topic Sounds good, thanks for the in OpenCL* for CPU

Miscompiled control flow

Kloeckner__Andreas — Sat, 29 Jun 2013 01:55:56 GMT

Hi there,

I've attached a test case (a gutted version of a segmented scan) that gets miscompiled by intel_sdk_for_ocl_applications_2013_xe_runtime_3.0.67279_x64, at least when running on my i7-2620M.

Specifically, when running this self-contained test code using PyOpenCL, I get the line

gid:0 fsii:0

printed 16 times for each of the two work groups. If you look at the kernel, that means that the printf() in the trailing snippet:

    if (get_local_id(0) == 0)
    {
      printf("gid:%d fsii:%d\n", psc_GID_0, psc_first_segment_start_in_interval);
    }

got executed 16 times for group id 0. In my book, it should be executed exactly once. (Confirmed by running against other implementations. Intel OpenCL 2012 also gets this right.)

Thanks,

Andreas

Sorry for late response.

Raghupathi_M_Intel — Fri, 12 Jul 2013 17:09:33 GMT

Sorry for late response. Thanks for the code. We'll take a look and get back to you.

Hi Andreas,

Yuri_K_Intel — Mon, 15 Jul 2013 10:27:39 GMT

Hi Andreas, I was able to reproduce the issue on the release version (67279), but I don't see it on our latest development versions, the output I get is: gid:0 fsii:0 gid:1 fsii:0 So obviously it has been fixed since that. Thanks, Yuri

Sounds good, thanks for the

Kloeckner__Andreas — Wed, 17 Jul 2013 15:52:58 GMT

Sounds good, thanks for the heads-up. How soon do you think your dev version will turn into a release?

Thanks, Andreas

Hi, the control flow you have

JLuna5 — Tue, 23 Jul 2013 06:46:50 GMT

Hi, the control flow you have established on the estructure