Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / openxla/iree issues and pull requests

#16204 - Fix "remove unused files" step in `build_all.yml`.

Pull Request - State: open - Opened by ScottTodd 10 months ago
Labels: infrastructure

#16203 - Convert more CI jobs to be package-based

Issue - State: open - Opened by ScottTodd 10 months ago
Labels: infrastructure

#16202 - [GlobalOpt] Fix extract_slice transpose propagation bug

Pull Request - State: open - Opened by Max191 10 months ago

#16201 - Drop slow `vmvx_resnet_50_int8` test.

Pull Request - State: open - Opened by ScottTodd 10 months ago
Labels: infrastructure, integrations/tensorflow

#16200 - Switch lightweight CI jobs to use GitHub-hosted runners.

Pull Request - State: open - Opened by ScottTodd 10 months ago
Labels: infrastructure

#16199 - [compiler] Mesh -> IREE conversion

Pull Request - State: open - Opened by sogartar 10 months ago - 3 comments

#16198 - Fuse mmt4d ukernel with consumer and get perfect codegen thanks to store-to-load forwarding

Issue - State: open - Opened by bjacob 10 months ago - 1 comment
Labels: performance ⚡, codegen/llvm

#16197 - [CPU] Remove "clamping" when calculating vec tile sizes

Pull Request - State: open - Opened by banach-space 10 months ago - 1 comment
Labels: benchmarks:x86_64, benchmarks:comp-stats, benchmarks:android-cpu

#16196 - Optimize unit dims and simplify pack unpack try unroll 3d transpose perf

Pull Request - State: open - Opened by hanhanW 10 months ago - 1 comment
Labels: benchmarks:x86_64

#16195 - Bump LLVM to llvm/llvm-project@9261ab708e37

Pull Request - State: open - Opened by MaheshRavishankar 10 months ago - 1 comment
Labels: benchmarks:cuda, benchmarks:x86_64, benchmarks:comp-stats, benchmarks:android-cpu, benchmarks:android-gpu, benchmarks:vulkan-nvidia

#16194 - Reenable benchmarks that were previously failing

Pull Request - State: open - Opened by mariecwhite 10 months ago
Labels: benchmarks:cuda, benchmarks:x86_64, benchmarks:android-cpu, benchmarks:android-gpu, benchmarks:vulkan-nvidia

#16193 - [CPU] Add patterns to drop unit dimensions at vector level

Pull Request - State: closed - Opened by dcaballe 10 months ago - 8 comments
Labels: benchmarks:x86_64, benchmarks:comp-stats, benchmarks:android-cpu

#16192 - Reland "Freeze all statuses emitted by calls into dynamic modules"

Pull Request - State: closed - Opened by qedawkins 10 months ago - 2 comments

#16191 - Revert "Freeze all statuses emitted by calls into dynamic modules

Pull Request - State: closed - Opened by qedawkins 10 months ago - 4 comments

#16190 - [ROCM][Ukernel] Fix ukernel empty check.

Pull Request - State: closed - Opened by raikonenfnu 10 months ago

#16188 - [cuda] Drop cuda1 HAL implementation code

Pull Request - State: open - Opened by antiagainst 10 months ago - 2 comments
Labels: hal/cuda, cleanup 🧹

#16187 - Replace func.func/call with util.func/call.

Issue - State: open - Opened by benvanik 10 months ago
Labels: compiler/dialects, cleanup 🧹

#16186 - Delete unused Bazel deep_copy and template_rule files.

Pull Request - State: closed - Opened by ScottTodd 10 months ago
Labels: infrastructure, cleanup 🧹

#16185 - Check https://github.com/llvm/llvm-project/pull/77874

Pull Request - State: open - Opened by MaheshRavishankar 10 months ago
Labels: benchmarks:cuda, benchmarks:x86_64, benchmarks:comp-stats, benchmarks:android-cpu, benchmarks:android-gpu, benchmarks:vulkan-nvidia

#16184 - Remove `c2-standard-16` from codebase

Pull Request - State: closed - Opened by pzread 10 months ago - 1 comment
Labels: benchmarks:x86_64

#16183 - [GlobalOpt] Add patterns for simplifying 1D pack/unpack ops to reshapes

Pull Request - State: closed - Opened by hanhanW 10 months ago - 1 comment
Labels: benchmarks:x86_64, benchmarks:android-cpu

