[PATCH][Review Requested][Compilation Time] Calculate hash value and equality comparison within SCEV node itself

Duncan Sands baldrick at free.fr
Tue Feb 5 07:38:51 PST 2013


Hi Sriram,

On 04/02/13 20:39, Murali, Sriram wrote:
> SCEV is currently implemented as a FoldingSetNode to calculate the hashing value
> and for equality comparison. This causes some performance loss. This patch aims
> at improving the compilation time performance by implementing the hashing and
> equality comparison within SCEV class instead of inheriting FoldingSetNode. The
> compilation performance gain for this particular refactoring is around 2%.

sorry to be a pain, but such numbers are meaningless without an estimate
of the variance.  In short, are the apparent differences statistically
significant, or might they just be random noise?

Ciao, Duncan.

>
> This patch is part of a series of compile time improvements. Although these were
> originally produced by our colleague Wan Xiaofei, our team consisting of
> tyler.nowicki at intel.com <mailto:tyler.nowicki at intel.com>; preston.gurd at intel.com
> <mailto:preston.gurd at intel.com>; and sriram.murali at intel.com
> <mailto:sriram.murali at intel.com>, are assuming all responsibility for this work.
>
> Please Review
>
> Thanks
>
> Sriram Murali
>
> *Benchmark***
>
> 	
>
> *Trunk***
>
> 	
>
> *CSEMap.SCEV***
>
> *401.bzip2***
>
> 	
>
> 74.21
>
> 	
>
> 72.69
>
> *403.gcc***
>
> 	
>
> 73.88
>
> 	
>
> 73.55
>
> *429.mcf***
>
> 	
>
> 72.80
>
> 	
>
> 71.41
>
> *433.milc***
>
> 	
>
> 78.78
>
> 	
>
> 77.31
>
> *444.namd***
>
> 	
>
> 94.73
>
> 	
>
> 93.87
>
> *445.gobmk***
>
> 	
>
> 36.28
>
> 	
>
> 35.98
>
> *450.soplex***
>
> 	
>
> 71.41
>
> 	
>
> 72.09
>
> *456.hmmer***
>
> 	
>
> 86.80
>
> 	
>
> 85.54
>
> *458.sjeng***
>
> 	
>
> 96.38
>
> 	
>
> 94.73
>
> *464.h264ref***
>
> 	
>
> 87.61
>
> 	
>
> 85.40
>
> *470.lbm***
>
> 	
>
> 68.95
>
> 	
>
> 66.36
>
> *471.omnetpp***
>
> 	
>
> 89.07
>
> 	
>
> 88.41
>
> *bitmnp01***
>
> 	
>
> 84.06
>
> 	
>
> 82.65
>
> *cjpegv2data6***
>
> 	
>
> 59.70
>
> 	
>
> 57.83
>
> *idctrn01***
>
> 	
>
> 40.18
>
> 	
>
> 38.99
>
> *libquake2***
>
> 	
>
> 48.48
>
> 	
>
> 48.02
>
> *libquake_portable***
>
> 	
>
> 63.54
>
> 	
>
> 62.94
>
> *libxcsoar***
>
> 	
>
> 47.44
>
> 	
>
> 47.21
>
> *linpack***
>
> 	
>
> 142.14
>
> 	
>
> 135.50
>
> *matrix01***
>
> 	
>
> 24.75
>
> 	
>
> 24.38
>
> *nbench***
>
> 	
>
> 108.04
>
> 	
>
> 103.82
>
> *tblook01***
>
> 	
>
> 44.03
>
> 	
>
> 42.37
>
> *ttsprk01***
>
> 	
>
> 39.23
>
> 	
>
> 38.30
>
> *Geomean***
>
> 	
>
> 65.85
>
> 	
>
> 64.60
>
> 		
>
> *401.bzip2***
>
> 	
>
> 100.00
>
> 	
>
> 102.09
>
> *403.gcc***
>
> 	
>
> 100.00
>
> 	
>
> 100.45
>
> *429.mcf***
>
> 	
>
> 100.00
>
> 	
>
> 101.95
>
> *433.milc***
>
> 	
>
> 100.00
>
> 	
>
> 101.90
>
> *444.namd***
>
> 	
>
> 100.00
>
> 	
>
> 100.92
>
> *445.gobmk***
>
> 	
>
> 100.00
>
> 	
>
> 100.83
>
> *450.soplex***
>
> 	
>
> 100.00
>
> 	
>
> 99.06
>
> *456.hmmer***
>
> 	
>
> 100.00
>
> 	
>
> 101.47
>
> *458.sjeng***
>
> 	
>
> 100.00
>
> 	
>
> 101.74
>
> *464.h264ref***
>
> 	
>
> 100.00
>
> 	
>
> 102.59
>
> *470.lbm***
>
> 	
>
> 100.00
>
> 	
>
> 103.90
>
> *471.omnetpp***
>
> 	
>
> 100.00
>
> 	
>
> 100.75
>
> *bitmnp01***
>
> 	
>
> 100.00
>
> 	
>
> 101.71
>
> *cjpegv2data6***
>
> 	
>
> 100.00
>
> 	
>
> 103.23
>
> *idctrn01***
>
> 	
>
> 100.00
>
> 	
>
> 103.05
>
> *libquake2***
>
> 	
>
> 100.00
>
> 	
>
> 100.96
>
> *libquake_portable***
>
> 	
>
> 100.00
>
> 	
>
> 100.95
>
> *libxcsoar***
>
> 	
>
> 100.00
>
> 	
>
> 100.49
>
> *linpack***
>
> 	
>
> 100.00
>
> 	
>
> 104.90
>
> *matrix01***
>
> 	
>
> 100.00
>
> 	
>
> 101.52
>
> *nbench***
>
> 	
>
> 100.00
>
> 	
>
> 104.06
>
> *tblook01***
>
> 	
>
> 100.00
>
> 	
>
> 103.92
>
> *ttsprk01***
>
> 	
>
> 100.00
>
> 	
>
> 102.43
>
> *Geomean***
>
> 	
>
> 100.00
>
> 	
>
> 101.94
>
> --
>
> Sriram Murali
>
> SSG/DPD/ECDL/DMP
>
> +1 (519) 772 – 2579
>
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>




More information about the llvm-commits mailing list