Replies: 1 comment 2 replies
-
好的,等我写完2.0.0的rapidocr。
感兴趣的话,可以加个微信。微信号发我邮箱,我加你。
***@***.***
…---- 回复的原邮件 ----
| 发件人 | ***@***.***> |
| 发送日期 | 2025年02月15日 17:41 |
| 收件人 | RapidAI/RapidOCR ***@***.***> |
| 抄送人 | Subscribed ***@***.***> |
| 主题 | [RapidAI/RapidOCR] onnxruntime-rocm可以使用,识别速度挺快的 (Discussion #340) |
问题描述 / Problem Description
上次用核显ROCM加速效果一般,这次换RX6600加速效果可以。刚好和pytorch对比速度。
运行环境 / Runtime Environment
Ubuntu 24.04
Python 3.12.3
复现代码 / Reproduction Code
修改infer_engine.py,把CPUExecutionProvider替换为ROCMExecutionProvider。参考https://rocm.docs.amd.com/projects/radeon/en/latest/docs/install/native_linux/install-onnx.html
pip3 uninstall onnxruntime-rocm
pip3 install onnxruntime-rocm -f https://repo.radeon.com/rocm/manylinux/rocm-rel-6.3.2/
结果(十次平均)
效率提高3.5倍
GPU RX6600:[0.09092569040001308, 0.0501723051071167, 0.2702143430709839]
CPU 5600: [0.3116320829999495, 0.02931690216064453, 1.090583086013794]
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>
|
Beta Was this translation helpful? Give feedback.
2 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
问题描述 / Problem Description
上次用核显ROCM加速效果一般,这次换RX6600加速效果可以。刚好和pytorch对比速度,看看A卡有没有必要多安装一个pytorch。
运行环境 / Runtime Environment
复现代码 / Reproduction Code
修改
infer_engine.py
,把CPUExecutionProvider
替换为ROCMExecutionProvider
。参考https://rocm.docs.amd.com/projects/radeon/en/latest/docs/install/native_linux/install-onnx.html,安装ROCM和onnxruntime-rocm。结果(十次平均)
效率提高3.5倍
[0.09092569040001308, 0.0501723051071167, 0.2702143430709839]
[0.3116320829999495, 0.02931690216064453, 1.090583086013794]
Beta Was this translation helpful? Give feedback.
All reactions