search for: geomean

...() { // rest of the code in bar } Here are the numbers on a Power8 PPCLE running Ubuntu 15.04 in ST-mode ---------------------------------------------- Runtime performance (speed) ---------------------------------------------- Workload Improvement -------- ----------- SPEC2006(C/C++) 0.06% (geomean) SPEC2017(C/C++) 0.10% (geomean) ---------------------------------------------- Compile time performance for Bootstrapped LLVM ---------------------------------------------- Workload Improvement -------- ----------- SPEC2006(C/C++) 0.41% (cumulative) SPEC2017(C/C++) -0.16% (cumulative) lnt 0....

...instead of opening a new window. c. Next Steps: Split the results table into multiple sub-tables depending upon the suite (MultiSource/Applications, ...SingleSource/UnitTest) and make use of select-all checkbox and graph all the workloads in the corresponding suite. d. Next Steps: Add a Geomean button to plot the geomean of all the workloads instead of separate plots for each workload. If you feel that some of these improvements are valuable, then I can create patches and upload them in Phabricator for review. Thanks! With regards, Sriram -- Sriram Murali SSG/DPD/ECDL/DMP +1 (519) 772...

...by clang, as other front-ends could run into this even at -O0. Due to this assertion failure and the infrastructure I used, the numbers below do not include test-suite/MultiSource/Benchmarks results. On the non-correctness aspects, LNT tells me that: - The programs that report execution time, on geomean are about 17% slower. - The programs that report scores, on geomean are about 21% slower. - Code size is up on geomean about 11%. I'm afraid I don't have compile time numbers, nor any feel for debug info quality. I'll need quite a bit more time to dig into the details to come up with s...

...this there are several tunable parameters for the cost model. -- Performance -- More results including clang, llvm-tblgen, and more specific numbers about benefits/regressions can be found in the notes section below. * Size Reduction: - Test Suite(X86_64): - Early+Late outlining provides a geomean of 10.5% reduction over clang Oz, with a largest improvement of ~67% and largest regression of ~7.5%. - Late outlining provides a geomean of 4.65% reduction, with a largest improvement of ~51% and largest regression of ~6.4%. - Spec 2006(X86_64) - Early+Late outlining provides a geomean re...

...cost model. > > -- Performance -- > > More results including clang, llvm-tblgen, and more specific numbers about > benefits/regressions can be found in the notes section below. > > * Size Reduction: > > - Test Suite(X86_64): > > - Early+Late outlining provides a geomean of 10.5% reduction over clang > Oz, with a largest improvement of ~67% and largest regression of ~7.5%. > > - Late outlining provides a geomean of 4.65% reduction, with a largest > improvement of ~51% and largest regression of ~6.4%. > > - Spec 2006(X86_64) > > - Early...

...correctness issues (in ARM64) and lack of performance tuning (in AArch64) are addressed. However much more work has to be done to catch up with GCC 4.9 middle-end and backend optimizations. Benchmark ARM64 vs GCC 4.9 % ARM64 vs AArch64 % ARM64 vs AArch64 patched % EEMBC (no consumer) geomean -17 1 -2 EEMBC (consumer only) geomean -21 -2 -5 Linpack Double -29 45 -1 Linpack Single -51 40 1 SPEC2000 geomean -6 0 1 Thanks, Ana. -----Original Message----- From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at cs.uiuc.edu] On Behalf Of Tim North...

...9.22% -3.68% 256.bzip2 2263 2054 2241 10.18% 0.98% 54008 53922 58123 0.16% -7.08% 300.twolf 3706 3578 3388 3.58% 9.39% 223957 219748 234142 1.92% -4.35% ----------------------------------------------------------------------------------------------- geomean 3183 2794 2788 13.91% 14.15% 219228 225405 236945 -2.74% -7.48% Perf Size CFP2000 x32 32 64 x32/32 x32/64 x32 32 64 x32/32 x32/64 ===...

...> > tried (where the main reason to use a global is file- or function- > > static variables, only accessed through a single getter function). > > > > Across several runs on the entire test-suite, when disabling the pass, > > I measured: > > without LTO, a -0.19% geomean improvement with LTO, a +0.11% geomean > > regression. > > > > As for just SPEC2006, there are two big regressions: 400.perlbench > > (10.6% w/ LTO, 2.7% w/o) and 471.omnetpp (2.3% w/, 3.9% w/o). > > > > Numbers are attached. > > > > > > -- A wa...

...**c. **Next Steps: Split the results table into multiple sub-tables > depending upon the suite (MultiSource/Applications, …SingleSource/UnitTest) > and make use of select-all checkbox and graph all the workloads in the > corresponding suite.**** > > **d. **Next Steps: Add a Geomean button to plot the geomean of all > the workloads instead of separate plots for each workload.**** > > ** ** > > If you feel that some of these improvements are valuable, then I can > create patches and upload them in Phabricator for review.**** > > Thanks!**** > > **...

...unning Ubuntu 15.04 in ST-mode > > ---------------------------------------------- > Runtime performance (speed) > ---------------------------------------------- > Workload Improvement > -------- ----------- > SPEC2006(C/C++) 0.06% (geomean) > SPEC2017(C/C++) 0.10% (geomean) > ---------------------------------------------- > Compile time performance for Bootstrapped LLVM > ---------------------------------------------- > Workload Improvement > -------- ----------- > S...

...correctness issues (in ARM64) and lack of performance tuning (in AArch64) are addressed. However much more work has to be done to catch up with GCC 4.9 middle-end and backend optimizations. Benchmark ARM64 vs GCC 4.9 % ARM64 vs AArch64 % ARM64 vs AArch64 patched % EEMBC (no consumer) geomean -17 1 -2 EEMBC (consumer only) geomean -21 -2 -5 Linpack Double -29 45 -1 Linpack Single -51 40 1 SPEC2000 geomean -6 0 1 Thanks, Ana. -----Original Message----- From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at cs.uiuc.edu] On Behalf Of Tim North...

