PointPillars_MultiHead_40FPS icon indicating copy to clipboard operation
PointPillars_MultiHead_40FPS copied to clipboard

about the performance in xavier

Open likegogogo opened this issue 3 years ago • 4 comments

I used the provided onnx and run it in xavier in MAXN mode. Running onnx model, takes about 130ms. Running trt model, takes about 130ms. I'm sure it had used GPU.

这个速度 正常吗?

likegogogo avatar Oct 09 '21 01:10 likegogogo

正常的,这个设置的点云范围比较大,加速还得再把模型搞小点

rolan123 avatar Oct 12 '21 09:10 rolan123

@ideafold 你好,请问你用的xavier是哪一款?NX 还是 AGX?

speshowBUAA avatar Nov 19 '21 03:11 speshowBUAA

@ideafold 你好,请问你用的xavier是哪一款?NX 还是 AGX?

agx,32G mem, 32 emmc.

likegogogo avatar Nov 20 '21 00:11 likegogogo

正常的,这个设置的点云范围比较大,加速还得再把模型搞小点

@rolan123 请问有没有尝试过在OpenPCDet里更改cbgs_pp_multihead.yaml和nuscenes_dataset.yaml中的点云范围参数 POINT_CLOUD_RANGE: [-11.2, -11.2, -5.0, 11.2, 11.2, 3.0],来达到压缩模型大小的效果?我尝试后训练结果出现了异常

speshowBUAA avatar Dec 27 '21 03:12 speshowBUAA