Instruction Throughput - Nvidia
Instruction Throughput - Nvidia
Instruction Throughput - Nvidia
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Instruction</strong> <strong>Throughput</strong>: Summary<br />
� Analyze:<br />
© NVIDIA Corporation 2011<br />
� Check achieved instruction throughput<br />
� Compare to HW peak (note: must take instruction mix into consideration)<br />
� Check percentage of instructions due to serialization<br />
� Optimizations:<br />
� Intrinsics, compiler options for expensive operations<br />
� Group threads that are likely to follow same execution path<br />
� Avoid SMEM bank conflicts (pad, rearrange data)<br />
16