[PATCH] [AArch64] Enable partial unrolling and runtime unrolling for AArch64 target

[Kevin Qin]

Hi Eric,

I list the top 5 performance improvements and regressions along with the
code size changes shown in bracket on SPEC2000 & SPEC2006. They are:


SPEC2006

Top 5 performance improvements
403_gcc    95.86% (107.71%)
482_sphinx3 96.36% (139.90%)
433_milc   96.43% (120.59%)
456_hmmer 97.45% (140.49%)
470_lbm    97.76% (108.70%)


Top 5 regressions
400_perlbench 102.54% (108.20%)
401_bzip2 101.71% (126.37%)
445_gobmk 101.63% (103.01%)
458_sjeng 100.96% (110.19%)
450_soplex 100.67% (135.32%)

geomean changes 99.15% (119.76%)

SPEC2000

Top 5 improvements
179_art     92.89% (158.06%)
175_vpr    97.90% (123.87%)
183_equake 98.28% (112.50%)
197_parser 98.59% (115.23%)
254_gap   98.66% (128.01%)

Top 5 regressions
177_mesa 103.54% (151.85%)
253_perlbmk 101.15% (110.19%)
164_gzip   100.86% (128.13%)
256_bzip2 100.62% (135.00%)
255_vortex 100.35% (100.72%)

geomean changes 99.41% (122.34%)

Thanks,
Kevin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140909/28d2810d/attachment.html>