Repository Issue Activity (beta)

#24064 `iree_linalg_ext.online_attention` codegen fails on gfx1151 (Strix Halo)

#24378 [CPU] Improve test coverage for onnx_ops test suite

#24914 [CPU] Eliminate cross-dispatch pack/unpack overhead when tensor.pad intervenes between producer and consumer

#24932 Tracking: [RVV] inner_tiled performance vs mmt4d ukernels

#24937 [VMVX] Support vector operands in Math → VM conversions

#24936 Non-linear compile-time blowup around VectorReductionToGPUPass

#24928 [Stream] Cubic scaling compile time in ScheduleConcurrencyPass for dispatches sharing a read-only resource

#24930 Multi-device Zero-Copy Path: Missing lowering for stream.cmd.flush/invalidate and zero-copy path

#24761 CPU multi threading doesn't work on Intel's hybrid core architecture

#24927 Silent wrong results for an integer-only StableHLO program from jax.random (llvm-cpu and vmvx agree with each other, disagree with XLA)

#24929 Runtime abort (OUT_OF_RANGE, length=2^39+N) invoking un-batched JAX Threefry on a runtime-input key; two or more lanes is fine

#24920 [Stream] SplitParameterEncoding in consolidate mode doesn't work

#24764 `publish_website` pushes to the main GH pages even from release candidate workflows

#24876 LLVMCPU with data tiling triggers "one or more operations with large vector sizes" error on llama-3.2-3B MLP

#24910 [IREE][HAL] buffer-view shape product overflows before allocation

#22681 [Codegen] Convert tensort.insert_slice to tensor.pad ops

#24886 [LLVMCPU] Compile time grows exponentially in the length of the def chain feeding a data-dependent tensor.extract`

#22235 [Codegen][GPU] Refactor and consolidate padding passes and configurations

#24830 Remove `--iree-hal-target-backends`

#24888 [LLVMCPU] Data-tiling pack/unpack is vectorized on x86 but not on AArch64

#24766 [LLVMGPU] Track paged KV attention lowering for NVIDIA GPUs

#24885 [Stream] stablehlo.while: loop-carried resource is deallocated on the branch into the loop (transient buffer has not been committed

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

#24881 [StableHLO] GatherToTorchIndexSelect drops the index clamp, so out-of-bounds gathers read arbitrary memory

#24880 Consider migration from MkDocs to Zensical