Collective Knowledge Aggregator proof-of-concept
Crowd results Raw CK browser Graphs Reports Datasets Models Home

This page is outdated! New version is available here.


Distinct solutions after online classification (auto/crowd-tune GCC compiler flags (minimize execution time))

Scenario UID8289e0cf24346aa7 (experiment.tune.compiler.flags.gcc.e)
Data UID04da0bfe4e5dd961
Discuss (optimizations to improve compilers,
semantic/data set/hardware features
to improve predictions
, etc):
GitHub wiki, Google group
Download:[ All solutions in JSON ], [ Solutions' classification in JSON ]
Reproduce all (with reactions):ck replay 8289e0cf24346aa7:04da0bfe4e5dd961
CompilerGCC 4.9
CPUIntel(R) Atom(TM) CPU Z3580 @ 1.33GHz
Objectivemin
Improvement key IK1Main kernel execution time speedup [min]
Improvement key IK2Code size improvement

Improvements (<4% variation) Distinct workload for highest improvement
# Solution UID IK1 IK2 New distinct optimization choices Ref Best species Worst species Touched Iters Program CMD Dataset Dataset file CPU freq (MHz) Cores Platform OS Replay
S1 9e9a676796c48dab 2.30 0.38 -O3 -fno-auto-inc-dec -fno-delayed-branch -fdelete-null-pointer-checks -fno-float-store -fno-gcse -finline-functions -flive-range-shrinkage -fno-loop-block -flto -fno-branch-count-reg -fno-trapping-math -freschedule-modulo-scheduled-loops -fno-sched-spec-load -fno-sched-spec-load-dangerous -fsched-group-heuristic -fsched-last-insn-heuristic -fno-schedule-insns -fsel-sched-pipelining -ftree-copyrename -ftree-dominator-opts -ftree-dse -fno-tree-loop-if-convert-stores -funswitch-loops -ffp-contract=on --param max-inline-insns-single=573 --param partial-inlining-entry-probability=5 --param min-vect-loop-bound=1 --param max-pending-list-length=14 --param gcse-after-reload-partial-fraction=3 --param unlikely-bb-count-fraction=7225 --param builtin-expect-probability=10 --param scev-max-expr-size=66 --param vect-max-version-for-alias-checks=0 --param integer-share-limit=160 --param l1-cache-size=80 --param sccvn-max-scc-size=9890 --param lra-max-considered-reload-pseudos=432 --param loop-max-datarefs-for-datadeps=931 --param ipa-cp-value-list-size=13 --param lto-min-partition=1300 -O3 1 0 2 1 milepost-codelet-mibench-office-rsynth-src-nsynth-codelet-5-1 default image-tiff-0001-nocomp data.tiff 1 ASUS Z00A Android 5.0
S2 5d9130ca164989be 1.66 1.00 -O3 -fno-auto-inc-dec -fno-devirtualize -fno-ipa-pure-const -fno-isolate-erroneous-paths-attribute -fno-keep-static-consts -fno-trapping-math -fomit-frame-pointer -fno-prefetch-loop-arrays -fno-sched-spec-load-dangerous -fno-sched-last-insn-heuristic -fno-split-wide-types -fno-tree-ch -ftree-coalesce-vars -ftree-copyrename -ftree-reassoc -fno-tree-ter --param max-early-inliner-iterations=2 --param max-hoist-depth=5 --param max-iterations-computation-cost=6 --param builtin-expect-probability=92 --param iv-consider-all-candidates-bound=16 --param omega-max-keys=355 --param vect-max-peeling-for-alignment=34 --param ipa-cp-loop-hint-bonus=114 --param max-stores-to-sink=4 --param asan-memintrin=0 -O3 1 0 2 1 cbench-security-rijndael decode enc-0001 data.enc 1 ASUS Z00A Android 5.0
