openxla/iree issues and pull requests

#15968 - [CPU] Remove special distribution tile sizes from setting matmul config.

Pull Request - State: closed - Opened by hanhanW 11 months ago - 2 comments
Labels: benchmarks:x86_64, benchmarks:android-cpu

#15967 - [CPU][NFC] Update variable names to be camelCase.

Pull Request - State: closed - Opened by hanhanW 11 months ago

#15966 - [compiler] NFC: Move op equivalence into a file in Utils/

Pull Request - State: closed - Opened by antiagainst 11 months ago

#15966 - [compiler] NFC: Move op equivalence into a file in Utils/

Pull Request - State: closed - Opened by antiagainst 11 months ago

#15965 - Bad unrolling/vectorization for `linalg.generic` implementing group-dequantization-reduction

Issue - State: open - Opened by bjacob 11 months ago
Labels: codegen, performance ⚡

#15965 - Bad unrolling/vectorization for `linalg.generic` implementing group-dequantization-reduction

Issue - State: open - Opened by bjacob 11 months ago
Labels: codegen, performance ⚡

#15964 - [GlobalOpt] Add quantized matmul reassociation support for f16 types

Pull Request - State: closed - Opened by Max191 11 months ago - 1 comment

#15963 - [GPU] Simplify gpu warp reduction lowering

Pull Request - State: closed - Opened by kuhar 11 months ago - 5 comments
Labels: benchmarks:cuda, benchmarks:android-gpu, benchmarks:vulkan-nvidia

#15963 - [GPU] Simplify gpu warp reduction lowering

Pull Request - State: closed - Opened by kuhar 11 months ago - 5 comments
Labels: benchmarks:cuda, benchmarks:android-gpu, benchmarks:vulkan-nvidia

#15962 - Add Falcon7b int4 to CI benchmarks

Pull Request - State: closed - Opened by mariecwhite 11 months ago - 5 comments
Labels: benchmarks:x86_64, benchmarks:comp-stats, benchmarks:android-cpu

#15962 - Add Falcon7b int4 to CI benchmarks

Pull Request - State: open - Opened by mariecwhite 11 months ago - 5 comments
Labels: benchmarks:x86_64, benchmarks:comp-stats, benchmarks:android-cpu

#15961 - Hoist constexpr from nested control flow regions

Pull Request - State: open - Opened by pzread 11 months ago - 15 comments
Labels: benchmarks:cuda, benchmarks:x86_64, benchmarks:android-cpu, benchmarks:android-gpu, benchmarks:vulkan-nvidia

#15961 - Hoist constexpr from nested control flow regions

Pull Request - State: open - Opened by pzread 11 months ago - 15 comments
Labels: benchmarks:cuda, benchmarks:x86_64, benchmarks:android-cpu, benchmarks:android-gpu, benchmarks:vulkan-nvidia

#15960 - Add support for serializing complex resource attributes

Pull Request - State: closed - Opened by qedawkins 11 months ago

#15959 - Align the Flow::DispatchTensorStoreOp verifier with tensor.insert_slice

Pull Request - State: closed - Opened by qedawkins 11 months ago

#15958 - [spirv] NFC: Cleanup and reorder functions in SPIR-V passes

Pull Request - State: closed - Opened by antiagainst 11 months ago
Labels: codegen/spirv, cleanup 🧹

#15957 - [Codegen] Switch to Util helper for getting bit widths

Pull Request - State: closed - Opened by qedawkins 11 months ago

#15956 - Bump torch-mlir to 9c655d0bfb166785dc17e51b0afa1f937c227cef

Pull Request - State: closed - Opened by qedawkins 11 months ago

#15955 - Add a pass for parameterizing large constants

Pull Request - State: closed - Opened by qedawkins 11 months ago - 2 comments

#15954 - Integrate llvm/llvm-project@395f9ce30e88 (2023-12-16)

Pull Request - State: closed - Opened by antiagainst 11 months ago - 1 comment

#15953 - Add folding arithmetic extensions

Pull Request - State: open - Opened by erman-gurses 11 months ago

#15953 - Add folding arithmetic extensions

Pull Request - State: closed - Opened by erman-gurses 11 months ago - 2 comments

#15952 - Updating HAL passes to use tablegen.