Hello A 4.4% geomean codesize improvement is really impressive. That stuff is hard to come by, you usually have to nibble away at it bit at a time. I ran some codesize benchmarks we have and they were in the same ballpark. Some of these are quite small so had less opportunity for outlining, but the average was still ov...

...reaker in my opinion. I'm very much in favor of switching in all configurations. To address some of the concerns, it may make sense to lower the threshold somewhat to minimize impact at this time (we won't have benefits as large at the time of the switch). I'm talking about getting the geomean closer to 1% in all configurations if possible. I believe that the regressions introduced by this flag flip can be undone by further using MemorySSA in the other passes currently using MemDepAnalysis, and offsetting the cost of computing MemorySSA in the first place. The threshold could be raised a...

....bzip2 13.9 13.6 2.21% 403.gcc 19.5 19.8 -1.52% 429.mcf 20.5 20.5 0.00% 445.gobmk 18.6 18.6 0.00% 456.hmmer 11.1 11.1 0.00% 458.sjeng 19.3 19.3 0.00% 462.libquantum 39.5 39.5 0.00% 464.h264ref 28.5 28.5 0.00% 471.omnetpp 15.6 15.6 0.00% 473.astar 13 13 0.00% 483.xalancbmk 21.9 21.9 0.00% GEOMEAN 19.0929865 19.00588287     0.46% 410.bwaves  15.2 15.2 0.00% 416.gamess CE CE #VALUE! 433.milc  19 18.6 2.15% 434.zeusmp    14.2 14.2 0.00% 435.gromacs       11.6 11.3 2.65% 436.cactusADM 8.31 7.89 5.32% 437.leslie3d 11 11 0.00% 444.namd   16 16 0.00% 447.dealII 25.4 25.4 0.00% 450.soplex...

...Here are the numbers on a Power8 PPCLE running Ubuntu 15.04 in ST-mode > > ---------------------------------------------- > Runtime performance (speed) > ---------------------------------------------- > Workload Improvement > -------- ----------- > SPEC2006(C/C++) 0.06% (geomean) > SPEC2017(C/C++) 0.10% (geomean) > ---------------------------------------------- > Compile time performance for Bootstrapped LLVM > ---------------------------------------------- > Workload Improvement > -------- ----------- > SPEC2006(C/C++) 0.41% (cumulative) >...

...uld run into this even > at -O0. > > Due to this assertion failure and the infrastructure I used, the numbers > below do not include test-suite/MultiSource/Benchmarks results. > > On the non-correctness aspects, LNT tells me that: > - The programs that report execution time, on geomean are about 17% slower. > - The programs that report scores, on geomean are about 21% slower. > - Code size is up on geomean about 11%. > I'm afraid I don't have compile time numbers, nor any feel for debug info > quality. > > I'll need quite a bit more time to dig into...

...SPASS 101.2% 120.0% > CTMark/consumer-typeset/consumer-typeset 105.7% 138.2% > CTMark/kimwitu++/kc 103.1% 122.6% > CTMark/lencod/lencod 106.2% 143.4% > CTMark/mafft/pairlocalalign 96.2% 135.4% > CTMark/sqlite3/sqlite3 109.1% 155.1% > CTMark/tramp3d-v4/tramp3d-v4 109.1% 132.0% > GEOMEAN 103.5% 128.0% > > > Thanks, > > Kristof Thanks, -Quentin -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170524/77fd12b2/attachment.html> -------------- next part -------------- A non-te...

...9.22% -3.68% 256.bzip2 2263 2054 2241 10.18% 0.98% 54008 53922 58123 0.16% -7.08% 300.twolf 3706 3578 3388 3.58% 9.39% 223957 219748 234142 1.92% -4.35% ----------------------------------------------------------------------------------------------- geomean 3183 2794 2788 13.91% 14.15% 219228 225405 236945 -2.74% -7.48% Perf Size CFP2000 x32 32 64 x32/32 x32/64 x32 32 64 x32/32 x32/64 ===...

...20.5 0.00% > 445.gobmk 18.6 18.6 0.00% > 456.hmmer 11.1 11.1 0.00% > 458.sjeng 19.3 19.3 0.00% > 462.libquantum 39.5 39.5 0.00% > 464.h264ref 28.5 28.5 0.00% > 471.omnetpp 15.6 15.6 0.00% > 473.astar 13 13 0.00% > 483.xalancbmk 21.9 21.9 0.00% > GEOMEAN 19.0929865 19.00588287 0.46% > 410.bwaves 15.2 15.2 0.00% > 416.gamess CE CE #VALUE! > 433.milc 19 18.6 2.15% > 434.zeusmp 14.2 14.2 0.00% > 435.gromacs 11.6 11.3 2.65% > 436.cactusADM 8.31 7.89 5.32% > 437.leslie3d 11 11 0.00% > 444.namd 16 16 0.00...

...a lot of macros and is a very C like program. Program Speedup 471.omnetpp 16.03% 473.astar 5.00% 483.xalancbmk 58.57% 444.namd -0.90% 447.dealII 60.47% 450.soplex 8.20% 453.povray 11.34% 252.eon 35.33% ------------------------- Geomean 21.01% 3.3 Statistics of LLVM profiles for SPEC2006 C/C++ programs We also collect some statistic of the profiles generated by FE based instrumentation and late instrumentation, namely, the following information: 1. the number of functions that being instrumented, 2. the result pro...