-
Notifications
You must be signed in to change notification settings - Fork 775
Pull requests: pytorch/FBGEMM
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[fbgemm_gpu] Add opt-in device sync before permute_2D / jagged_to_pad…
cla signed
#6195
opened Aug 21, 2026 by
RupengWang
Loading…
Use distinct float16 and bfloat16 CPU template types
cla signed
#6194
opened Aug 21, 2026 by
cyyever
Contributor
Loading…
Honor disable_random_init in DRAM inference embedding
cla signed
meta-exported
#6193
opened Aug 21, 2026 by
lizhe-ji
Loading…
Enable the prefetch pipeline for embedding cache enrichment
cla signed
meta-exported
#6192
opened Aug 21, 2026 by
lizhe-ji
Loading…
Add a CUDA kernel for masked_select_jagged_1d
cla signed
meta-exported
#6190
opened Aug 20, 2026 by
MengXiangge
Loading…
Observe shutdown on teardown instead of running loops to completion
cla signed
meta-exported
#6189
opened Aug 20, 2026 by
FriedCosey
Loading…
Drop redundant mask pre-scan in masked_select_jagged_1d via O(1) upper-bound alloc
cla signed
#6188
opened Aug 20, 2026 by
DenisYaroshevskiy
Loading…
[ROCm][CI] Restore ROCm GPU CI: switch to the MI350 runner label
ciflow/rocm
cla signed
module: rocm
#6186
opened Aug 19, 2026 by
aryaman-gupta
Contributor
Loading…
Fix B001: Replace bare except with except Exception
cla signed
meta-exported
#6184
opened Aug 19, 2026 by
meta-codesync
Bot
Loading…
Fix B001: Replace bare except with except Exception
cla signed
meta-exported
#6181
opened Aug 19, 2026 by
meta-codesync
Bot
Loading…
Drop rocm 7.14 from the GenAI x86 wheels matrix until supported
ciflow/rocm
cla signed
meta-exported
module: rocm
#6180
opened Aug 19, 2026 by
meyering
Contributor
Loading…
Make x86-only conv kernels fail loudly on arm instead of faulting or returning zeros
cla signed
meta-exported
#6177
opened Aug 18, 2026 by
meyering
Contributor
Loading…
Ignore rocm.docs.amd.com in Sphinx linkcheck
ciflow/rocm
cla signed
meta-exported
module: rocm
#6176
opened Aug 18, 2026 by
meyering
Contributor
Loading…
Remove unused type error suppressions
cla signed
meta-exported
#6171
opened Aug 18, 2026 by
maggiemoss
Loading…
Port u8s8 quantized GEMM to arm with a portable compute kernel (#6170)
cla signed
meta-exported
#6170
opened Aug 18, 2026 by
meyering
Contributor
Loading…
Compute warning flags before library creation
cla signed
meta-exported
#6166
opened Aug 17, 2026 by
q10
Contributor
Loading…
Reland: runtime arch query for NFP8 dtype selection
ci-no-td
cla signed
meta-exported
#6165
opened Aug 17, 2026 by
q10
Contributor
Loading…
Apply the last k scale block in _kernel_matmul_fp8_block_fastacc
cla signed
#6160
opened Aug 14, 2026 by
truong-v
Loading…
Mask the column-scale load in _kernel_scale_fp8_row
cla signed
#6158
opened Aug 14, 2026 by
truong-v
Loading…
Add -Wmissing-braces (#6154)
cla signed
meta-exported
#6154
opened Aug 13, 2026 by
q10
Contributor
Loading…
Make four -Wall-implied warnings explicit
cla signed
meta-exported
#6145
opened Aug 12, 2026 by
q10
Contributor
Loading…
Keep grid-overflow tests out of opcheck generation
cla signed
meta-exported
#6137
opened Aug 10, 2026 by
q10
Contributor
Loading…
Reduce TBE backward arg count — pack weights + aux tensors into TensorLists (#6132)
cla signed
meta-exported
#6132
opened Aug 10, 2026 by
gchalump
Contributor
Loading…
Bounds-assert permute index in permute_1D/2D_lengths_kernel (S683387)
cla signed
meta-exported
#6129
opened Aug 10, 2026 by
q10
Contributor
Loading…
[ROCm] support warpSize 32 and 64 in a single TBE codegen build
ciflow/rocm
cla signed
module: rocm
#6123
opened Aug 7, 2026 by
jeffdaily
Contributor
Loading…
Previous Next
ProTip!
Updated in the last three days: updated:>2026-08-18.