iree-org/iree

Issues

• [tosa] failed to legalize operation 'arith.extsi'

  #19402 opened 4 months ago by lundong
  5

• LLama-3.1 Toy example gives different numbers on CPU and on HIP target

  #19347 opened 4 months ago by MaheshRavishankar
  37

• pkg_ci failure: test_sdxl_rocm_benchmark

  #19442 opened 4 months ago by amd-chrissosa
  7

• [CPU] Shortfin LLM Accuracy Degradation Followup

  #19411 opened 4 months ago by IanWood1
  10

• Tracking issue for dropping the local revert of https://github.com/llvm/llvm-project/pull/116470

  #19448 opened 4 months ago by MaheshRavishankar
  2

• CLIP text model bf16 fails to compile for HIP with `unsupported dynamic alloca` in pass `AMDGPU DAG->DAG Pattern Instruction Selection`

  #19385 opened 4 months ago by sogartar
  2

• Llama 3.1 8b f16 sharded TP8 compiles but fails to run

  #19428 opened 4 months ago by aviator19941
  4

• [regression][gpu] Regression due to cb59389

  #19436 opened 4 months ago by pdhirajkumarprasad
  3

• Switching data-tiling to late materialization path (being compatible with multi-device)

  #19378 opened 4 months ago by hanhanW
  2

• [numeric] Numeric error for Sigmoid with Conv operator

  #19415 opened 4 months ago by pdhirajkumarprasad
  2

• [numeric] Numeric error for Conv operator with quantize/dequantize

  #19416 opened 4 months ago by pdhirajkumarprasad
  2

• Samples workflow broken by LLVM webassembly backend changes

  #19425 opened 4 months ago by ScottTodd
  0

• error: One or more operations with large vector sizes (16384 bytes) were found

  #19386 opened 4 months ago by pdhirajkumarprasad
  1

• [CPU] Llama 3b bs1 compilation error

  #19410 opened 4 months ago by IanWood1
  1

• (compiler) (bf16) Flux.1-dev transformer dispatch fails due to memref element type mismatch.

  #19397 opened 4 months ago by monorimet
  5

• [GPU] Softmax dispatch has two reductions and one of them which doesnt have a lowering config is not getting tiled

  #19406 opened 4 months ago by nirvedhmeshram
  6

• [Codegen] vulkan-spirv backend fails to compile collapsed softmax

  #19409 opened 4 months ago by IanWood1
  0

• Llama 3.1 8B fp16 TP8 sharded fails to compile for CPU and GPU

  #19263 opened 4 months ago by aviator19941
  8

• [Codegen] Llama 3.1 8B fp16 TP8 compile time regression

  #19377 opened 4 months ago by IanWood1
  0

• PJRT plugin fails to execute in JAX while multiple GPUs are provided

  #19366 opened 4 months ago by PragmaTwice
  0

• Correctness issue related to EmplaceAllocations

  #19355 opened 4 months ago by qedawkins
  9

• error: 'func.func' op exceeded stack allocation limit of 32768 bytes for function. Got 229376 bytes

  #19333 opened 4 months ago by pdhirajkumarprasad
  3

• [numeric] Numeric failures with Conv operator

  #19339 opened 4 months ago by pdhirajkumarprasad
  3

• [LLVMGPUVectorDistribute] Make layout conversions more read friendly from shared memory

  #19384 opened 4 months ago by manupak
  0

• 'vector.transfer_write' op inferred mask type ('vector<1x1x8xi1>') and mask operand type ('vector<1x8x1xi1>') don't match

  #19383 opened 4 months ago by pdhirajkumarprasad
  0

• General support of i1 mask in attention

  #19380 opened 4 months ago by lialan
  1

• [Feature Request] Frontend to emit i1 packed tensor for attention masks

  #19382 opened 4 months ago by lialan
  0

• Regression in Llama3.1 Causing `Abort` Failure

  #19360 opened 4 months ago by stbaione
  2

• failed to legalize operation 'arith.sitofp'

  #19375 opened 4 months ago by jinchen62
  0

• Compiling ConfigTrackingCanonicalizer fails on aarch64

  #19264 opened 4 months ago by marbre
  4

• CI lint failure: bazel_to_cmake.py truncating output CMakeLists.txt at 16,384 bytes.

  #19365 opened 4 months ago by bjacob
  2

• TileAndFuse configuration gives numeric error

  #19343 opened 4 months ago by nirvedhmeshram
  5

• Gather fusion + warp reduction fails

  #19353 opened 4 months ago by manupak
  5

• [regression][cpu] error: Transpose bubbling patterns failed

  #19320 opened 4 months ago by pdhirajkumarprasad
  3

• Elementwise Op Fusion fails due to `complex` tensor operands

  #19288 opened 4 months ago by rsuderman
  2

• (compiler) erroneous collapse_shape in flux text encoder

  #19284 opened 4 months ago by monorimet
  3

• [compiler] FLUX.1 transformer compilation for gfx942 hangs with pad-to-intrinsics

  #19249 opened 4 months ago by monorimet
  3

• Strange behavior trying to upgrade iree-base-compiler Python package

  #19305 opened 4 months ago by ScottTodd
  3

• [compiler] SDXL EulerDiscreteScheduler Compilation Failure

  #19262 opened 4 months ago by saienduri
  11

• Can not execute from shark_turbine.aot import *

  #19301 opened 4 months ago by FlintWangacc
  1

• Can not execute from shark_turbine.aot import *

  #19303 opened 4 months ago by FlintWangacc
  1

• failed to legalize operation 'arith.extf'

  #19299 opened 4 months ago by jinchen62
  0

• Llama 3.1 8B f16 decomposed fails to compile after commit 2602a2a730b5c1439b79360644c9884377110080

  #19290 opened 4 months ago by aviator19941
  2

• (compiler) (gfx942) Layout shape mismatch error resulting from vector distribution in Flux.1 VAE

  #19286 opened 4 months ago by monorimet
  0

• Fix Global Opt for 1x1 Filter Convs

  #19270 opened 4 months ago by IanWood1
  2

• [Python Bindings] Conflict between Linalg and IREEGPU iterator types in python bindings

  #19269 opened 4 months ago by Max191
  0

• Peel away conditional check from TileAndFuse padding kernel inner K loop

  #19276 opened 4 months ago by jerryyin
  0

• TileAndFuse padding enablement feature enhancements

  #19275 opened 4 months ago by jerryyin
  0

• Add versioning to documentation website(s)

  #19251 opened 4 months ago by ScottTodd
  0

• configure_ci.py interprets `Notes:` as "trailer lines"

  #19240 opened 4 months ago by bjacob
  3