Mercurial > hg > graal-jvmci-8
comparison src/cpu/sparc/vm/c2_globals_sparc.hpp @ 1365:6476042f815c
6940701: Don't align loops in stubs for Niagara sparc
Summary: Don't align loops in stubs for Niagara sparc since NOPs are expensive.
Reviewed-by: twisti, never
author | kvn |
---|---|
date | Wed, 07 Apr 2010 09:37:47 -0700 |
parents | 473cce303f13 |
children | c18cbe5936b8 |
comparison
equal
deleted
inserted
replaced
1364:0dc88ad3244e | 1365:6476042f815c |
---|---|
58 define_pd_global(intx, FLOATPRESSURE, 52); // C2 on V9 gets to use all the float/double registers | 58 define_pd_global(intx, FLOATPRESSURE, 52); // C2 on V9 gets to use all the float/double registers |
59 define_pd_global(intx, FreqInlineSize, 175); | 59 define_pd_global(intx, FreqInlineSize, 175); |
60 define_pd_global(intx, INTPRESSURE, 48); // large register set | 60 define_pd_global(intx, INTPRESSURE, 48); // large register set |
61 define_pd_global(intx, InteriorEntryAlignment, 16); // = CodeEntryAlignment | 61 define_pd_global(intx, InteriorEntryAlignment, 16); // = CodeEntryAlignment |
62 define_pd_global(intx, NewSizeThreadIncrease, ScaleForWordSize(4*K)); | 62 define_pd_global(intx, NewSizeThreadIncrease, ScaleForWordSize(4*K)); |
63 // The default setting 16/16 seems to work best. | |
64 // (For _228_jack 16/16 is 2% better than 4/4, 16/4, 32/32, 32/16, or 16/32.) | |
65 define_pd_global(intx, OptoLoopAlignment, 16); // = 4*wordSize | |
66 define_pd_global(intx, RegisterCostAreaRatio, 12000); | 63 define_pd_global(intx, RegisterCostAreaRatio, 12000); |
67 define_pd_global(bool, UseTLAB, true); | 64 define_pd_global(bool, UseTLAB, true); |
68 define_pd_global(bool, ResizeTLAB, true); | 65 define_pd_global(bool, ResizeTLAB, true); |
69 define_pd_global(intx, LoopUnrollLimit, 60); // Design center runs on 1.3.1 | 66 define_pd_global(intx, LoopUnrollLimit, 60); // Design center runs on 1.3.1 |
70 | 67 |