Skip to content

Actions: NVIDIA/TransformerEngine

Actions

Build

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
build.yml
2,500+ workflow runs
2,500+ workflow runs

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
fix: add missing NVTE_BHSD_BHSD_BHSD to layout to_string Build #20030: Pull request #3361 reopened by andrewwhitecdw
Action required andrewwhitecdw:bugfix/fused-attn-add-missing-nvte-bhsd-bhsd-bhsd-to
andrewwhitecdw:bugfix/fused-attn-add-missing-nvte-bhsd-bhsd-bhsd-to
Action required
fix: add missing NVTE_BHSD to qkv format to_string Build #20029: Pull request #3362 reopened by andrewwhitecdw
Action required andrewwhitecdw:bugfix/fused-attn-add-missing-nvte-bhsd-to-qkv-format-to
andrewwhitecdw:bugfix/fused-attn-add-missing-nvte-bhsd-to-qkv-format-to
Action required
fix: simplify redundant softmax version condition Build #20028: Pull request #3363 reopened by andrewwhitecdw
Action required andrewwhitecdw:codequality/fused-attn-simplify-redundant-softmax-version
andrewwhitecdw:codequality/fused-attn-simplify-redundant-softmax-version
Action required
fix: simplify qkv_format checks using a membership tuple Build #20027: Pull request #3365 reopened by andrewwhitecdw
Action required andrewwhitecdw:codequality/run-attention-with-cp-simplify-qkv-format-checks-using-a
andrewwhitecdw:codequality/run-attention-with-cp-simplify-qkv-format-checks-using-a
Action required
fix: Optional segment position annotations in mask helpers Build #20026: Pull request #3369 reopened by andrewwhitecdw
Action required andrewwhitecdw:codequality/test-fused-attn-optional-segment-position-annotations
andrewwhitecdw:codequality/test-fused-attn-optional-segment-position-annotations
Action required
fix: remove redundant self-assignment out_ = out_ Build #20025: Pull request #3367 reopened by andrewwhitecdw
-1s andrewwhitecdw:codequality/run-attention-with-cp-remove-redundant-self-assignment-out-out
andrewwhitecdw:codequality/run-attention-with-cp-remove-redundant-self-assignment-out-out
-1s
fix: make_mask uses deprecated jnp.bool instead of jnp.bool_ Build #20024: Pull request #3371 reopened by andrewwhitecdw
Action required andrewwhitecdw:bugfix/test-fused-attn-make-mask-uses-deprecated-jnp-bool
andrewwhitecdw:bugfix/test-fused-attn-make-mask-uses-deprecated-jnp-bool
Action required
fix: comment typo 'mas' -> 'mask' in two TODOs Build #20023: Pull request #3372 reopened by andrewwhitecdw
Action required andrewwhitecdw:codequality/test-fused-attn-comment-typo-mas-mask-in-two-todos
andrewwhitecdw:codequality/test-fused-attn-comment-typo-mas-mask-in-two-todos
Action required
fix: Unreachable backend check after earlier backend skip Build #20022: Pull request #3368 reopened by andrewwhitecdw
Action required andrewwhitecdw:codequality/test-fused-attn-unreachable-backend-check-after
andrewwhitecdw:codequality/test-fused-attn-unreachable-backend-check-after
Action required
[CI] Publish PyTorch 2.11 CUDA 13 release wheels Build #20021: Pull request #3311 synchronize by bvolpato
Action required bvolpato:bvolpato/publish-gb200-aarch64-wheels
bvolpato:bvolpato/publish-gb200-aarch64-wheels
Action required
surface comm overlap rank errors Build #20020: Pull request #3384 synchronize by sudhakarsingh27
6h 0m 7s sudhakarsingh27:sudhakars/surface-comm-overlap-rank-errors
sudhakarsingh27:sudhakars/surface-comm-overlap-rank-errors
6h 0m 7s
[All] Refactor fused attention APIs with cuDNN-frontend support checks and opaque config/params handles Build #20019: Pull request #2964 synchronize by pre-commit-ci Bot
6h 0m 22s cyanguwa:fe_check_support
cyanguwa:fe_check_support
6h 0m 22s
[All] Refactor fused attention APIs with cuDNN-frontend support checks and opaque config/params handles Build #20018: Pull request #2964 synchronize by cyanguwa
2m 14s cyanguwa:fe_check_support
cyanguwa:fe_check_support
2m 14s
Prototype NVFP4 with FP8 UE5M3 block scales Build #20017: Pull request #3325 synchronize by timmoon10
6h 0m 10s timmoon10:nvfp4-ue5m3-prototype
timmoon10:nvfp4-ue5m3-prototype
6h 0m 10s
[Common] Use wide instructions in SR to reduce issue-bound bottleneck Build #20016: Pull request #3357 synchronize by janekb04
6h 0m 7s janekb04:wide_instruction
janekb04:wide_instruction
6h 0m 7s
[Pytorch] Optimize CPU overhead for NCCL EP eager mode Build #20015: Pull request #3378 synchronize by phu0ngng
6h 0m 9s YangFei1990:ncclep_cpu_overhead
YangFei1990:ncclep_cpu_overhead
6h 0m 9s
surface comm overlap rank errors Build #20014: Pull request #3384 synchronize by pre-commit-ci Bot
6h 0m 23s sudhakarsingh27:sudhakars/surface-comm-overlap-rank-errors
sudhakarsingh27:sudhakars/surface-comm-overlap-rank-errors
6h 0m 23s
surface comm overlap rank errors Build #20013: Pull request #3384 opened by sudhakarsingh27
1m 54s sudhakarsingh27:sudhakars/surface-comm-overlap-rank-errors
sudhakarsingh27:sudhakars/surface-comm-overlap-rank-errors
1m 54s
Add a backwards linear function to be used with the fused mla q up-proj Build #20012: Pull request #3330 synchronize by pre-commit-ci Bot
6h 0m 20s chaseblock:lin_bwd
chaseblock:lin_bwd
6h 0m 20s
Add a backwards linear function to be used with the fused mla q up-proj Build #20011: Pull request #3330 synchronize by chaseblock
1m 19s chaseblock:lin_bwd
chaseblock:lin_bwd
1m 19s
Produce BOLT-compatible libraries Build #20010: Pull request #3383 synchronize by fheinecke
6h 0m 22s fheinecke:fred/bolt-compatible-libs-1
fheinecke:fred/bolt-compatible-libs-1
6h 0m 22s
Produce BOLT-compatible libraries Build #20009: Pull request #3383 opened by fheinecke
39s fheinecke:fred/bolt-compatible-libs-1
fheinecke:fred/bolt-compatible-libs-1
39s
[Common] Experimental CuTeDSL MXFP8 backends in C++ via TVM-FFI Build #20008: Pull request #3137 synchronize by kainzhong
12m 7s kainzhong:cutedsl_mxfp8_common
kainzhong:cutedsl_mxfp8_common
12m 7s
[All] Refactor fused attention APIs with cuDNN-frontend support checks and opaque config/params handles Build #20007: Pull request #2964 synchronize by cyanguwa
6h 0m 7s cyanguwa:fe_check_support
cyanguwa:fe_check_support
6h 0m 7s
[JAX] Move ffi type definition before ffi target definitions Build #20006: Pull request #3382 synchronize by jberchtold-nvidia
6h 0m 25s jberchtold-nvidia:jberchtold/fix-ffi-ep-type-registration-order
jberchtold-nvidia:jberchtold/fix-ffi-ep-type-registration-order
6h 0m 25s