Search results

BWDW JIT 256 error reproducible through gtests, but not through benchdnn

component:testsCodeowner: @oneapi-src/onednn-archCodeowner: @oneapi-src/onednn-arch

help wanted

platform:cpu-aarch64Codeowner: @oneapi-src/onednn-cpu-aarch64Codeowner: @oneapi-src/onednn-cpu-aarch64

sightingSuspicious library behavior. Should be promoted to a bug when confirmedSuspicious library behavior. Should be promoted to a bug when confirmed

Status: Open.

#4124 In uxlfoundation/oneDNN;

· gassan-arm opened on Oct 13, 2025

Large kernels can generate xbyak_aarch64 exceptions

help wanted

platform:cpu-aarch64Codeowner: @oneapi-src/onednn-cpu-aarch64Codeowner: @oneapi-src/onednn-cpu-aarch64

sightingSuspicious library behavior. Should be promoted to a bug when confirmedSuspicious library behavior. Should be promoted to a bug when confirmed

Status: Open.

#4089 In uxlfoundation/oneDNN;

· Sqvid opened on Oct 8, 2025

How to use dnnl_rnn_flags_diff_weights_overwrite in RNN LSTM primitives creation?

question

Status: Open.

#3995 In uxlfoundation/oneDNN;

· lacak-sk opened on Sep 24, 2025

Temporary fix nightly crash for large l_pad failing test

bugA confirmed library bugA confirmed library bug

help wanted

platform:cpu-aarch64Codeowner: @oneapi-src/onednn-cpu-aarch64Codeowner: @oneapi-src/onednn-cpu-aarch64

Status: Open.

#3981 In uxlfoundation/oneDNN;

· kasturedeeksha opened on Sep 22, 2025

[RV64] test_benchdnn_modeC_matmul_ci_cpu fails with dropout attribute Description

bugA confirmed library bugA confirmed library bug

help wanted

platform:cpu-rv64RISC-VRISC-V

Status: Open.

#3934 In uxlfoundation/oneDNN;

· xiazhuozhao opened on Sep 12, 2025

AArch64: graph fusion and graph ci cases failed in AArch64 platform

bugA confirmed library bugA confirmed library bug

platform:cpu-aarch64Codeowner: @oneapi-src/onednn-cpu-aarch64Codeowner: @oneapi-src/onednn-cpu-aarch64

Status: Open.

Bug

#3825 In uxlfoundation/oneDNN;

· xiang1guo opened on Aug 26, 2025

Please privide more 4bit matmul docs and example?

platform:cpu-x64Intel64/AMD64 processors. Codeowner: @oneapi-src/onednn-cpu-x64Intel64/AMD64 processors. Codeowner: @oneapi-src/onednn-cpu-x64

question

Status: Open.

#3641 In uxlfoundation/oneDNN;

· alanzhai219 opened on Jul 21, 2025

Example of dnnl::vanilla_rnn_backward

question

Status: Open.

#3257 In uxlfoundation/oneDNN;

· dtecta opened on May 13, 2025

Test test_bfloat16_plus_float fails on s390x

help wanted

platform:cpu-s390xIBM Z systems based on s390x architectureIBM Z systems based on s390x architecture

sightingSuspicious library behavior. Should be promoted to a bug when confirmedSuspicious library behavior. Should be promoted to a bug when confirmed

Status: Open.

#2228 In uxlfoundation/oneDNN;

· skriesch opened on Nov 26, 2024

[ARM] Support fp16 data type in JIT Reorder kernel

enhancementA feature or an optimization requestA feature or an optimization request

help wanted

platform:cpu-aarch64Codeowner: @oneapi-src/onednn-cpu-aarch64Codeowner: @oneapi-src/onednn-cpu-aarch64

Status: Open.

#2185 In uxlfoundation/oneDNN;

· dmitry-gorokhov opened on Oct 28, 2024

Extend support for JIT Backward Convolution Operators with ARM SVE 128bit

enhancementA feature or an optimization requestA feature or an optimization request

help wanted

platform:cpu-aarch64Codeowner: @oneapi-src/onednn-cpu-aarch64Codeowner: @oneapi-src/onednn-cpu-aarch64

Status: Open.

#2165 In uxlfoundation/oneDNN;

· snadampal opened on Oct 14, 2024

[ARM] Support 8bit/4bit weights decompression for Matmul primitive

enhancementA feature or an optimization requestA feature or an optimization request

help wanted

platform:cpu-aarch64Codeowner: @oneapi-src/onednn-cpu-aarch64Codeowner: @oneapi-src/onednn-cpu-aarch64

Status: Open.

#2081 In uxlfoundation/oneDNN;

· dmitry-gorokhov opened on Sep 4, 2024