jxt1234

icon indicating an email [email protected]

Results 338 comments of jxt1234

linux x86_64 平台， 多线程情况下， mnn 和推理速度不如 onnx，有可能是什么原因？

相关模型发一下吧，有可能是 featuremap 尺寸较小。不利于 mnn 并行。AVX512 上多线程并行需要以 48 * numberThread 的 featuremap 尺寸

Facing issue with version incapitability between protobuffer library and MNN library while performing mnnquant quantization process using python API of mnn

It seems libc++ not the same

Facing issue with version incapitability between protobuffer library and MNN library while performing mnnquant quantization process using python API of mnn

You can try to use gdb mnnquant to debug the crash stack. It may be caused by the image path has invalid picture. In the same time you can use...

MNN.cv has the rotation issue in loading some photos from the phone

MNN.cv use stbi. It currently don't support load rotation info from jpg's exif. We will fix it later.

mnncompress量化模型以后大小没有变化

估计是 mnncompress 里面没有识别出 linear 层出来量化

vs2022 x86下编译错误

可能是你的编译器不支持 AVX2 ，加上 -DMNN_AVX2=OFF 可以关掉 AVX2 相关的编译。也可以加 -DMNN_USE_SSE=OFF 关掉所有 simd 优化（不推荐）。

有一些算子不支持

用 mnn ，这些都支持：https://github.com/alibaba/MNN/

yolov11-obb 检测demo

已经解决就关闭了

doc多处链接失效

修改了一部分，还未修改完全

doc多处链接失效

2.9.5 修正