S3 49c09c6f42c047a6 1.42 1.00 -O3 -fcrossjumping -fgcse-las -foptimize-sibling-calls -fno-peel-loops -fpredictive-commoning -fno-prefetch-loop-arrays -freorder-functions -fsched-pressure -fno-strict-aliasing -fno-tree-bit-ccp -ftree-ccp -fno-tree-dominator-opts -ftree-fre -fno-tree-switch-conversion -fno-vpt -fuse-linker-plugin -falign-loops=0 --param partial-inlining-entry-probability=20 --param gcse-after-reload-partial-fraction=3 --param gcse-after-reload-critical-fraction=20 --param max-iterations-to-track=972 --param max-iterations-computation-cost=16 --param sms-loop-average-count-threshold=0 --param tracer-max-code-growth=123 --param scev-max-expr-size=103 --param omega-max-wild-cards=28 --param vect-max-peeling-for-alignment=27 --param max-sched-region-insns=185 --param min-size-for-stack-sharing=18 --param max-dse-active-local-stores=9162 --param ira-max-conflict-table-size=162 --param ipa-sra-ptr-growth-factor=1 --param ipa-cp-eval-threshold=630 --param lto-min-partition=1490 --param asan-stack=0 --param asan-globals=1 -O3 1 0 2 1 milepost-codelet-mibench-telecomm-gsm-src-short-term-codelet-2-1 default cdataset-patricia-0001 data.txt 1 ASUS Z00A Android 5.0
S4 4da6a122fd724439 1.40 1.02 -O3 -fcombine-stack-adjustments -fno-cprop-registers -fipa-sra -fno-float-store -fno-guess-branch-probability -fno-prefetch-loop-arrays -fno-rounding-math -fno-selective-scheduling -fno-tree-bit-ccp -fno-tree-coalesce-vars -fno-tree-forwprop -ftree-loop-im -fno-tree-switch-conversion -falign-functions=0 --param max-inline-insns-single=323 --param max-inline-insns-recursive=172 --param partial-inlining-entry-probability=96 --param inline-unit-growth=19 --param max-gcse-insertion-ratio=20 --param max-completely-peel-loop-nest-depth=15 --param sms-max-ii-factor=185 --param max-goto-duplication-insns=16 --param max-cse-path-length=18 --param vect-max-version-for-alignment-checks=5 --param max-sched-region-blocks=3 --param max-fields-for-field-sensitive=0 --param prefetch-latency=238 --param sccvn-max-alias-queries-per-access=1938 --param tm-max-aggregate-size=6 --param ipa-cp-loop-hint-bonus=121 --param cxx-max-namespaces-for-diagnostic-help=371 --param sched-pressure-algorithm=1 -O3 1 0 2 1 cbench-telecom-crc32 default image-jpeg-fgg photo.jpg 1 ASUS Z00A Android 5.0
S5 0a0a4377be1a842a 1.38 1.07 -O3 -fno-associative-math -fauto-inc-dec -fdelayed-branch -fgcse-las -finline-functions -fmodulo-sched -fno-math-errno -fno-zero-initialized-in-bss -fpredictive-commoning -frounding-math -fsched-dep-count-heuristic -fsignaling-nans -fno-tree-bit-ccp -ftree-loop-distribute-patterns -ftree-vrp -fno-unsafe-loop-optimizations -fwhole-program -fira-algorithm=CB --param predictable-branch-outcome=20 --param max-inline-recursive-depth=15 --param large-function-growth=121 --param max-unroll-times=7 --param tracer-min-branch-ratio=77 --param tracer-min-branch-probability=80 --param min-crossjump-insns=5 --param omega-max-wild-cards=0 --param omega-hash-table-size=339 --param max-reload-search-insns=172 --param lra-max-considered-reload-pseudos=8 --param slp-max-insns-in-bb=462 --param ipa-cp-loop-hint-bonus=74 --param max-stores-to-sink=0 -O3 1 0 2 1 cbench-network-dijkstra default cdataset-dijkstra-0001 data.txt 1 ASUS Z00A Android 5.0
