DeepSpeed
fix(zero3): async grad offload + pinned offload buffers by default
#8207
Merged

fix(zero3): async grad offload + pinned offload buffers by default #8207

delock
delock fix(zero3): async grad offload + pinned offload buffers by default
d6f43c5d
delock refactor(zero3): remove orphaned async grad copy stub
d537ae37
delock delock marked this pull request as ready for review 16 days ago
delock delock requested a review from tjruwase tjruwase 16 days ago
delock delock requested a review from tohtana tohtana 16 days ago
delock Merge branch 'master' into gma/async_zero_offload
d4bc2fc3
chatgpt-codex-connector
chatgpt-codex-connector commented on 2026-08-04
delock delock marked this pull request as draft 16 days ago
delock docs(zero): document pin_memory default, mechanism, and memlock caveat
12bdbc51
delock delock force pushed from 4648e397 to a15185d1 16 days ago
delock delock force pushed from a15185d1 to 6b1af38e 16 days ago
delock delock force pushed from 6b1af38e to 707f2d08 16 days ago
delock feat(accel): track pinned host memory for OOM diagnostics
fbb02770
delock delock force pushed from 707f2d08 to fbb02770 16 days ago
delock delock marked this pull request as ready for review 15 days ago
delock delock requested a review from loadams loadams 15 days ago
delock Merge branch 'master' into gma/async_zero_offload
0b8426f4
sfc-gh-truwase
sfc-gh-truwase commented on 2026-08-07
sfc-gh-truwase Merge branch 'master' into gma/async_zero_offload
ef6a75dc
sfc-gh-truwase
sfc-gh-truwase approved these changes on 2026-08-07
sfc-gh-truwase sfc-gh-truwase merged 7dfa67ae into master 13 days ago

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone