search

PaddlePaddle / PaddleOCR

Awesome multilingual OCR toolkits based on PaddlePaddle (practical ultra lightweight OCR system, support 80+ languages recognition, provide data annotation and synthesis tools, support training and deployment among server, mobile, embedded and IoT devices)
https://paddlepaddle.github.io/PaddleOCR/
Apache License 2.0
42.41k stars 7.65k forks source link

PaddleOCR v4模型使用Mkldnn在非AVX512 CPU上变得非常慢 #10337

Closed sdcb closed 7 months ago

sdcb commented 1 year ago

请提供下述完整信息以便快速定位问题/Please provide the following information to quickly locate the problem

AVX512 CPU上速度正常，速度大约是1.3s 在其它CPU上奇慢无比，速度大约是60秒~100秒 （虚拟机CPU都是4核）

LFOV6P1M3% Z@$(DB5TM7HB

使用Openblas或者Onnx速度都正常，使用ppocr-v3识别模型也正常。

raoyutian commented 1 year ago

预测库的Mkldnn.dll问题。

shiyutang commented 1 year ago

以上回答已经充分解答了问题，如果有新的问题欢迎随时提交issue，或者在此条issue下继续回复～ 我们开启了飞桨套件的ISSUE攻关活动，欢迎感兴趣的开发者参加：PaddlePaddle/PaddleOCR#10223

sdcb commented 1 year ago

不对，上面只解决了2.5推理库准确率下降的问题，未解决ppocr-v4性能慢的问题，它们是两个问题，由于我没办法修改问题状态，因此我重开了新问题。

github-actions[bot] commented 8 months ago

This issue has been automatically marked as stale because it has not had recent activity. It will be closed in 7 days if no further activity occurs. Thank you for your contributions.