-
Notifications
You must be signed in to change notification settings - Fork 294
Pull requests: NVIDIA/cudnn-frontend
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Account for launch geometry in SM100 SDPA heuristics
area:frost
area:global_attention
cat-perf-bug
Performance regressions or cases where behavior is correct but too slow.
orig-nv-eng
Reported or requested by NVIDIA engineering.
#1158
opened Sep 20, 2026 by
YangXu1990uiuc
Collaborator
Loading…
4 tasks done
frost(sdpa_fwd_sm80): serve native packed THD graphs
#1155
opened Sep 20, 2026 by
Levius-Fubuki
Loading…
3 of 4 tasks
Rank FROST SDPA plans by workload
area:frost
area:global_attention
cat-bugfix
cat-feature
Requests for new functionality, APIs, examples, or behavior improvements.
orig-nv-eng
Reported or requested by NVIDIA engineering.
Reduce prepared SDPA execution overhead
area:frost
area:global_attention
cat-perf-bug
Performance regressions or cases where behavior is correct but too slow.
orig-nv-eng
Reported or requested by NVIDIA engineering.
update-heuristics
area:frost
cat-enhancements
cat-feature
Requests for new functionality, APIs, examples, or behavior improvements.
mod-frost
[test][sdpa] Cover explicit-stream and device-context boundaries
#1149
opened Sep 19, 2026 by
0z5a
Loading…
[docs][sdpa] Document native execute-time shape override boundaries
#1148
opened Sep 19, 2026 by
0z5a
Loading…
engines: offer opt-in FROST engines without the flag when the cuDNN backend proposes no plan or declines every plan at build
area:frost
cat-feature
Requests for new functionality, APIs, examples, or behavior improvements.
mod-frost
orig-nv-eng
Reported or requested by NVIDIA engineering.
Enable Rubin MXFP8 clamped GeGLU forward and backward
#1144
opened Sep 18, 2026 by
ZeYang1025
Loading…
3 of 4 tasks
Frost sdpa fwd d64
area:frost
cat-feature
Requests for new functionality, APIs, examples, or behavior improvements.
orig-nv-eng
Reported or requested by NVIDIA engineering.
test(sdpa): CP numerics reference, TE merge oracle and CP1/CP2/CP4 replay
#1142
opened Sep 18, 2026 by
0z5a
Loading…
sdpa: CUDA graph topology evidence and an opt-in direct populate/update path
#1140
opened Sep 18, 2026 by
0z5a
Loading…
sdpa: validate the derived O layout after inference, not in pre_validate_node
#1139
opened Sep 18, 2026 by
0z5a
Loading…
Add native Mamba-2 SSD forward and backward graph engine
area:frost
area:linear_attention
cat-feature
Requests for new functionality, APIs, examples, or behavior improvements.
orig-nv-eng
Reported or requested by NVIDIA engineering.
#1133
opened Sep 18, 2026 by
YangXu1990uiuc
Collaborator
Loading…
fix: acquire SM100 grouped-wgrad tensor maps before TMA use
cat-bugfix
mod-cutedsl
CuTeDSL kernels, generated kernels, examples, or related integration work.
orig-nv-eng
Reported or requested by NVIDIA engineering.
[Do not merge] CI heartbeat
cat-ci
CI failures, test flakiness, workflow breakage, or automation issues.
mod-infra
Infrastructure, CI/CD, build systems, packaging, releases, or repo maintenance.
orig-nv-eng
Reported or requested by NVIDIA engineering.
[DSv4.1] Fuse native-layout vision RoPE backward and QKV gradient packing
area:frost
cat-feature
Requests for new functionality, APIs, examples, or behavior improvements.
orig-nv-eng
Reported or requested by NVIDIA engineering.
#1122
opened Sep 17, 2026 by
YangXu1990uiuc
Collaborator
•
Draft
[DSv4.1] Add FROST prepared BF16 tail RoPE for prefill
area:frost
cat-feature
Requests for new functionality, APIs, examples, or behavior improvements.
orig-nv-eng
Reported or requested by NVIDIA engineering.
#1116
opened Sep 17, 2026 by
YangXu1990uiuc
Collaborator
•
Draft
[DSv4.1] Extend FROST RoPE QDQ with group16 E4M3-scale FP4
area:frost
cat-feature
Requests for new functionality, APIs, examples, or behavior improvements.
orig-nv-eng
Reported or requested by NVIDIA engineering.
#1115
opened Sep 17, 2026 by
YangXu1990uiuc
Collaborator
•
Draft
[DSv4.1] Add FROST tail RoPE and group32 FP4/FP8 QDQ
area:frost
cat-feature
Requests for new functionality, APIs, examples, or behavior improvements.
orig-nv-eng
Reported or requested by NVIDIA engineering.
#1114
opened Sep 17, 2026 by
YangXu1990uiuc
Collaborator
•
Draft
[DSv4.1] Add FROST saved-state Engram gate forward and backward
area:frost
cat-feature
Requests for new functionality, APIs, examples, or behavior improvements.
orig-nv-eng
Reported or requested by NVIDIA engineering.
#1113
opened Sep 17, 2026 by
YangXu1990uiuc
Collaborator
•
Draft
frost(sdpa_bwd_sm80): serve THD ports at their own head stride
area:frost
area:global_attention
cat-enhancements
orig-nv-eng
Reported or requested by NVIDIA engineering.
Previous Next
ProTip!
Updated in the last three days: updated:>2026-09-17.