- User Since
- Apr 4 2014, 4:14 AM (206 w, 6 d)
Fri, Mar 16
Wed, Mar 14
LGTM, just fix formatting before submit.
Tue, Mar 13
I guess that is impossible to have two nodes which only differ by this bit. LGTM.
Fri, Mar 9
Thu, Mar 8
Wed, Mar 7
Given performance benefit is somewhat unclear can you put it under an option?
Tue, Mar 6
Mon, Mar 5
Fri, Mar 2
You can leverage higher alignment after you have actual allocation map.
I am not sure why this new and instruction shall be needed. As far as I understand one of the arguments of SI_IF_BREAK shall be full copy of exec mask saved before the loop, isn't it?
If that is not so, shall we fix it instead?
I think it is simpler than that. If a local symbol must be 64 dword aligned, it should be declared as a such and not 4 byte aligned as we have.
Such logic shall not be needed at the first place, llvm should be able to deduce proper alignment given proper input.
Fri, Feb 23
Wed, Feb 21
Tue, Feb 20
Feb 17 2018
Feb 16 2018
Changes look non-essential to me.
Feb 15 2018
Please upload a full context diff (-U999999).
Feb 14 2018
Please fix the tests, otherwise looks good.
Feb 13 2018
Feb 8 2018
I agree theoretically. Anyway, are there any regressions?
Feb 7 2018
Rebased and merged with rL324440.
Feb 6 2018
Feb 5 2018
Feb 3 2018
Feb 2 2018
Jan 29 2018
I think you need a test with lds combining which does not merge on VI and does merge on GFX9 due to m0 defs.
Jan 22 2018
Jan 16 2018
I believe the source of inconsistency is the original definition of ID_SYMBOLIC_LAST_, which does not point to a last value in the enum, but a last + 1. This can be changed in a separate patch.
Jan 15 2018
Jan 5 2018
Jan 2 2018
Check for encoding in positive test.
Added ID_SYMBOLIC_FIRST_GFX9_ to enum.
Updated per review comments.
Dec 28 2017
Dec 19 2017
Dec 12 2017
Dec 11 2017
Dec 8 2017
Dec 7 2017
There actually can be problem with folding the node if we patch it after creation. At least this needs to be checked.
Dec 6 2017