DefaultCPUAllocator: not enough memory

backseason / PoolNet

Code for our CVPR 2019 paper "A Simple Pooling-Based Design for Real-Time Salient Object Detection"

MIT License

630 stars 153 forks source link

DefaultCPUAllocator: not enough memory #45

Open Omri-L opened 4 years ago

Omri-L commented 4 years ago

Hi, I keep getting this memory error every once in a while when I run "joint_main" test with batch size =1. I thought that you might have advice for me to overcome that.

backseason commented 4 years ago

I have never met with this issue. But I guess this may relates to the memory size of you hardware.

Adam-D-Lewis commented 4 years ago

I believe I experienced a similar problem. I was running out of memory (used > 10GB) when running the following configuration: --mode=test --model=results/run-1/models/final.pth --test_fold=results/run-1-sal-e --sal_mode=e --no-cuda --test_mode=1

I believe it is due to a memory leak in recent versions of pytorch (was using pytorch==1.3.1). I downgraded to pytorch==0.4.1, and the memory usage is stable now.