[PATCH] D29473: [AMDGPU] Unroll preferences improvements

Tom Stellard via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Fri Feb 3 08:33:45 PST 2017

tstellarAMD added inline comments.

Comment at: llvm/trunk/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp:88
         // programs way too big.
-        UP.Threshold = 800;
+        UP.Threshold = UnrollThresholdPrivate;
+        return;
vpykhtin wrote:
> tstellarAMD wrote:
> > Do you also want to set PartialThreshold here?
> I thought partialy unrolled loops won't make it possible to SROA private arrays. What are the benefits of partial unrolling on AMDGPU btw? What comes in mind: mem ops clustering/widening, less branches? What else?
I had a test case where bumping the PartialThreshold helped more non-partial loops be unrolled, but I looked at the case again and increasing the normal Threshold has the same affect, so I don't think this is needed.



More information about the llvm-commits mailing list