Skip to content

Devmem - #1

Merged
archlitchi merged 6 commits into
masterfrom
devmem
Sep 14, 2021
Merged

Devmem#1
archlitchi merged 6 commits into
masterfrom
devmem

Conversation

@archlitchi

Copy link
Copy Markdown
Member

initial merge

@archlitchi
archlitchi merged commit f332dbb into master Sep 14, 2021
@archlitchi
archlitchi deleted the devmem branch September 14, 2024 08:49
bobsongplus pushed a commit to bobsongplus/HAMi that referenced this pull request Oct 21, 2024
…e_type

add mul device type use  nvidia.com/gpu-h100 like requests/limits
youyos pushed a commit to youyos/HAMi that referenced this pull request Jul 30, 2025
hami-robot Bot pushed a commit that referenced this pull request May 6, 2026
#1820)

* fix(scheduler): guard against zero-value division in ComputeScore (#1780)

ComputeScore and ComputeDefaultScore divide by device capacity fields
(Count, Totalcore, Totalmem) without zero-checks. When any field is 0,
float32 division produces NaN/Inf, which breaks sort.Sort's ordering
requirement and causes an index-out-of-range panic — crashing the
entire scheduler extender and halting all GPU scheduling cluster-wide.

Return score 0 for devices/nodes with zero capacity so the scheduler
can gracefully mark them as unschedulable instead of panicking.

Signed-off-by: lin121291 <4jp33f9e@gmail.com>

* fix(scheduler): add nil check for Device pointer in ComputeScore

Signed-off-by: lin121291 <4jp33f9e@gmail.com>

---------

Signed-off-by: lin121291 <4jp33f9e@gmail.com>
mesutoezdil pushed a commit to mesutoezdil/HAMi that referenced this pull request Jun 14, 2026
…oject-HAMi#1… (Project-HAMi#1820)

* fix(scheduler): guard against zero-value division in ComputeScore (Project-HAMi#1780)

ComputeScore and ComputeDefaultScore divide by device capacity fields
(Count, Totalcore, Totalmem) without zero-checks. When any field is 0,
float32 division produces NaN/Inf, which breaks sort.Sort's ordering
requirement and causes an index-out-of-range panic — crashing the
entire scheduler extender and halting all GPU scheduling cluster-wide.

Return score 0 for devices/nodes with zero capacity so the scheduler
can gracefully mark them as unschedulable instead of panicking.

Signed-off-by: lin121291 <4jp33f9e@gmail.com>

* fix(scheduler): add nil check for Device pointer in ComputeScore

Signed-off-by: lin121291 <4jp33f9e@gmail.com>

---------

Signed-off-by: lin121291 <4jp33f9e@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant