search for: floatmm

...% 5.3663 5.2169 0.0008 > MultiSource/Applications/lua/lua.exec -2.50% 28.8779 28.1550 0.0036 > External/Povray/povray.exec -1.73% 4.3798 4.3040 0.0271 > MultiSource/Applications/lambda-0_1_3/lambda.exec -1.56% 6.6249 6.5214 0.0019 > > Added Tests > SingleSource/Benchmarks/Stanford/FloatMM.compile > SingleSource/Benchmarks/Stanford/FloatMM.exec > > > > _______________________________________________ > llvm-testresults mailing list > llvm-testresults at cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-testresults

...t;> MultiSource/Applications/lua/lua.exec -2.50% 28.8779 28.1550 0.0036 >> External/Povray/povray.exec -1.73% 4.3798 4.3040 0.0271 >> MultiSource/Applications/lambda-0_1_3/lambda.exec -1.56% 6.6249 6.5214 0.0019 >> >> Added Tests >> SingleSource/Benchmarks/Stanford/FloatMM.compile >> SingleSource/Benchmarks/Stanford/FloatMM.exec >> >> >> >> _______________________________________________ >> llvm-testresults mailing list >> llvm-testresults at cs.uiuc.edu >> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-testresults &...

Hi all, I've started looking at the GlobalMerge pass, enabled by default on ARM and AArch64. I think we should reconsider that, at least for AArch64. As is, the pass just merges all globals together, in groups of 4KB (AArch64, 128B on ARM). At the time it was enabled, the general thinking was "it's almost free, it doesn't affect performance much, we might as well use it".

...rce/Benchmarks/Stanford/Bubblesort | * | * | SingleSource/Benchmarks/Stanford/Queens | * | * | SingleSource/Benchmarks/Stanford/Oscar | * | * | SingleSource/Benchmarks/Stanford/FloatMM | * | * | SingleSource/Benchmarks/Stanford/Puzzle | * | * | SingleSource/Benchmarks/Misc/flops-3 | * | * | SingleSource/Benchmarks/Misc/flops-5...

...|5.240930496| -0.02|30531918346| 0|5.239798764| 0|30531918345| 0| |SingleSource/Benchmarks/Stanford/Bubblesort.test | 441|0.015584076| 76543755|0.015530674| 0.34| 76543751| 0|0.015531506| 0.34| 76543751| 0| |SingleSource/Benchmarks/Stanford/FloatMM.test | 275|0.069482096| 549642410|0.069534833| -0.08| 549642406| 0|0.069590491| -0.16| 549642406| 0| |SingleSource/Benchmarks/Stanford/Oscar.test | 563|0.001015371| 7107593|0.000947407| 7.17| 7107590| 0|0.00...

...0002 > MultiSource/Benchmarks/Prolangs-C/assembler/assembler.compile > <http://llvm.org/perf/db_default/simple/nts/380/graph?pset.0=on&test.MultiSource%2FBenchmarks%2FProlangs-C%2Fassembler%2Fassembler.compile=on> > 1.51% 0.6343 0.6439 0.0011 > SingleSource/Benchmarks/Stanford/FloatMM.compile > <http://llvm.org/perf/db_default/simple/nts/380/graph?pset.0=on&test.SingleSource%2FBenchmarks%2FStanford%2FFloatMM.compile=on> > 1.51% 0.0463 0.0470 0.0002 > SingleSource/Benchmarks/BenchmarkGame/fannkuch.compile > <http://llvm.org/perf/db_default/simple/nts/380/...

Hi Stepan, Sorry for the delay. It's great that you are working on MergeFunctions as well and I agree, we should definitely try to combine our efforts to improve MergeFunctions. Just to give you some context, the pass (with the similar function merging patch) is already being used in a production setting. From my point of view, it would be better if we focus on improving its capability

...12 29027 0 0.01 29002 0 0.01 29002 fixoutput.ll 2 32311 0 0.01 32281 0 0.01 32281 fixpenal.ll 3 97288 0 0.02 97255 0 0.01 97255 fixups.ll 8 13196 0 0.01 13180 0 0.01 13180 flags.ll 31 51827 1 0.02 51596 * * * fldry.ll 13 32438 0 0.01 32418 0 0.01 32418 float.ll 11 117871 0 0.02 117854 0 0.02 117854 FloatMM.ll 6 12850 0 0.01 12831 0 0.01 12831 FloatPrecision.ll 2 2147 0 0.01 2138 0 0.01 2138 floatrep.ll 2 2588 0 0.01 2571 0 0.01 2571 flops-1.ll 1 10454 0 0.01 10439 0 0.01 10439 flops-2.ll 1 10782 0 0.01 10767 0 0.01 10767 flops-3.ll 1 10676 0 0.01 10661 0 0.01 10661 flops-4.ll 1 12150 0 0.01 12135 0 0...