[Performance] Performance regression in SVMClassifier operator between v1.18.0 and v1.19.0

[junghyunpark2001]

Describe the issue

Description

We observed a performance regression in the SVMClassifier operator (RBF kernel with float32 and probabilities) between ONNXRuntime v1.18.0 and v1.19.0.

Affected Operator

SVMClassifier

• Opset Version: 1
• Data Type: float32
• Configuration: RBF kernel with probabilities
• Regression: +9.34% slowdown
• Suspected Commit: 08ecf30e (not confirmed, requires git bisect)

Test Case Details

Model Information:

• Operator: SVMClassifier
• Opset: 1
• Input data type: float32
• Kernel type: RBF (Radial Basis Function)
• Output: class predictions with probabilities

Performance Comparison:

• v1.18.0: 0.003 ms (baseline)
• v1.19.0: 0.003 ms
• Regression: +9.34% slowdown

Root Cause Analysis

Suspected Commit

• Commit 08ecf30e is suspected based on version range analysis
• Further investigation (git bisect) recommended to confirm

Suspected Causes

The 9.34% regression in SVMClassifier suggests:

1. Changes in RBF kernel computation
2. Modifications to probability calculation logic
3. Inefficiencies in class prediction algorithms
4. Changes in memory access patterns or vectorization

To reproduce

1. Download and unzip test.zip

2. Run benchmark:

   python script.py ./svmclassifier 1.18.0 1.19.0

3. Compare the reported latencies between the two versions.

test.zip

Urgency

No response

Platform

Linux

OS Version

Ubuntu 24.04.3 LTS

ONNX Runtime Installation

Released Package

ONNX Runtime Version or Commit ID

08ecf30

ONNX Runtime API

Python

Architecture

X64

Execution Provider

Default CPU

Execution Provider Library Version

No response

Model File

No response

Is this a quantized model?

Yes

[microsoft-github-policy-service]

Applying stale label due to no activity in 30 days. Please react to the issue by assigning it or labeling it as 'contributions welcome', 'documentation', 'feature request', 'regression', or 'no stale'.