Out of memory when compiling 'big' kernel for HD4600 GPU

[Hua_Z_]

In fact my kernel file is not big. It contains about 400 lines code, 10 funtions and one kernel function. Some code in kernel function were commented out. Gradually recovering the code from comments, I found when total quantity of code exceeds some threadhold, clBuildProgram will complain error:

Error in clBuildProgram (-11): CL_BUILD_PROGRAM_FAILURE
fcl build 1 succeeded.
fcl build 2 succeeded.
Error: out of memory.

As far as I know, Intel Core GPU does not have its own memory. It uses CPU memory to work. The physicsal memory is very big. Bigger quantity of code can be successfully compiled on AMD HD5870 descrete GPU card which contains only 1GB memory.

So, are there some solutions to solve the compilation OOM issue?
the CPU is i7-4710MQ , Windows 10
 

[Robert_I_Intel]

Hi Hua,

Sorry for replying late: somehow this post fell thru the cracks.

Without looking at your kernel code it is hard to tell what is going on. The error that you are seeing is in the backend compiler, which generates final GPU binary. Could you please either attach the code or send me a private email and attach email there? Also, please specify which Graphics Driver version are you using?

Thanks!

[Hua_Z_]

I terminated the programming on Intel platform due to the bug https://software.intel.com/en-us/forums/topic/559322 and accomplished my program on AMD platform . If the bug in topic 559322 was fixed, I may consider to continue the work on Intel platform.

[Robert_I_Intel]

Hi Hua,

Very sorry to hear that! Unfortunately, a lot of our people that could fix this bug are out for the summer, so it would be a while until your bug is fixed: I promptly filed it but the driver team is severely understaffed right now. I will notify you when the bug is fixed and maybe you will come back to our platform again.