Pull Request - State: closed - Opened by benvanik 11 months ago
Labels: compiler/dialects, cleanup 🧹

#15951 - [CodeGen] GPU Subgroup Reduction Pipeline for MatVec

Issue - State: open - Opened by kuhar 11 months ago - 1 comment
Labels: codegen, codegen/spirv, codegen/rocm

#15951 - [CodeGen] GPU Subgroup Reduction Pipeline for MatVec

Issue - State: open - Opened by kuhar 11 months ago - 1 comment
Labels: codegen, codegen/spirv, codegen/rocm

#15951 - [CodeGen] GPU Subgroup Reduction Pipeline for MatVec

Issue - State: open - Opened by kuhar 11 months ago - 1 comment
Labels: codegen, codegen/spirv, codegen/rocm

#15950 - [spirv] NFC: sort pass pipelines update comment headers

Pull Request - State: closed - Opened by antiagainst 11 months ago
Labels: codegen/spirv, cleanup 🧹

#15950 - [spirv] NFC: sort pass pipelines update comment headers

Pull Request - State: closed - Opened by antiagainst 11 months ago
Labels: codegen/spirv, cleanup 🧹

#15950 - [spirv] NFC: sort pass pipelines update comment headers

Pull Request - State: closed - Opened by antiagainst 11 months ago
Labels: codegen/spirv, cleanup 🧹

#15949 - Vulkan validation layer errors in tests / on different devices

Issue - State: open - Opened by ScottTodd 11 months ago
Labels: bug 🐞, hal/vulkan

#15948 - `arith.truncf: f32 -> bf16` is lowered to "software" bf16 implementation

Issue - State: closed - Opened by bjacob 11 months ago - 1 comment
Labels: codegen, performance ⚡

#15947 - Issue with Only Vulkan-Related Test Codes Failing on Android Device

Issue - State: open - Opened by hiroto01230 11 months ago - 19 comments
Labels: platform/android 🤖, hal/vulkan, support

#15946 - Add a f32-to-bf16 demotion pass for matmuls, and new compiler flag `--iree-global-opt-enable-demote-contraction-inputs-to-bf16`

Pull Request - State: closed - Opened by bjacob 11 months ago

#15946 - Add a f32-to-bf16 demotion pass for matmuls, and new compiler flag `--iree-global-opt-enable-demote-contraction-inputs-to-bf16`

Pull Request - State: open - Opened by bjacob 11 months ago

#15945 - [Flow] Enable fusion of `tensor.unpack` with non-intersecting reductions

Pull Request - State: closed - Opened by Max191 11 months ago - 2 comments
Labels: benchmarks:cuda, benchmarks:x86_64, benchmarks:comp-stats, benchmarks:android-cpu, benchmarks:android-gpu, benchmarks:vulkan-nvidia

#15944 - [NFC][Codegen] Move Codegen/Dialect to Codegen/Dialect/Codegen/IR

Pull Request - State: open - Opened by Groverkss 11 months ago - 1 comment

#15944 - [NFC][Codegen] Move Codegen/Dialect to Codegen/Dialect/Codegen/IR

Pull Request - State: closed - Opened by Groverkss 11 months ago - 1 comment

#15944 - [NFC][Codegen] Move Codegen/Dialect to Codegen/Dialect/Codegen/IR

Pull Request - State: closed - Opened by Groverkss 11 months ago - 1 comment

#15943 - [RFC] e2e microkernel pipeline on ROCM backend.

