how to know if the installation is ok?

[nOon29]

Hi so i have install drivers,toolkit and sdk. I can run the sdk program but how to know if the code run on gpu?
Because most of the nvidia example have a very short time execution an most of them use the cpu to compare results. So with a top i can see the use of cpu but not the use of cpu.
It is possible to find a big matrix multiplication for exemple who one code use the gpu and the other in C use the cpu.
Thanks for answer

[porQue]

Hi so i have install drivers,toolkit and sdk. I can run the sdk program but how to know if the code run on gpu?

Because most of the nvidia example have a very short time execution an most of them use the cpu to compare results. So with a top i can see the use of cpu but not the use of cpu.

It is possible to find a big matrix multiplication for exemple who one code use the gpu and the other in C use the cpu.

Thanks for answer

You might be able to write a short Matrix Multiplication, run it on the CPU, and time it, I think NVIDIA offers a sample Matrix Multiplication that you can also run and time that as well. You should be able to run it faster on the GPU.

[lixingjian]

Hi so i have install drivers,toolkit and sdk. I can run the sdk program but how to know if the code run on gpu?

Because most of the nvidia example have a very short time execution an most of them use the cpu to compare results. So with a top i can see the use of cpu but not the use of cpu.

It is possible to find a big matrix multiplication for exemple who one code use the gpu and the other in C use the cpu.

Thanks for answer

You can run deviceQuery or bandwithtest in SDK to verify if your GPU is working.