iree-org/iree

Issues

• SDXL punet_quant.mlir fails to compile at ConvertConvToChannelsLastPass

  #17643 opened in 11 hours by aviator19941
  8

• (Windows/Python) -- Latest windows wheels cannot import iree.compiler.ir.Context

  #17700 opened 2 days ago by monorimet
  3

• Compiler crashed when compiling a simple module with single conv layer

  #17701 opened 2 days ago by Marrrb1e
  1

• Can't compile iree within conda environment: "error: expected parameter declarator"

  #17660 opened 8 days ago by chudur-budur
  4

• In `SetEncoding`, only set one of `matmul_narrow_{M,N}`

  #17641 opened 2 days ago by bjacob
  0

• CI runner for test_nvidia_a100 is running out of disk space

  #17677 opened 2 days ago by ScottTodd
  4

• Fix consistency between MaterializeEncoding tile sizes and max padding size

  #17615 opened 13 days ago by bjacob
  1

• Undo spaghettification of `#hal.device.*` attrs and analysis.

  #17694 opened 2 days ago by benvanik
  0

• Add `stream.async.move` as an equivalent to `stream.async.transfer` but with move semantics.

  #17693 opened 2 days ago by benvanik
  0

• Compiler assert (crash) during fusion of iree_linalg_ext.scatter

  #17691 opened 3 days ago by ScottTodd
  3

• iree-compiler python package leaves directory artifact after pip uninstall

  #17688 opened 3 days ago by dan-garvey
  0

• Unexpected failure to compile i32 matmul with vanilla flags for llvm-cpu

  #17683 opened 4 days ago by newling
  0

• Number of dims and results of reindexed AffineMap doesn't match on Vectorization

  #17591 opened 14 days ago by jinchen62
  11

• [codegen] softmax nans

  #17670 opened 7 days ago by dan-garvey
  4

• Crash / ASan issues using VmModule.wrap_buffer from Python

  #17635 opened 10 days ago by ScottTodd
  46

• Handling reshape propagation for attention ops.

  #17673 opened 7 days ago by MaheshRavishankar
  1

• Matmul dispatch failing to compile due to shared memory allocation

  #17669 opened 7 days ago by IanNod
  0

• The parallel dimensions are not collapsed in unpack + elementwsie fusion

  #17594 opened 14 days ago by hanhanW
  6

• Make tensors flat at boundaries / make row-major indexing explicit during mid-codegen

  #17656 opened 8 days ago by krzysz00
  0

• Upgrade Github runner to v2.317.0 before 2024-06-28

  #17565 opened 8 days ago by yuennancy
  4

• Runtime stuck on execution of inference of YOLOv5 compiled from ONNX

  #17590 opened 15 days ago by maximiliankir
  4

• Wrong `flow.dispatch.tensor.store` op creation; leading to shape mismatch.

  #17526 opened 22 days ago by pashu123
  2

• [Codegen] TileAndDistributeToWorkgroups for operations with multiple results and related producers

  #17632 opened 10 days ago by qedawkins
  6

• Dequantization + Extract Slice Problems

  #17642 opened 9 days ago by IanWood1
  0

• (gfx1103/Windows) Numerics issues on HIP driver for SDXL Unet

  #17579 opened 16 days ago by monorimet
  6

• (znver4/cpu) numerics issues/bad results from different compile flags on SDXL VAE

  #17596 opened 14 days ago by monorimet
  3

• Cannot run posenet example in tflite

  #17622 opened 12 days ago by Sylvexia
  3

• Compiler failure: cannot get concrete layout for contraction in vector distribution

  #17585 opened 10 days ago by IanNod
  0

• (Codegen) --iree-opt-aggressively-propagate-transposes does not work with fx-decomposed attention.

  #17624 opened 11 days ago by monorimet
  1

• Heap buffer overflow in `tensor.pack` with `bf16` element type and transposition.

  #17621 opened 13 days ago by bjacob
  1

• Dependency graph of data-tiling issues

  #17608 opened 13 days ago by bjacob
  0

• MaterializeEncoding for conv on CPU with Winograd

  #17607 opened 14 days ago by bjacob
  0

• Implement MaterializeEncoding for conv ops on GPU with IGEMM

  #17606 opened 14 days ago by bjacob
  0

• Implement SetEncoding for conv ops.

  #17605 opened 14 days ago by bjacob
  0

• Let `EncodingAttr` track the op type

  #17604 opened 14 days ago by bjacob
  0

• Drop `EncodingRole` enum, replace by operand index

  #17602 opened 14 days ago by bjacob
  0

• Data race between worker threads sharing a buffer

  #17525 opened 14 days ago by lialan
  43

• Missing patterns to canonicalize the vectorized result of tensor.unpack

  #17593 opened 14 days ago by hanhanW
  2

• Torch-mlir input conversion pipeline incompatible with `--iree-execution-model=inline-dynamic`

  #17584 opened 15 days ago by zero9178
  1

• Sharding into multiple IREE logical devices or queues with peered memory

  #17550 opened 18 days ago by sogartar
  5

• Bufferization issue with redundant allocation and copy

  #17571 opened 16 days ago by yzhang93
  0

• Compiler crashes for some models when folding TensorSliceOp

  #17566 opened 17 days ago by maxbartel
  1

• `malloc(): corrupted top size` in `iree-compile` and `iree-opt` when building with existing LLVM

  #17561 opened 17 days ago by makslevental
  5

• The rocm links in the doc is outdated

  #17560 opened 17 days ago by hanhanW
  1

• Data tiling encodings: simplify `round_dims_to` and implement narrower padding for narrow dimensions

  #17553 opened 17 days ago by bjacob
  0

• [Winograd] Semantics and fusion changes for winograd ops

  #17548 opened 18 days ago by Max191
  0

• a handle passed as operand #0 and consumed by this operation points to a payload entity more than once

  #17547 opened 18 days ago by jn-shen
  0

• pass demote-i64-to-i32 error

  #17546 opened 18 days ago by bailuan
  0

• Dialect Resources dropped when compiling to flow

  #17544 opened 18 days ago by harsh-amd
  0

• LinalgExt ops should not have Variadic ins/outs

  #17511 opened 23 days ago by Groverkss
  0