-
Notifications
You must be signed in to change notification settings - Fork 743
Pull requests: pytorch/FBGEMM
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Add unit tests for warp primitives, bitonic sort, and ROCm warpSize guards (#5715)
ciflow/rocm
cla signed
fb-exported
meta-exported
module: rocm
#5715
opened Apr 29, 2026 by
q10
Contributor
Loading…
Add spin-loop termination to for AMD GPU hang on MP-ZCH
cla signed
fb-exported
meta-exported
#5714
opened Apr 29, 2026 by
Ali-Tehrani
Contributor
Loading…
Fix pyspark_deps.par crash (coredump f160mb8e201359l3) (#2646)
cla signed
fb-exported
meta-exported
#5713
opened Apr 29, 2026 by
excelle08
Contributor
Loading…
TBE backward hip_mixed_d warp kernel for ROCm (#5074)
ciflow/rocm
cla signed
fb-exported
meta-exported
module: rocm
#5712
opened Apr 29, 2026 by
spcyppt
Contributor
Loading…
TBE backward CTA kernel optimization for ROCm (#5074)
ciflow/rocm
cla signed
fb-exported
meta-exported
module: rocm
#5711
opened Apr 29, 2026 by
spcyppt
Contributor
Loading…
Add OSS benchmark scripts for TBE training benchmarks
cla signed
fb-exported
meta-exported
#5710
opened Apr 29, 2026 by
spcyppt
Contributor
Loading…
Fix AMD GPU hang of MP-ZCH (#5708)
cla signed
fb-exported
meta-exported
#5708
opened Apr 28, 2026 by
Ali-Tehrani
Contributor
Loading…
Fix backward cache flush on col_tile change in group_index_select
cla signed
fb-exported
meta-exported
module: rocm
#5707
opened Apr 28, 2026 by
q10
Contributor
Loading…
Fix sign-extension bug in fbgemm MX4 Python reference dequantize (#5706)
cla signed
fb-exported
meta-exported
#5706
opened Apr 28, 2026 by
purvisa-at-meta
Loading…
Fix sign-extension bug in fbgemm MX4 Triton dequantize kernel (#5705)
cla signed
fb-exported
meta-exported
#5705
opened Apr 28, 2026 by
purvisa-at-meta
Loading…
Add unit tests for FusedNBitRowwise fp32 intermediate precision and fp16→int4→bf16 roundtrip
cla signed
fb-exported
meta-exported
#5703
opened Apr 28, 2026 by
zhaozhul
Contributor
Loading…
Add SVE-FP16 version of EmbeddingSpMDM8Bit
cla signed
#5702
opened Apr 27, 2026 by
ShuyangLiu
Loading…
Add SVE-FP16 version of EmbeddingSpMDMNbit
cla signed
#5701
opened Apr 27, 2026 by
ShuyangLiu
Loading…
Remove GPU sync stalls in _prefetch zero-row invalidation
cla signed
#5699
opened Apr 27, 2026 by
EddyLXJ
Contributor
Loading…
[ROCm] support warpSize 32 and 64 in the same build
ciflow/rocm
cla signed
module: rocm
#5696
opened Apr 25, 2026 by
jeffdaily
Loading…
Remove unnecessary if __name__ == "__main__": unittest.main() boilerplate in deeplearning/fbgemm/fbgemm_gpu/test (#5689)
cla signed
fb-exported
meta-exported
#5689
opened Apr 24, 2026 by
meta-codesync
Bot
Loading…
Add diagnostic output to debug OSS CI torch import failure
cla signed
fb-exported
meta-exported
#5686
opened Apr 23, 2026 by
gchalump
Contributor
Loading…
Investigate OSS CI nightly failure: revert Python 3.10+ typing changes
ci-no-td
cla signed
fb-exported
meta-exported
#5685
opened Apr 23, 2026 by
gchalump
Contributor
Loading…
Fix OSS CI nightly failures: setuptools downgrade + cu128 deprecation
cla signed
fb-exported
meta-exported
#5684
opened Apr 23, 2026 by
gchalump
Contributor
Loading…
Refactor bounds_check_indices offset checks to condition-first (Phase 1) (#5682)
cla signed
fb-exported
meta-exported
#5682
opened Apr 23, 2026 by
gchalump
Contributor
Loading…
fbcode/deeplearning/fbgemm/fbgemm_gpu/test/tbe/utils/split_embeddings_utils_test.py
cla signed
fb-exported
meta-exported
#5680
opened Apr 23, 2026 by
meta-codesync
Bot
Loading…
Fix OOM (exit code 137) in CI builds for CUDA 13.2+ (#5679)
cla signed
fb-exported
meta-exported
#5679
opened Apr 23, 2026 by
gchalump
Contributor
Loading…
fbcode/deeplearning/fbgemm/fbgemm_gpu/test/tbe/dram_kv/dram_kv_test.py (#2620)
cla signed
fb-exported
meta-exported
#5678
opened Apr 23, 2026 by
meta-codesync
Bot
Loading…
Add FP8 rowwise padding to quantized AllToAll pooled embeddings (#5673)
cla signed
fb-exported
meta-exported
#5673
opened Apr 22, 2026 by
RohanVardhan
Loading…
fbcode/deeplearning/fbgemm/fbgemm_gpu/test/tbe/training/merge_vbe_test.py
cla signed
fb-exported
meta-exported
#5672
opened Apr 22, 2026 by
meta-codesync
Bot
Loading…
Previous Next
ProTip!
no:milestone will show everything without a milestone.