1

I found the error when training using Caffe. Seems both of the errors are connected with memory because when I reduce the batch size, both the errors are solved.

Just, seems the misaligned address comes first before out of memory. For example, when I put batch size = 32 (or more), I got out of memory error. When I put batch size = 16, I got misaligned address error. When I put batch size = 8 (or less), there is no error.

Actually what is the difference between those errors and the meaning of the errors?

I use CUDA 8.0 and CuDNN 5.1

talonmies
  • 70,661
  • 34
  • 192
  • 269
Kalkaneus
  • 39
  • 7
  • check http://stackoverflow.com/questions/37323053/misaligned-address-in-cuda – pSoLT Jan 16 '17 at 12:09
  • 2
    CUDA alignment requirements are discussed [in the programming guide](http://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#device-memory-accesses). And presumably "out of memory" doesn't require much explaining - something that Caffe is doing or you are asking Caffe to do requires more memory than what is available on your GPU – Robert Crovella Jan 16 '17 at 12:13
  • This appears to be a known problem in Caffe with certain hardware https://github.com/BVLC/caffe/issues/2417 – talonmies Jan 16 '17 at 13:10

0 Answers0