headmyshoulder
Reputation: 6310
Cuda performance problem
maybe this question is a little too general but maybe someone has experience the same problem.
I have a small cuda program using thrust and everything is fine. It works as expected.
But when I port the same program to another machine the performance of the whole program is about 10 to 100 times slower than the original machine. The hardware of the second machine should be faster (GTX480 instead of GTX265), so I guess the problem is on the second machine itsself. Does anyone has an idea what might go wrong?
Upvotes: 1
Views: 285
Answers (1)
Related Questions
- CUDA Thrust performance
- CUDA: slower thrust::reduce after calling thrust::for_each_n
- CUDA performance issue
- Simple Thrust code performs about half as fast as my naive cuda kernel. Am I using Thrust wrong?
- Cuda thrust global memory writing very slow
- Thrust issue with CUDA 6 managed memory
- NVidia CUDA Thrust device vector allocation is too slow
- CUDA Thrust slow when operating large vectors on my machine
- CUDA: Why Thrust is so slow on uploading data to GPU?
- cuda: different answer between cpu and gpu reduce