Skip to content
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.

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

Bump IREE requirement pins to their latest versions. #835

Merged
merged 4 commits into from
Jan 20, 2025
Merged

Conversation

shark-pr-automator[bot]
Copy link
Contributor

@shark-pr-automator shark-pr-automator bot commented Jan 16, 2025

@@ -55,7 +55,6 @@ def setUp(self):
"--iree-opt-data-tiling=false",
"--iree-preprocessing-pass-pipeline='builtin.module(util.func(iree-preprocessing-generalize-linalg-matmul-experimental))'",
"--iree-stream-resource-memory-model=discrete",
"--iree-hip-legacy-sync=false",
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I can't tell how to update the SDXL files on Azure to fix https://github.com/nod-ai/shark-ai/actions/runs/12834154632/job/35790824168?pr=835#step:7:54 . There are multiple versions of the files on Azure and different parts of this project use different files. Each folder has a full copy of the 12GB+ model artifacts, and it doesn't make sense to fork that many files to remove 1 flag.

subprocess.CalledProcessError: Command '['/home/nod/actions-runner-shark-ai/_work/_tool/Python/3.12.7/x64/bin/python', '-m', 'iree.build', '/home/nod/actions-runner-shark-ai/_work/shark-ai/shark-ai/shortfin/python/shortfin_apps/sd/components/builders.py', '--model-json=sdxl_config_i8.json', '--target=gfx942', '--splat=True', '--build-preference=precompiled', '--output-dir=/home/nod/.cache/shark', '--model=clip', '--iree-hal-target-device=amdgpu', '--iree-hip-target=gfx942', "--iree-compile-extra-args=--iree-hal-target-backends=rocm --iree-hip-target=gfx942 --iree-execution-model=async-external --iree-preprocessing-pass-pipeline='builtin.module(util.func(iree-global-opt-raise-special-ops, iree-flow-canonicalize), iree-preprocessing-transpose-convolution-pipeline, iree-preprocessing-pad-to-intrinsics, util.func(iree-preprocessing-generalize-linalg-matmul-experimental))' --iree-global-opt-propagate-transposes=1 --iree-opt-const-eval=0 --iree-opt-outer-dim-concat=1 --iree-opt-aggressively-propagate-transposes=1 --iree-dispatch-creation-enable-aggressive-fusion --iree-hal-force-indirect-command-buffers --iree-codegen-llvmgpu-use-vector-distribution=1 --iree-llvmgpu-enable-prefetch=1 --iree-codegen-gpu-native-math-precision=1 --iree-hip-legacy-sync=0 --iree-opt-data-tiling=0 --iree-vm-target-truncate-unsupported-floats"]' returned non-zero exit status 1.
ERROR
tests/apps/sd/e2e_test.py::test_sd_server_bs4_dense ERROR
[2/9] Fetch https://sharkpublic.blob.core.windows.net/sharkpublic/sdxl/01062025/configs/sdxl_flagfile_gfx942.txt
[3/9] Fetch https://sharkpublic.blob.core.windows.net/sharkpublic/sdxl/01062025/configs/attention_and_matmul_spec_gfx942.mlir
[4/9] Fetch https://sharkpublic.blob.core.windows.net/sharkpublic/sdxl/01062025/configs/topology_config_spx_single.txt
[5/9] Fetch https://sharkpublic.blob.core.windows.net/sharkpublic/sdxl/01062025/configs/sdxl_config_i8.json
[6/9] Fetch https://sharkpublic.blob.core.windows.net/sharkpublic/sdxl/01062025/configs/sdxl_config_i8.json
[7/9] Fetch https://sharkpublic.blob.core.windows.net/sharkpublic/sdxl/01062025/configs/sdxl_config_i8.json
[8/9] Fetch https://sharkpublic.blob.core.windows.net/sharkpublic/sdxl/01062025/configs/sdxl_config_i8.json
[9/9] Fetch https://sharkpublic.blob.core.windows.net/sharkpublic/sdxl/01062025/configs/sdxl_config_i8.json
Successfully built 9 actions
tests/apps/sd/e2e_test.py::test_sd_server_bs8_percall [2025-01-17 17:52:21.262] [info] [manager.py:28] Created local system with ['amdgpu:0:0@0'] devices
[2025-01-17 17:52:21.[57](https://github.com/nod-ai/shark-ai/actions/runs/12834154632/job/35790824168?pr=835#step:7:58)6] [info] [server.py:250] Preparing runtime artifacts for clip...
unused_prog_name: Unknown command line argument '--iree-hip-legacy-sync=0'.  Try: 'unused_prog_name --help'
unused_prog_name: Did you mean '--iree-hip-bc-dir=0'?

Can someone point me to some playbook docs for how to update the hosted files or help get the CI passing again?

Signed-off-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
@ScottTodd
Copy link
Member

Aw, the new automation force pushed over one fix patch: 9285135.

https://github.com/peter-evans/create-pull-request has some configuration we can try out.

@ScottTodd
Copy link
Member

Except for some preexisting failures in the "Llama Benchmarking 8B Tests" workflow, this is passing CI now.

@ScottTodd
Copy link
Member

Tests seem to be passing now... going to sync the branch and run tests again then merge 🤞

@ScottTodd ScottTodd merged commit f743696 into main Jan 20, 2025
37 checks passed
@ScottTodd ScottTodd deleted the integrates/iree branch January 20, 2025 23:42
ScottTodd added a commit to iree-org/iree-turbine that referenced this pull request Jan 21, 2025
This will allow downstream workflows like
https://github.com/nod-ai/shark-ai/blob/main/.github/workflows/update_iree_requirement_pins.yml
to get nightly IREE and iree-turbine packages without waiting longer.

See that nod-ai/shark-ai#835 got these versions
due to the timing:
```
iree-base-compiler==3.2.0rc20250120
iree-base-runtime==3.2.0rc20250120
iree-turbine==3.2.0rc20250119
```
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.

2 participants