Skip to content

Commit

Permalink
[CPU EP] Add blocked quantization to DequantizeLinear op kernel (micr…
Browse files Browse the repository at this point in the history
…osoft#20901)

### Description
Added blocked quantization to DequantizeLinear op kernel. All existing
[input types and output
types](https://github.com/microsoft/onnxruntime/blob/main/docs/ContribOperators.md#commicrosoftdequantizelinear)
are supported. All axes are supported.

The implementation in the PR is naive - single thread and scalar
instructions. Multi-threading and vector instructions are planned in the
future based on the needs.


### Motivation and Context
onnx introduced blocked quantization in opset 21 for
[DequantizeLinear](https://github.com/microsoft/onnxruntime/blob/main/docs/ContribOperators.md#commicrosoftdequantizelinear).
This PR adds the spec support in onnx runtime.
  • Loading branch information
fajin-corp authored Jun 4, 2024
1 parent 5faeaf6 commit 3ecb012
Show file tree
Hide file tree
Showing 2 changed files with 1,024 additions and 91 deletions.
Loading

0 comments on commit 3ecb012

Please sign in to comment.