Skip to content

Near-roofline NVFP4 quantization kernel #4516

Closed
santoshmo wants to merge 11 commits into
pytorch:mainfrom
santoshmo:nvfp4-rht-cutedsl
Closed

Near-roofline NVFP4 quantization kernel #4516
santoshmo wants to merge 11 commits into
pytorch:mainfrom
santoshmo:nvfp4-rht-cutedsl

nvfp4: max-bandwidth CuTeDSL quantize kernel + selectable NVFP4Tensor…

c0390a7
Select commit
Loading
Failed to load commit list.
PyTorch Bot / Dr.CI completed Jun 19, 2026 in 0s

Dr.CI classification results

{"FAILED":[{"workflowId":27843140344,"workflowUniqueId":89543087,"id":82406467902,"runnerName":"i-033b67ba8002484ce","authorEmail":"sanmohan@fb.com","name":"Run Regression Tests / test-nightly (CUDA Nightly, linux.g5.12xlarge.nvidia.gpu, --pre torch torchvision --index-url htt... / linux-job","jobName":"test-nightly (CUDA Nightly, linux.g5.12xlarge.nvidia.gpu, --pre torch torchvision --index-url htt... / linux-job","conclusion":"failure","completed_at":"2026-06-19T20:58:14.000000000Z","html_url":"https://github.com/pytorch/ao/actions/runs/27843140344/job/82406467902","logUrl":"https://ossci-raw-job-status.s3.amazonaws.com/log/pytorch/ao/82406467902","head_branch":"nvfp4-rht-cutedsl","pr_number":4516,"head_sha":"c0390a7cca4d82d0d63a0b03070bb2f33fe22d9e","head_sha_timestamp":"2026-06-19T18:46:31Z","failure_captures":["test/test_low_bit_optim.py::TestFSDP2::test_uneven_shard"],"failure_lines":["FAILED test/test_low_bit_optim.py::TestFSDP2::test_uneven_shard - AssertionError: Scalars are not equal!"],"failure_context":[],"time":"2026-06-19T18:49:30.000000000Z"},{"workflowId":27843142245,"workflowUniqueId":127020490,"id":82406473876,"runnerName":"GitHub Actions 1019528636","authorEmail":"sanmohan@fb.com","name":"PR Label Check / Check PR Labels","jobName":"Check PR Labels","conclusion":"failure","completed_at":"2026-06-19T18:49:41.000000000Z","html_url":"https://github.com/pytorch/ao/actions/runs/27843142245/job/82406473876","logUrl":"https://ossci-raw-job-status.s3.amazonaws.com/log/pytorch/ao/82406473876","head_branch":"nvfp4-rht-cutedsl","pr_number":4516,"head_sha":"c0390a7cca4d82d0d63a0b03070bb2f33fe22d9e","head_sha_timestamp":"2026-06-19T18:46:31Z","failure_captures":["Process completed with exit code 1."],"failure_lines":["##[error]Process completed with exit code 1."],"failure_context":[],"time":"2026-06-19T18:49:33.000000000Z"}],"FLAKY":[],"BROKEN_TRUNK":[],"UNSTABLE":[],"UNKNOWN":[],"AWAITING_APPROVAL":[]}