S6 3b6b4683b4f4c52c 1.35 0.78 -O3 -fno-branch-target-load-optimize2 -fdse -floop-strip-mine -fsched-spec-load -fsched-rank-heuristic -ftree-bit-ccp -fno-tree-copy-prop -ftree-loop-if-convert -ftree-sink -ftree-vectorize -funroll-loops -finline-limit=0 -fira-region=one -fsched-stalled-insns-dep=0 --param min-inline-recursive-probability=8 --param max-delay-slot-live-search=304 --param max-pending-list-length=25 --param large-unit-insns=13007 --param early-inlining-insns=16 --param large-stack-frame=250 --param large-stack-frame-growth=372 --param gcse-after-reload-critical-fraction=11 --param gcse-unrestricted-cost=4 --param max-peeled-insns=144 --param iv-max-considered-uses=443 --param max-dse-active-local-stores=6514 --param prefetch-latency=129 --param use-canonical-types=0 --param sccvn-max-scc-size=5329 --param ira-max-conflict-table-size=1334 --param min-insn-to-prefetch-ratio=14 --param max-tail-merge-iterations=2 --param sched-pressure-algorithm=2 --param max-slsr-cand-scan=289512 -O3 1 0 2 1 milepost-codelet-mibench-security-pgp-d-src-mpilib-codelet-1-1 default SGEMM_NN SGEMM_NN_1x1.json 1 ASUS Z00A Android 5.0
S7 2a404aa49a961096 1.27 0.67 -O3 -fbtr-bb-exclusive -fdata-sections -fno-delayed-branch -fno-dse -fno-if-conversion2 -fno-defer-pop -fno-math-errno -fno-toplevel-reorder -fno-reorder-blocks -fsched-spec-load-dangerous -fschedule-insns -fno-tree-builtin-call-dce -fno-tree-forwprop -ftree-fre -fno-tree-loop-vectorize -ftree-pre -ftree-slsr -ftree-ter -funroll-loops -funsafe-math-optimizations -fexcess-precision=standard -finline-limit=0 --param max-inline-insns-single=465 --param min-inline-recursive-probability=11 --param large-unit-insns=9252 --param builtin-expect-probability=29 --param vect-max-version-for-alias-checks=7 --param max-sched-insn-conflict-delay=4 --param max-partial-antic-length=106 --param max-vartrack-expr-depth=11 --param tm-max-aggregate-size=2 --param ipa-cp-loop-hint-bonus=93 --param sched-pressure-algorithm=1 -O3 1 0 2 1 cbench-security-sha default bzip2-0001 data.bz2 1 ASUS Z00A Android 5.0
S8 8c8014fd6964f75b 1.24 1.00 -O3 -fauto-inc-dec -fcse-follow-jumps -fno-devirtualize -fno-expensive-optimizations -ffinite-math-only -fgcse-sm -fno-inline-functions -fira-loop-pressure -fno-ira-share-save-slots -fira-share-spill-slots -fno-peephole -fno-zero-initialized-in-bss -freorder-blocks -fsched-dep-count-heuristic -fno-sel-sched-pipelining-outer-loops -fno-strict-aliasing -ftracer -fno-tree-fre -ftree-loop-if-convert-stores -fno-tree-vrp -funsafe-math-optimizations -fweb -fuse-linker-plugin -ffp-contract=off --param large-stack-frame-growth=1798 --param max-iterations-to-track=1554 --param vect-max-version-for-alignment-checks=0 --param sink-frequency-threshold=63 --param sched-state-edge-prob-cutoff=80 --param max-jump-thread-duplication-stmts=11 --param lra-max-considered-reload-pseudos=266 --param max-vartrack-size=57797540 --param max-tracked-strlens=468 -O3 1 0 2 1 cbench-automotive-susan corners image-pgm-clean-gray-square-600-450-8 data.pgm 1 ASUS Z00A Android 5.0