#16182 - [GlobalOpt] Add patterns for simplifying 1D pack/unpack ops to reshapes

Pull Request - State: closed - Opened by hanhanW 10 months ago
Labels: benchmarks:x86_64, benchmarks:android-cpu

#16181 - reshape-like pack/unpack ops are not folded into reshapes

Issue - State: open - Opened by hanhanW 10 months ago
Labels: good first issue 🌱, codegen

#16180 - Bump LLVM to https://github.com/llvm/llvm-project/pull/78329

Pull Request - State: closed - Opened by MaheshRavishankar 10 months ago - 1 comment
Labels: benchmarks:cuda, benchmarks:x86_64, benchmarks:comp-stats, benchmarks:android-cpu, benchmarks:android-gpu, benchmarks:vulkan-nvidia

#16179 - Start LLVM integrate integrate-llvm-20240122

Pull Request - State: closed - Opened by MaheshRavishankar 10 months ago - 1 comment
Labels: benchmarks:cuda, benchmarks:x86_64, benchmarks:comp-stats, benchmarks:android-cpu, benchmarks:android-gpu, benchmarks:vulkan-nvidia

#16178 - Teach "auto" input detection about ONNX in Torch plugin.

Pull Request - State: open - Opened by ScottTodd 10 months ago
Labels: integrations/pytorch

#16177 - Update IREE architecture diagram targets.

Pull Request - State: closed - Opened by ScottTodd 10 months ago
Labels: documentation

#16175 - [GPU] linalg.conv_2d_nhwc_hwcf result is wrong on cuda backend

Issue - State: open - Opened by kircle888 10 months ago
Labels: bug 🐞

#16173 - Need to bump tf-nightly to drop the revert in #16114 integrate

Issue - State: open - Opened by hanhanW 10 months ago - 10 comments
Labels: integrations/tosa

#16172 - [VectorDistribution] Add pattern to distribute contractions

Pull Request - State: closed - Opened by harsh-nod 10 months ago

#16171 - Add support for hoisting bounded alloc ops

Pull Request - State: closed - Opened by yzhang93 10 months ago

#16171 - Add support for hoisting bounded alloc ops

Pull Request - State: closed - Opened by yzhang93 10 months ago

#16170 - Update IREE architecture diagrams.

Pull Request - State: closed - Opened by ScottTodd 10 months ago - 2 comments
Labels: documentation

#16169 - try-partial-unrolling-threshold-4

Pull Request - State: open - Opened by bjacob 10 months ago - 1 comment
Labels: benchmarks:x86_64

#16169 - try-partial-unrolling-threshold-4

Pull Request - State: open - Opened by bjacob 10 months ago - 1 comment
Labels: benchmarks:x86_64

#16168 - Exported I/O buffer views improperly alias with tied ops.

Issue - State: open - Opened by benvanik 10 months ago
Labels: bug 🐞, compiler/dialects

#16168 - Exported I/O buffer views improperly alias with tied ops.

Issue - State: open - Opened by benvanik 10 months ago
Labels: bug 🐞, compiler/dialects

#16167 - try-partial-unrolling-threshold-16

Pull Request - State: open - Opened by bjacob 10 months ago - 1 comment
Labels: benchmarks:x86_64

#16167 - try-partial-unrolling-threshold-16

Pull Request - State: open - Opened by bjacob 10 months ago - 1 comment
Labels: benchmarks:x86_64

#16166 - [hal] Share event-backed semaphore and deferred queue

Pull Request - State: open - Opened by antiagainst 10 months ago - 4 comments
Labels: hal/cuda, hal/rocm

#16165 - [CPU] "heuristics" -> "undefined" in the definition of `clPProcStrategy`

Pull Request - State: closed - Opened by banach-space 10 months ago - 7 comments

#16163 - [CPU][DataTiling] Extend encodings to linalg.generic ops

Pull Request - State: open - Opened by Max191 10 months ago - 4 comments

#16161 - [CPU][DataTiling] Remove LinalgExt::UserEncoding attribute and ContractionOpType enum

Pull Request - State: closed - Opened by Max191 10 months ago - 1 comment

#16160 - [CPU] Introduce vector unit dims optimization in vector lowering.

