3

I am trying to run the Lua based program from the OpenNMT. I have followed the procedure from here : http://forum.opennmt.net/t/text-summarization-on-gigaword-and-rouge-scoring/85

I have used the command:

 th train.lua -data textsum-train.t7 -save_model textsum1 -gpuid 0 1 2 3 4 5 6 7

I am using 8 GPUs but still the process is damn slow as if the process is working on the CPU. kindly, let me know what might be the solution for the optimizing the GPU usage.
Here is the stats of the GP usage:
GPUSTATS

Kindly, let me know how I can make the process run faster using the complete GPUs. I am available with 11GBs, but the process only consumes 2 GB or less. Hence the process is damn slow.

Community
  • 1
  • 1
Jaffer Wilson
  • 7,029
  • 10
  • 62
  • 139
  • I suggest you contact Jean or Guillaume on the OpenNMT forum, they are very active and ready to help. Have you solved the issue yet? – Wiktor Stribiżew Jul 20 '17 at 12:28
  • Yes solved .. got it. – Jaffer Wilson Jul 20 '17 at 12:34
  • Then feel free to post as an answer. – Wiktor Stribiżew Jul 20 '17 at 12:34
  • 1
    The index of gpu starts with 1 not 0. 0 means CPU. – Jaffer Wilson Jul 20 '17 at 12:34
  • @WiktorStribiżew Dear friend, Thank you for your caring. But it is almost year passed. I was working on it previous but then it went from my hand to another. Now I am working on somethig different concept. Hence, I won't be able to give you more details about it. My apologize. – Jaffer Wilson Jul 06 '18 at 10:44
  • @WiktorStribiżew hey.. no need for any sorry.. what you did is extremely liked by me. there is nothing to express my feelings on SO otherwise I would have used those expression... What I was expecting is help and what you gave is same.... :) :) ;) – Jaffer Wilson Jul 06 '18 at 10:49

1 Answers1

3

As per OpenNMT documentation, you need to remove 0 from right after the gpuid option since 0 stands for the CPU, and you are effectively reduce the training speed to that of a CPU-powered one.

To use data parallelism, assign a list of GPU identifiers to the -gpuid option. For example:

th train.lua -data data/demo-train.t7 -save_model demo -gpuid 1 2 4

will use the first, the second and the fourth GPU of the machine as returned by the CUDA API.

Wiktor Stribiżew
  • 607,720
  • 39
  • 448
  • 563