S9 ead54774209bd841 1.19 1.06 -O3 -fno-btr-bb-exclusive -fno-delayed-branch -fdelete-null-pointer-checks -fno-ipa-reference -fira-share-save-slots -fno-keep-inline-functions -flive-range-shrinkage -fno-merge-constants -fmodulo-sched -fno-move-loop-invariants -fguess-branch-probability -fno-peephole2 -fno-rerun-cse-after-loop -fno-sched-spec-insn-heuristic -ftree-ch -fno-tree-dominator-opts -fno-tree-loop-if-convert-stores -falign-functions=0 -fsched-stalled-insns-dep=0 --param inline-min-speedup=19 --param max-inline-insns-single=180 --param max-inline-insns-recursive=158 --param max-inline-insns-recursive-auto=740 --param max-inline-recursive-depth=9 --param max-delay-slot-live-search=36 --param max-hoist-depth=44 --param max-average-unrolled-insns=44 --param max-peeled-insns=17 --param max-unswitch-insns=3 --param vect-max-version-for-alignment-checks=0 --param vect-max-peeling-for-alignment=11 --param max-sched-region-insns=8 --param selsched-max-sched-times=1 --param selsched-insns-to-rename=4 --param max-last-value-rtl=8594 --param prefetch-latency=315 --param tm-max-aggregate-size=15 --param max-tail-merge-comparisons=2 -O3 1 0 2 1 milepost-codelet-mibench-office-rsynth-src-nsynth-codelet-9-1 default image-ppm-0001 data.ppm 1 ASUS Z00A Android 5.0
S10 d0e0694b9a925a76 1.17 0.86 -O3 -fno-branch-target-load-optimize2 -fcx-fortran-rules -fno-float-store -fgcse-after-reload -fno-keep-inline-functions -fmerge-all-constants -fmath-errno -ftrapping-math -fno-sched-pressure -fsched-spec-load -ftree-coalesce-vars -fno-tree-dse -ftree-slsr -fno-unit-at-a-time -funroll-loops -fno-unsafe-math-optimizations -fno-wpa -fira-region=mixed -ftree-parallelize-loops=0 --param max-inline-insns-recursive=428 --param min-inline-recursive-probability=5 --param max-modulo-backtrack-attempts=46 --param large-function-growth=57 --param large-stack-frame-growth=1156 --param gcse-unrestricted-cost=5 --param max-peeled-insns=136 --param max-completely-peel-loop-nest-depth=1 --param max-unswitch-insns=14 --param sms-max-ii-factor=130 --param max-cse-path-length=14 --param omega-max-vars=127 --param max-pipeline-region-insns=274 --param sched-mem-true-dep-cost=2 --param integer-share-limit=424 --param prefetch-latency=273 --param max-partial-antic-length=144 --param allow-packed-load-data-races=0 --param max-tail-merge-comparisons=18 -O3 1 0 2 1 milepost-codelet-mibench-consumer-lame-src-newmdct-codelet-10-1 default audio-mp3-0001 data.mp3 1 ASUS Z00A Android 5.0
S11 a442cc94478ba9f2 1.14 1.03 -O3 -fno-compare-elim -fno-fast-math -fno-gcse-las -findirect-inlining -fno-ira-hoist-pressure -fno-loop-block -floop-parallelize-all -fno-merge-constants -fmodulo-sched -fsigned-zeros -fno-optimize-sibling-calls -fno-ree -frerun-cse-after-loop -fsplit-wide-types -ftree-tail-merge --param max-inline-recursive-depth=9 --param sms-min-sc=3 --param max-predicted-iterations=142 --param scev-max-expr-complexity=17 --param omega-hash-table-size=611 --param vect-max-peeling-for-alignment=28 --param max-dse-active-local-stores=5602 --param switch-conversion-max-branch-ratio=7 --param min-insn-to-prefetch-ratio=0 --param lto-partitions=51 --param lto-min-partition=930 --param allow-load-data-races=1 --param allow-store-data-races=1 -O3 1 0 2 1 cbench-bzip2 encode image-tiff-0001 data.tiff 1 ASUS Z00A Android 5.0



[ Participated users, platforms, OS, CPU, GPU, GPGPU, NN, NPU ] [ How to participate ] [ Motivation (PPT) (PDF) ] [ Papers 1 , 2 , 3] [ Android app ] [ Collective training set ] [ Unified AI ]
View entry in raw format

Developed by Grigori Fursin           
Implemented as a CK workflow
                         
   
                      Hosted at