Pull Request - State: closed - Opened by raikonenfnu 11 months ago - 3 comments

#15942 - [Transform] Add transform.iree.gpu_distribute_shared_memory_copy

Pull Request - State: closed - Opened by Groverkss 11 months ago

#15942 - [Transform] Add transform.iree.gpu_distribute_shared_memory_copy

Pull Request - State: closed - Opened by Groverkss 11 months ago

#15942 - [Transform] Add transform.iree.gpu_distribute_shared_memory_copy

Pull Request - State: closed - Opened by Groverkss 11 months ago

#15941 - [LLVMGPU] Add multi-row vector reduction configuration for ROCm

Pull Request - State: closed - Opened by kuhar 11 months ago - 1 comment

#15941 - [LLVMGPU] Add multi-row vector reduction configuration for ROCm

Pull Request - State: closed - Opened by kuhar 11 months ago - 1 comment

#15940 - [SPIRV] Switch SPIRVTile to use common methods.

Pull Request - State: open - Opened by hanhanW 11 months ago - 2 comments

#15940 - [SPIRV] Switch SPIRVTile to use common methods.

Pull Request - State: open - Opened by hanhanW 11 months ago - 2 comments

#15940 - [SPIRV] Switch tiling on reductions to use common helpers.

Pull Request - State: closed - Opened by hanhanW 11 months ago - 3 comments

#15939 - [CodeGen] Switch tiling only cases to use scf::tileUsingSCFForOp

Pull Request - State: closed - Opened by hanhanW 11 months ago - 3 comments

#15939 - [CodeGen] Switch tiling only cases to use scf::tileUsingSCFForOp

Pull Request - State: closed - Opened by hanhanW 11 months ago - 3 comments

#15938 - [CodeGen][NFC] Rename tileLinalgOpsWithFilter to distributeLinalgOpsWithFilter.

Pull Request - State: closed - Opened by hanhanW 11 months ago

#15938 - [CodeGen][NFC] Rename tileLinalgOpsWithFilter to distributeLinalgOpsWithFilter.

Pull Request - State: closed - Opened by hanhanW 11 months ago

#15938 - [CodeGen][NFC] Rename tileLinalgOpsWithFilter to distributeLinalgOpsWithFilter.

Pull Request - State: closed - Opened by hanhanW 11 months ago

#15937 - [HIP] Adds support for native executable and cache

Pull Request - State: closed - Opened by nithinsubbiah 11 months ago

#15937 - [HIP] Adds support for native executable and cache

Pull Request - State: closed - Opened by nithinsubbiah 11 months ago

#15936 - The polynomial approximation for f16 math.powf generates NAN and INF

Issue - State: open - Opened by hanhanW 11 months ago - 8 comments
Labels: codegen, codegen/llvm

#15935 - [spirv] Provide same entry point list in variants when linking

Pull Request - State: closed - Opened by antiagainst 11 months ago - 2 comments
Labels: codegen/spirv

#15934 - Set --iree-llvmcpu-link-embedded=false when --iree-llvmcpu-sanitize=address is enabled.(#15519)

Pull Request - State: open - Opened by TSWorld1314 11 months ago - 6 comments

#15931 - [CPU] Retire CPUDoubleTilingPadExpert pipeline.

Pull Request - State: closed - Opened by hanhanW 11 months ago - 2 comments
Labels: benchmarks:x86_64, benchmarks:android-cpu

#15931 - [CPU] Retire CPUDoubleTilingPadExpert pipeline.

Pull Request - State: closed - Opened by hanhanW 11 months ago - 2 comments
Labels: benchmarks:x86_64, benchmarks:android-cpu

#15931 - [CPU] Retire CPUDoubleTilingPadExpert pipeline.

Pull Request - State: closed - Opened by hanhanW 11 months ago - 2 comments
Labels: benchmarks:x86_64, benchmarks:android-cpu

#15928 - [CPU] Unify distribution tile size selection heuristic.

Pull Request - State: closed - Opened by hanhanW 11 months ago - 2 comments
Labels: benchmarks:x86_64, benchmarks:android-cpu

#15928 - [CPU] Unify distribution tile size selection heuristic.

Pull Request - State: open - Opened by hanhanW 11 months ago - 2 comments
Labels: benchmarks:x86_64, benchmarks:android-cpu

#15928 - [CPU] Unify distribution tile size selection heuristic.

