = graph.control.get<ControlGraph::LoadLDSTile>(tag);
if(!load)
return false;
auto [macTileTag, macTile] = graph.getDimension<CoordinateGraph::MacroTile>(tag);
auto elementBits = DataTypeInfo::Get(load->varType).elementBits;
return elementBits == 6 && load->isTransposedTile
&& (isMatrixA(macTile) || isMatrixB(macTile));
};
return graph.control.findElements(isLoadLDSTileOfTransposedTile)
.to<std::unordered_set>();
}
KernelGraph AddF6LDSPadding::apply(KernelGraph const& graph)
{
auto const& arch = m_context->targetArchitecture();
auto candidates = findCandidates(graph);
// Return unchanged graph if padding is not needed or if no LoadLDSTile of transposed tile found.
if(!arch.HasCapability(GPUCapability::DSReadTransposeB6PaddingBytes)
|| std::ranges::empty(candidates))
{
return graph;
}
const auto paddingBytes
= arch.GetCapability(GPUCapability::DSReadTransposeB6PaddingBytes);
auto kgraph{graph};
std::unordered_set<int> visitedCoordinates;
for(auto tag : candidates)
{
auto connections = graph.mapper.getConnections(tag);
for(auto conn : connections)
{
auto coordTag = conn.coordinate;
if(visitedCoordinates.contains(coordTag))
continue;
visitedCoordinates.insert(coordTag);
auto maybeLDS = graph.coordinates.get<CoordinateGraph::LDS>(coordTag);
if(maybeLDS)
{
auto ldsTag = coordTag;
for(auto conn : graph.mapper.getCoordinateConnections(ldsTag))
{
auto coordTag = conn.coordinate;
auto controlTag = conn.control;
auto storeLDSTile
= graph.control.get<ControlGraph::StoreLDSTile>(controlTag);
if(storeLDSTile)
{
auto storeLDSTileTag{controlTag};
auto [macroTileTag, macroTile]
= graph.getDimension<CoordinateGraph::MacroTile>(
storeLDSTileTag);
Log::debug("Padding Tile {}", macroTileTag);
kgraph.coordinates.setElement(
macroTileTag, paddedMacroTile(macroTile, paddingBytes));
for(auto conn : graph.mapper.getConnections(storeLDSTileTag))
{
auto maybeMacroTile
= graph.coordinates.get<CoordinateGraph::MacroTile>(
conn.coordinate);
if(maybeMacroTile)
{
Log::debug("Padding Tile {}", conn.coordinate);
kgraph.coordinates.setElement(
conn.coordinate,
paddedMacroTile(*maybeMacroTile, paddingBytes));
}
}
}
}
}
}
}
return kgraph;
}
}
}