Multi GPU possible error ?

[Ashwin Nair]

up vote down votefavorite

I have installed caffe on a PC with Multi GPU. I have data with is in HDF5 data format each with 2000 samples and each sample is a matrix in the format 48 x 58x58. Its of the format 48 layers with a resolution 58x58. The solver and model is here. In this case I have only used build/tools/caffe train -solver models/mv16f/solver.prototxt -gpu 0 The output I get when i do the train is here. But When I run the same program on a single GPU pc I get output like this. Which leads me to think maybe there is some setting or something else wrong with the one running on the multi GPU pc.

Same Question Posted on Stackoverflow http://stackoverflow.com/questions/34766292/training-issue-with-caffe-hdf5-data-is-this-output-normal