Pull Request - State: open - Opened by hanhanW 11 months ago - 2 comments
Labels: benchmarks:x86_64, benchmarks:android-cpu

#15927 - [CPU] Add support for converting math.powf from fp16 to fp32.

Pull Request - State: open - Opened by hanhanW 11 months ago

#15927 - [CPU] Add support for converting math.powf from fp16 to fp32.

Pull Request - State: open - Opened by hanhanW 11 months ago

#15926 - Mark tensorflow_resnet.ipynb as xfail.

Pull Request - State: closed - Opened by ScottTodd 11 months ago
Labels: integrations/tensorflow

#15926 - Mark tensorflow_resnet.ipynb as xfail.

Pull Request - State: closed - Opened by ScottTodd 11 months ago
Labels: integrations/tensorflow

#15925 - Fix experimental/web[gpu] builds after HAL changes.

Pull Request - State: closed - Opened by ScottTodd 11 months ago

#15924 - [LLVMGPU] Re-enable multidim distribution on gridDimension.

Pull Request - State: open - Opened by raikonenfnu 11 months ago - 1 comment

#15923 - Switch JAX pjrt-plugin link.

Pull Request - State: open - Opened by ScottTodd 11 months ago
Labels: documentation

#15923 - Switch JAX pjrt-plugin link.

Pull Request - State: closed - Opened by ScottTodd 11 months ago - 2 comments
Labels: documentation

#15922 - [LinalgExt] Switch distribution tests to use transform dialect.

Pull Request - State: closed - Opened by hanhanW 11 months ago

#15921 - [LinalgExt] Delete LinalgExt tiling patterns and passes.

Pull Request - State: closed - Opened by hanhanW 11 months ago - 1 comment

#15921 - [LinalgExt] Delete LinalgExt tiling patterns and passes.

Pull Request - State: open - Opened by hanhanW 11 months ago - 1 comment

#15920 - [onnx] Add ONNX importer and iree-import-onnx tool to compiler package.

Pull Request - State: closed - Opened by stellaraccident 11 months ago

#15919 - Removing transfer_range from the HAL device vtable.

Pull Request - State: closed - Opened by benvanik 11 months ago
Labels: hal/api, cleanup 🧹

#15919 - Removing transfer_range from the HAL device vtable.

Pull Request - State: closed - Opened by benvanik 11 months ago
Labels: hal/api, cleanup 🧹

#15918 - Add tests for iree-run-module inputs.

Pull Request - State: closed - Opened by ScottTodd 11 months ago - 4 comments
Labels: runtime/tools

#15918 - Add tests for iree-run-module inputs.

Pull Request - State: closed - Opened by ScottTodd 11 months ago - 4 comments
Labels: runtime/tools

#15917 - [CPU] Add missing passes for ukernel fallback

Pull Request - State: closed - Opened by pzread 11 months ago - 1 comment
Labels: benchmarks:x86_64, benchmarks:android-cpu

#15916 - Replacing hal.ex.shared_device with hal.devices.* ops.

Pull Request - State: closed - Opened by benvanik 11 months ago - 4 comments
Labels: compiler/dialects, hal/api

#15916 - Replacing hal.ex.shared_device with hal.devices.* ops.

Pull Request - State: closed - Opened by benvanik 11 months ago - 4 comments
Labels: compiler/dialects, hal/api

#15915 - Quote path to Python executable for lit tests on Windows.

Pull Request - State: closed - Opened by ScottTodd 11 months ago
Labels: platform/windows 🚪

#15914 - [DNS] I15760 E2E patches

Pull Request - State: open - Opened by pzread 11 months ago

#15914 - [DNS] I15760 E2E patches

Pull Request - State: open - Opened by pzread 11 months ago

#15913 - Fix producer-consumer graph in `ConstExprAnalysis`

Pull Request - State: closed - Opened by pzread 11 months ago - 1 comment
Labels: benchmarks:cuda, benchmarks:x86_64, benchmarks:comp-stats, benchmarks:android-cpu, benchmarks:android-gpu, benchmarks:vulkan-nvidia