-
Notifications
You must be signed in to change notification settings - Fork 251
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
- Status: Open.#571 In NVIDIA/cudnn-frontend;
[Bug][CI] frost_tests:sm120: crashed in FP8 ragged SDPA tests on SM120
cat-bugReports of incorrect behavior, crashes, regressions, or unexpected results.Reports of incorrect behavior, crashes, regressions, or unexpected results.orig-qaReported or owned by quality assurance, validation, or testing.Reported or owned by quality assurance, validation, or testing.Status: Open.#563 In NVIDIA/cudnn-frontend;[CI] Multiple oss:rel jobs fail with AttributeError: module 'cudnn.gemm.cutedsl.grouped.wgrad.api' has no attribute 'torch'
cat-bugReports of incorrect behavior, crashes, regressions, or unexpected results.Reports of incorrect behavior, crashes, regressions, or unexpected results.cat-ciCI failures, test flakiness, workflow breakage, or automation issues.CI failures, test flakiness, workflow breakage, or automation issues.orig-qaReported or owned by quality assurance, validation, or testing.Reported or owned by quality assurance, validation, or testing.Status: Open.#560 In NVIDIA/cudnn-frontend;- Status: Open.#552 In NVIDIA/cudnn-frontend;
[CI] deploy:test_bdist_wheels_ga_cudnn_3.12: [Hopper] test_gdn_bprop.py OOM under xdist
cat-bugReports of incorrect behavior, crashes, regressions, or unexpected results.Reports of incorrect behavior, crashes, regressions, or unexpected results.cat-ciCI failures, test flakiness, workflow breakage, or automation issues.CI failures, test flakiness, workflow breakage, or automation issues.orig-qaReported or owned by quality assurance, validation, or testing.Reported or owned by quality assurance, validation, or testing.Status: Open.#551 In NVIDIA/cudnn-frontend;- Status: Open.#550 In NVIDIA/cudnn-frontend;
frost(sdpa): migrate the SM100 dense compact-BSHD normalization to serve-or-decline (AGENTS Hard Rule 2 cleanup)
orig-nv-engReported or requested by NVIDIA engineering.Reported or requested by NVIDIA engineering.Status: Open.#541 In NVIDIA/cudnn-frontend;test(sdpa): fuzz head-axis stride gaps in the ragged sweeps
mod-infraInfrastructure, CI/CD, build systems, packaging, releases, or repo maintenance.Infrastructure, CI/CD, build systems, packaging, releases, or repo maintenance.orig-nv-engReported or requested by NVIDIA engineering.Reported or requested by NVIDIA engineering.Status: Open.#540 In NVIDIA/cudnn-frontend;frost(sdpa): THD backward engines must adopt the declared-stride contract from day one
orig-nv-engReported or requested by NVIDIA engineering.Reported or requested by NVIDIA engineering.Status: Open.#539 In NVIDIA/cudnn-frontend;frost(sdpa): cu_seq_len (cu_ragged) form assumes packed strides
cat-bugReports of incorrect behavior, crashes, regressions, or unexpected results.Reports of incorrect behavior, crashes, regressions, or unexpected results.orig-nv-engReported or requested by NVIDIA engineering.Reported or requested by NVIDIA engineering.Status: Open.#538 In NVIDIA/cudnn-frontend;frost(sdpa): FP8/MXFP8 THD declared-stride support
orig-nv-engReported or requested by NVIDIA engineering.Reported or requested by NVIDIA engineering.Status: Open.#537 In NVIDIA/cudnn-frontend;[Bug]: [CI][OSS]oss:rel:cutlass_dsl_4.6: [Blackwell, shard1] SDPA backward dq NaN mismatch in oss:rel shard1
cat-bugReports of incorrect behavior, crashes, regressions, or unexpected results.Reports of incorrect behavior, crashes, regressions, or unexpected results.orig-qaReported or owned by quality assurance, validation, or testing.Reported or owned by quality assurance, validation, or testing.Status: Open.#536 In NVIDIA/cudnn-frontend;