Archives Discussions

[acl075]

Memory Bench Marks and Instances

Hi Experts,

I am a new bee to GPU programming. I was wondering if there are any benchmarks for the CPU<->GPU memory transfers and internal memory transfers inside the device?

Also I would appreciate someone can point on the impact when multiple instances of a data intensive component e.g. video encoders are run in a single GPU card (any slow down expected, any cap on the number of such invocations).

Thank you in anticipation, greatly appreciate your inputs

[nou]

in OpenCL SDK samples there are some benchmarks and this http://developer.amd.com/gpu/ATISTREAMPOWERTOY/Pages/default.aspx

[sorasak]

thanks very much