Skip to content

Update ort and ort-genai package - #934

Open
baijumeswani wants to merge 3 commits into
mainfrom
baijumeswani/upgrade-packages
Open

Update ort and ort-genai package#934
baijumeswani wants to merge 3 commits into
mainfrom
baijumeswani/upgrade-packages

Conversation

@baijumeswani

Copy link
Copy Markdown
Collaborator

No description provided.

Copilot AI review requested due to automatic review settings July 30, 2026 23:40
@vercel

vercel Bot commented Jul 30, 2026

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated (UTC)
foundry-local Ready Ready Preview Aug 2, 2026 7:16am

Request Review

Copilot AI left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Pull request overview

Updates ONNX Runtime dependencies and standardizes native package consumption on the CPU package.

Changes:

  • Bumps ORT to 1.28.0 and ORT GenAI to 0.15.0.
  • Replaces Foundry/GPU-specific ORT packages with Microsoft.ML.OnnxRuntime.
  • Updates pipeline prefetching, packaging, and documentation.

Reviewed changes

Copilot reviewed 11 out of 11 changed files in this pull request and generated 1 comment.

Show a summary per file
File Description
sdk_v2/js/script/install-native.cjs Uses the standard ORT package.
sdk_v2/deps_versions.json Updates shared dependency versions.
sdk_v2/cpp/nuget/pack.py Updates ORT package help text.
sdk_v2/cpp/nuget/Microsoft.AI.Foundry.Local.Runtime.nuspec Changes the runtime dependency.
sdk_v2/cpp/cmake/FindOnnxRuntime.cmake Simplifies standard ORT acquisition.
.pipelines/v2/templates/steps-prefetch-nuget.yml Updates package prefetching.
.pipelines/v2/templates/steps-build-windows.yml Updates parameter documentation.
.pipelines/v2/templates/steps-build-linux.yml Removes GPU-package prefetching.
.pipelines/v2/sdk_v2-pipeline-plan.md Documents CPU-only Linux builds.
.pipelines/v2/sdk_v2-js-pipeline-plan.md Documents the standard ORT package.
.pipelines/foundry-local-packaging.yml Synchronizes pipeline version pins.
Comments suppressed due to low confidence (1)

.pipelines/v2/templates/steps-prefetch-nuget.yml:132

  • After removing the GPU package, this list contains only two packages, so “All four packages” is now incorrect in the bash branch as well.
      declare -a entries=(
        "genai:Microsoft.ML.OnnxRuntimeGenAI.Foundry:${{ parameters.genaiVersion }}"
        "ort:Microsoft.ML.OnnxRuntime:${{ parameters.ortVersion }}"
      )

Comment on lines 82 to 84
@{ key = 'genai'; id = 'Microsoft.ML.OnnxRuntimeGenAI.Foundry'; version = '${{ parameters.genaiVersion }}' },
@{ key = 'ort'; id = 'Microsoft.ML.OnnxRuntime.Foundry'; version = '${{ parameters.ortVersion }}' }
@{ key = 'ort'; id = 'Microsoft.ML.OnnxRuntime'; version = '${{ parameters.ortVersion }}' }
)
…512 FP nondeterminism

The exact digit a 0.5B greedy model emits is not stable across CPU ISAs: ORT 1.28
routes FP32 GQA single-token decode through a new flash/online-softmax GEMV kernel
whose softmax reductions dispatch to AVX-512F variants on AVX-512 hosts (e.g. AMD
EPYC 9V74) vs AVX2 elsewhere, giving a different FP accumulation order. A few-ULP
delta flips the greedy argmax between near-tie digit tokens (CI: turn 2 5->3,
turn 4 6->2). This is mathematically-equivalent FP reordering, not an SDK bug.

Keep all mechanics assertions strict (finish reason, token counts, turn count, and
the turn-3==turn-2 rewind-determinism check, which is bit-stable on a single host).
Emit the exact-digit expectations as non-fatal GTEST_LOG_(WARNING) warnings, and
document the cause plus less-invasive alternatives (ORT-level determinism test,
wider-margin prompt, ORT_GQA_DISABLE_FLASH_ATTENTION=1) in a comment.
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