Skip to content

EWLayer && oneDNN#217

Merged
aobolensk merged 40 commits into
mainfrom
Semyon1104/IntegrateOneDNN
Nov 2, 2025
Merged

EWLayer && oneDNN#217
aobolensk merged 40 commits into
mainfrom
Semyon1104/IntegrateOneDNN

Conversation

@Semyon1104

@Semyon1104 Semyon1104 commented Oct 22, 2025

Copy link
Copy Markdown
Collaborator

Comment thread CMakeLists.txt Outdated
Comment thread src/layers_oneDNN/EWLayer_oneDNN.cpp Outdated
Comment thread src/layers_oneDNN/EWLayer_oneDNN.cpp Outdated
Comment thread src/layers_oneDNN/EWLayer.cpp
Comment thread src/layers/CMakeLists.txt Outdated
Comment thread app/Graph/build.cpp
void build_graph(it_lab_ai::Tensor& input, it_lab_ai::Tensor& output,
const std::string& json_path, bool comments, bool parallel) {
const std::string& json_path, bool comments, bool parallel,
bool onednn) {

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

Why graph depends on onednn?

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

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

the parameter is passed through build_graph -> parse_json_model to create a layer with or without oneDNN. I'm still thinking about the best way to organize this

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

The fact that the layer is being executed using oneDNN (or any other way) is purely dependent on how do we execute the ready-to-use graph. Actually "parallel" parameter is also not very good idea, and we will need to get rid of this (I have created a task out of it: #218, may be resolved later)

So, theoretically we should be able to take a graph out, serialize it anyhow, store it and be able to execute (without any re-configuration on the graph level) it in different ways: sequentially, in parallel, using SYCL/Kokkos/oneDNN, etc.

@Semyon1104 Semyon1104 force-pushed the Semyon1104/IntegrateOneDNN branch from 8cf3438 to b57dba8 Compare October 28, 2025 09:07
@codecov

codecov Bot commented Oct 28, 2025

Copy link
Copy Markdown

Codecov Report

❌ Patch coverage is 88.54167% with 11 lines in your changes missing coverage. Please review.
✅ Project coverage is 83.84%. Comparing base (3d2f33b) to head (24a4934).
⚠️ Report is 1 commits behind head on main.

Files with missing lines Patch % Lines
src/layers_oneDNN/EWLayer.cpp 87.91% 7 Missing and 4 partials ⚠️
Additional details and impacted files
@@            Coverage Diff             @@
##             main     #217      +/-   ##
==========================================
- Coverage   84.06%   83.84%   -0.23%     
==========================================
  Files          42       44       +2     
  Lines        2667     2655      -12     
  Branches     1397     1466      +69     
==========================================
- Hits         2242     2226      -16     
+ Misses        220      219       -1     
- Partials      205      210       +5     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

Comment thread .github/workflows/ci.yml
- name: Set binary path
id: set_eval_binary
run: |
echo "EVAL_BINARY=build/bin/ACC" >> $GITHUB_OUTPUT

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

Please, leave this as is. I think this output is redundant and the CI script becomes more fragile due to this. Similar comment is applicable for other modifications of CI scripts. Please, remove debug output here and all temporary workarounds

@Semyon1104 Semyon1104 force-pushed the Semyon1104/IntegrateOneDNN branch 3 times, most recently from a002eaf to c6ac48e Compare October 31, 2025 08:15
@Semyon1104 Semyon1104 force-pushed the Semyon1104/IntegrateOneDNN branch from ed4739a to c6ac48e Compare November 2, 2025 09:45
@aobolensk aobolensk linked an issue Nov 2, 2025 that may be closed by this pull request
@aobolensk aobolensk merged commit 4d83114 into main Nov 2, 2025
23 checks passed
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.

Integrate oneDNN library

2 participants