- User Since
- Dec 29 2016, 12:44 AM (33 w, 3 d)
Sat, Aug 12
Fri, Aug 11
I have re-enabled the previous offloading tests and moved the new GPU offloading tests to a new file which is disabled for linux (for now).
Thu, Aug 10
First of all, I apologize if I've upset you with my previous post. I am actively working on understanding what is causing these issues. It is not my intention to write tests that work on local configurations only. I am upset to see that these tests keep failing for your and maybe other configurations. Without knowing the actual reason of the failures I can only speculate what is going wrong with them hence the flurry of changes.
Wed, Aug 9
310549 should solve this problem by using a default architecture that is supported by the current device version.
Thanks for running the test on your machine! This is very useful.
I've removed that test. Let's see if the other two tests pass or not.
Revision 310505 fixes the tests for this patch.
Already covered by D34888
Tue, Aug 8
I have just pushed a fix, revision 310433.
Is that the last access to CachedResults before the error?
Mon, Aug 7
Add -no-canonical-prefixes to tests.
Fix test comments.
Sun, Aug 6
Fix -march special casing.
Sat, Aug 5
Don't exclude flags when host matches offload toolchain.
New way to handle OpenMP target flags.
Jul 10 2017
I think I have something that works which is similar to what you were requesting. Please let me know your thoughts!
Jul 6 2017
Jul 5 2017
Rebase on latest master.
Does this also include the fixes in the following revision?
Jun 30 2017
@hfinkel I've add the flag as suggested. There is one minor change, I used "=" instead of ":" when specifying the toolchain/triple. I also support the triple being omitted when there is only one offloading toolchain specified with -fopenmp-targets.
Check -fopenmp-targets has one entry when using default toolchain in -Xopenmp-target.
Pass OpenMP target options.
Jun 29 2017
Jun 28 2017
[Update regression tests] Add a test for propagating the compute capability to the OpenMP device offloading toolchain which targets NVIDIA GPUs.
This is a test for patch D34784 which is enabled by this patch.
Split previous diff into a "device offloading kind" patch (show here) and a new patch D34784 which relies on a new compiler flag.
These changes are related to looking up the .bc library for inlining purposes. I believe @arpith-jacob has already handled this in trunk. Therefore this is obsolete code.
Updated diff to address comments.
May 12 2017
May 9 2017
May 2 2017
May 1 2017
@rnk any further thought on the changes to this patch? Thanks :)
Change output of unbundler to produce cubin files when using OpenMP to offload to NVIDIA GPUs.
Apr 21 2017
Refactor if condition.
Apr 20 2017
Apr 19 2017
Avoid renaming by enabling PTXAS to generate an output file with the appropriate extension, in this case a cubin extension.
Apr 17 2017
Merge IF statements.
Apr 13 2017
Re-use an already existing flag rather than creating a new one.
Use the rename() utility function of LLVM for renaming the PTXAS output before invoking NVLINK.
Remove tests which belong into a different patch.
Apr 12 2017
Update check before loop.
Fix for loop range.