I want to tune my OpenCL kernel running on HD 4000 of my i5-3317U processor, on Windows 7. I spent days to exploit the tools helping me to tackle the bottleneck of my kernel. However, I was quite disappointing about the tools provided by Intel.
For the Intel VTune Amplifer , it only support kernels running on the CPU. So when I using this to profile my kernel, it reported more than 80% EU stall of my kernel, which I assume, is not correct
For the GPA, I read the manual of Getting started, it provide a sample executable for you to display HUD. However, if I need passing parameters to my kernel, it report me incorrect executable. and I can not find any textbox in the GPA monitor to pass my kernel parameters.
So could any one with more experience suggest me how to tune the kernel on GPU with appropriate tools on the Ivy bridge processor?