P4 to Git Change 1075489 by marsenau@marsenau_ma on 2014/09/10 15:05:23

EPR #405194 - Change unroll threshold to LLVM default to partially work around Linpack performance problem.

	Prior to CL 1058428, which increased the unroll threshold to 200, this was only 100 which is lower than the LLVM default. Linpack's new ISA has increased register usage, but decreasing the unroll threshold to the previous level does not reduce the register count to its previous level. The increased register usage is probably a new SC problem, so this should probably be increased again in the future. There is no change in register usage with 100 vs. 150 on Linpack.

Affected files ...

... //depot/stg/opencl/drivers/opencl/compiler/lib/utils/OPTIONS.def#113 edit


[ROCm/clr commit: d5f7502ca8]

projects/clr/rocclr/compiler/lib/utils/OPTIONS.def

@@ -560,12 +560,12 @@ OPTION(OT_UINT32, \
        "Set loop unroll count to <number> for all loops")
 
 // -unroll-threshold=<positive integer> : loop unroll threshold
-// default is 200.
+// default is 150.
 OPTION(OT_UINT32, \
        OA_RUNTIME|OVIS_SUPPORT|OVA_REQUIRED|OA_SEPARATOR_EQUAL, \
        "unroll-threshold", NULL, \
        LUThreshold, \
-       200, 0, 0x3FFFFFFF, NULL, \
+       150, 0, 0x3FFFFFFF, NULL, \
        "Set loop unroll threshold to <number> for all loops")
 
 // -unroll-allow-partial : allow partial unroll