iGPU cache issue

[itjianggit]

OpenVINO Version

tag 2023.0.1

Operating System

Ubuntu 20.04 (LTS)

Device used for inference

iGPU

OpenVINO installation

Build from source

Programming Language

C++

Hardware Architecture

x86 (64 bits)

Model used

BSRN

Model quantization

No

Target Platform

models: model.zip

Performance issue description

Question 1: If the input size is different, the pre processed model cannot be reused and needs to be re read into the model for processing. At this time, the GPU cache is not released, greatly increasing memory usage. Is there a solution? The result is in the first compressed package.

Question 2: The cache consumed by the GPU is much higher than the CPU, for example, the 1080p image has already experienced OOM when the GPU zooms in twice. The result is in the second compressed package.

1020x678.zip

1080p.zip

Step-by-step reproduction

Read in the model, default model size 350x640, modify the model input tensor, and use CPU/GPU for inference.

Issue submission checklist

• [X] I'm reporting a performance issue. It's not a question.
• [X] I checked the problem with the documentation, FAQ, open issues, Stack Overflow, etc., and have not found a solution.
• [X] There is reproducer code and related data files such as images, videos, models, etc.

[github-actions[bot]]

This issue will be closed in a week because of 9 months of no activity.

[github-actions[bot]]

This issue was closed because it has been stalled for 9 months with no activity.