Skip to content

sGPU Test Scheduling: Global Work Queue #696

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
VeeraRajasekhar wants to merge 9 commits into
base: dev
Choose a base branch
from
Draft

sGPU Test Scheduling: Global Work Queue #696

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
935b4b7
Ci test run optimization
VeeraRajasekhar Jul 31, 2026
6f4c3e3
Removed mgpu runs and use previous built wheels
VeeraRajasekhar Jul 31, 2026
8594f49
Revert "Removed mgpu runs and use previous built wheels"
VeeraRajasekhar Aug 2, 2026
b32560c
Refactor logging
VeeraRajasekhar Aug 2, 2026
cc8f0fb
Refactor implementation
VeeraRajasekhar Aug 5, 2026
0f812af
Refactored the code
VeeraRajasekhar Aug 6, 2026
2f91be8
Refactored the code
VeeraRajasekhar Aug 7, 2026
d5a5ed8
Deleted unwanted files
VeeraRajasekhar Aug 7, 2026
e7c1d09
Refactor ci scripts
VeeraRajasekhar Aug 7, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
19 changes: 19 additions & 0 deletions .github/scripts/ci_sgpu_queue.conf
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,19 @@
# Copyright (c) 2026, Advanced Micro Devices, Inc. All rights reserved.
#
# See LICENSE for license information.
#
# sGPU suite definitions for the work-queue scheduler (run_queue_sgpu.sh).
# Format: <label> <logfile> <mode> <command> [args...]
#
# mode=list suite honours TE_CI_LIST_ONLY, so it is expanded into one work
# item per test invocation and those items are packed across all
# GPUs. Items are re-dispatched with TEST_FILTER=<tag>.
# mode=opaque suite is scheduled as a single work item (whole script).
#
# Unlike ci_sgpu_jobs.conf, line order carries no GPU assignment: every item
# from every suite goes into one queue and workers pull on completion.

examples examples.log opaque .github/scripts/run_examples.sh
torch torch.log list ci/pytorch.sh
jax jax.log list ci/jax.sh
core core.log opaque ci/core.sh
Loading