Pull Request - State: open - Opened by hanhanW 10 months ago - 11 comments
Labels: benchmarks:x86_64, benchmarks:android-cpu

#16159 - Revert "Disable loop unrolling in LLVM IR optimization passes"

Pull Request - State: open - Opened by bjacob 10 months ago - 9 comments

#16158 - [cuda] Collect tracing events after command buffer completion

Pull Request - State: closed - Opened by antiagainst 10 months ago - 1 comment
Labels: hal/cuda

#16157 - Benchmarking tracy capture unresponsive on A100 bots

Issue - State: open - Opened by antiagainst 10 months ago
Labels: hal/cuda

#16156 - Missing i8 * i4 = i32, i4 * i8 = i32 ukernels

Issue - State: open - Opened by mariecwhite 10 months ago - 3 comments
Labels: bug 🐞, codegen/llvm

#16156 - Missing i8 * i4 = i32, i4 * i8 = i32 ukernels

Issue - State: open - Opened by mariecwhite 10 months ago - 3 comments
Labels: bug 🐞, codegen/llvm

#16155 - Fix web and Colab sample CI builds.

Pull Request - State: closed - Opened by ScottTodd 10 months ago - 2 comments

#16154 - [ROCM][Ukernel] Fix index types.

Pull Request - State: closed - Opened by raikonenfnu 10 months ago

#16152 - [CPU][DataTiling] Remove EncodingUser attribute, extend encodings to `linalg.generic` ops, remove LiftGenericToTransposeBatchMatmulPass

Pull Request - State: open - Opened by Max191 10 months ago - 4 comments
Labels: benchmarks:x86_64, benchmarks:comp-stats, benchmarks:android-cpu

#16152 - [CPU][DataTiling] Remove EncodingUser attribute, extend encodings to `linalg.generic` ops, remove LiftGenericToTransposeBatchMatmulPass

Pull Request - State: open - Opened by Max191 10 months ago - 4 comments
Labels: benchmarks:x86_64, benchmarks:comp-stats, benchmarks:android-cpu

#16151 - Refresh website branding.

Pull Request - State: closed - Opened by ScottTodd 10 months ago - 4 comments
Labels: documentation

#16150 - WIP: Add sample for automatic custom kernel match+replace

Pull Request - State: open - Opened by qedawkins 10 months ago

#16149 - Several compiler flags like `--iree-llvmcpu-loop-unrolling` have no effect

Issue - State: closed - Opened by bjacob 10 months ago - 3 comments
Labels: bug 🐞, codegen/llvm

#16148 - Add e2e tests for packing on i8 types.

Pull Request - State: closed - Opened by hanhanW 10 months ago

#16147 - [CPU][RISC-V] Re-enable MobileNetV2 int8 benchmarks.

Pull Request - State: closed - Opened by hanhanW 10 months ago - 2 comments
Labels: benchmarks:comp-stats

#16147 - [CPU][RISC-V] Reenable MobileNetV2 int8 benchmarks.

Pull Request - State: open - Opened by hanhanW 10 months ago - 1 comment

#16146 - Provide IREE host tools for Mac OS for cross-compilation

Issue - State: open - Opened by RechieKho 10 months ago - 1 comment
Labels: enhancement ➕

#16145 - how can I compile target device with RISCV64 IMAF , not RISCV GC

Issue - State: open - Opened by unicormbu 10 months ago - 2 comments
Labels: enhancement ➕

#16145 - how can I compile target device with RISCV64 IMAF , not RISCV GC

Issue - State: open - Opened by unicormbu 10 months ago - 2 comments
Labels: enhancement ➕

#16144 - [NFC] Move getRootOperation utility from CPUUtils -> Utils

Pull Request - State: open - Opened by Abhishek-Varma 10 months ago - 8 comments

#16144 - [NFC] Move getRootOperation utility from CPUUtils -> Utils

Pull Request - State: open - Opened by Abhishek-Varma 10 months ago - 3 comments

#16143 - [CPU] Improve vector tile sizes for sub-byte matmuls on Aarch64

Pull Request - State: closed - Opened by dcaballe 10 months ago - 2 comments
Labels: benchmarks:x86_64, benchmarks:comp-stats, benchmarks:android-cpu

#16142 - Enforce CUDA >= 12 and fix its CMake search procedure

Pull Request - State: closed - Opened by sogartar 10 months ago
Labels: infrastructure, hal/cuda

#16141 - [stablehlo] Implementation of `cholesky` via `linalg`

Pull Request - State: open - Opened by rsuderman 10 months ago

#16140 - Fixing copypasta in codegen/flow tests that included ABI ops.

Pull Request - State: closed - Opened by benvanik 10 months ago
Labels: codegen, cleanup 🧹

#16139 - Selectively enable some LLVM loop unrolling options

Issue - State: open - Opened by bjacob 10 months ago - 1 comment
Labels: performance ⚡, codegen/llvm

#16138 - ElementsAttr does not provide iteration facilities for mlir::Attribute

Issue - State: open - Opened by monorimet 10 months ago
Labels: bug 🐞

#16138 - ElementsAttr does not provide iteration facilities for mlir::Attribute

Issue - State: open - Opened by monorimet 10 months ago
Labels: bug 🐞

#16137 - Simplify how mmt4d ukernels deal with the K=0 case.

Pull Request - State: closed - Opened by bjacob 10 months ago

#16136 - Bytecode execution fails when it targets both `vulkan` and `metal`.

Issue - State: open - Opened by RechieKho 10 months ago
Labels: bug 🐞

#16135 - Always print x86 in Intel syntax

Pull Request - State: open - Opened by bjacob 10 months ago

#16134 - Removing YAML tracing tooling and the libyaml dep.

Pull Request - State: closed - Opened by benvanik 10 months ago - 2 comments
Labels: bindings/python, runtime/tools, cleanup 🧹

#16133 - [VectorExt] Fix LayoutIterator iteration for step != 0

Pull Request - State: closed - Opened by Groverkss 10 months ago - 2 comments

#16132 - Allow specifying multiple --device= flags in tooling.

Pull Request - State: closed - Opened by benvanik 10 months ago
Labels: runtime/tools

#16131 - Fixing task worker utilization tracing plot.

Pull Request - State: closed - Opened by benvanik 10 months ago
Labels: bug 🐞, hal/cpu

#16131 - Fixing task worker utilization tracing plot.

Pull Request - State: closed - Opened by benvanik 10 months ago
Labels: bug 🐞, hal/cpu

#16130 - Replace iree-e2e-matmul-test reliance on YAML with IR.

Issue - State: closed - Opened by benvanik 10 months ago - 2 comments
Labels: compiler/tools, cleanup 🧹

#16130 - Replace iree-e2e-matmul-test reliance on YAML with IR.

Issue - State: closed - Opened by benvanik 10 months ago - 2 comments
Labels: compiler/tools, cleanup 🧹

#16129 - Preserve reflection attrs on functions when wrapping for the native ABI.

Pull Request - State: closed - Opened by benvanik 10 months ago
Labels: compiler/dialects

#16129 - Preserve reflection attrs on functions when wrapping for the native ABI.

Pull Request - State: closed - Opened by benvanik 10 months ago
Labels: compiler/dialects

#16128 - [Stream] Transient buffer adding extra copies to Llama2 inference

Issue - State: open - Opened by Max191 10 months ago - 45 comments

#16128 - [Stream] Transient buffer adding extra copies to Llama2 inference

Issue - State: open - Opened by Max191 10 months ago - 45 comments

#16127 - [Vector Distribution] Layout Choices

Issue - State: open - Opened by Groverkss 10 months ago - 1 comment
Labels: enhancement ➕, codegen

#16127 - [Vector Distribution] Layout Choices

Issue - State: open - Opened by Groverkss 10 months ago - 1 comment
Labels: enhancement ➕, codegen

#16126 - Ukernels: simplify the architecture-specific bitcode build.

Pull Request - State: closed - Opened by bjacob 10 months ago - 5 comments

#16125 - conv2d-to-img-to-col pattern failure diagnostic improvements

Pull Request - State: open - Opened by newling 10 months ago - 1 comment

#16125 - conv2d-to-img-to-col pattern failure diagnostic improvements

Pull Request - State: closed - Opened by newling 10 months ago - 2 comments

#16123 - [HIP] Add macro for HIP build deps update

Pull Request - State: closed - Opened by nithinsubbiah 10 months ago

#16123 - [HIP] Add macro for HIP build deps update

Pull Request - State: closed - Opened by nithinsubbiah 10 months ago

#16122 - iree-complie error for layer - UpSampling2D

Issue - State: closed - Opened by shreya-ibind 10 months ago - 1 comment
Labels: bug 🐞

#16122 - iree-complie error for layer - UpSampling2D

Issue - State: closed - Opened by shreya-ibind 10 months ago - 1 comment
Labels: bug 🐞