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This page is outdated!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 UIDf8879342fe761f2c
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:f8879342fe761f2c
CompilerGCC 4.9
CPUSamsung EXYNOS5433
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 40498dc26a016ed5 1.38 0.95 -O3 -fno-branch-probabilities -fno-compare-elim -fno-ipa-sra -fno-gcse-after-reload -fkeep-inline-functions -fno-lto -fno-peephole2 -ftoplevel-reorder -free -fno-thread-jumps -ftree-copy-prop -fno-tree-loop-optimize -ftree-loop-vectorize -fno-tree-reassoc -falign-functions=0 -falign-loops=0 -fexcess-precision=fast -ftree-parallelize-loops=0 --param max-inline-insns-single=455 --param max-unswitch-insns=22 --param sms-loop-average-count-threshold=0 --param hot-bb-frequency-fraction=569 --param iv-consider-all-candidates-bound=26 --param omega-max-vars=1 --param vect-max-peeling-for-alignment=51 --param max-sched-insn-conflict-delay=1 --param selsched-max-sched-times=1 --param max-fields-for-field-sensitive=0 --param max-sched-ready-insns=97 --param l1-cache-size=15 --param sccvn-max-alias-queries-per-access=1987 --param ira-max-loops-num=145 --param min-insn-to-prefetch-ratio=12 --param prefetch-min-insn-to-mem-ratio=4 --param allow-store-data-races=0 --param allow-packed-store-data-races=1 -O3 1 0 2 1 milepost-codelet-mibench-telecomm-gsm-src-short-term-codelet-2-1 default 1900, 1900, 1900, 1900, 1900, 1900, 1900, 1900 1 SAMSUNG SM-N910C Android 5.0.1
S2 d45dd2994de0a66a 1.15 1.00 -O3 -fno-branch-target-load-optimize2 -fno-fat-lto-objects -fgcse-las -fif-conversion -fno-ipa-pure-const -fno-partial-inlining -fshrink-wrap -fno-tree-ter -fno-tree-vrp -fvpt -fira-region=one --param max-early-inliner-iterations=0 --param large-unit-insns=7589 --param large-stack-frame=312 --param hot-bb-count-ws-permille=954 --param tracer-min-branch-probability-feedback=73 --param max-grow-copy-bb-insns=16 --param iv-consider-all-candidates-bound=17 --param vect-max-version-for-alignment-checks=8 --param sink-frequency-threshold=13 --param l1-cache-line-size=40 --param switch-conversion-max-branch-ratio=6 --param tree-reassoc-width=0 --param uninit-control-dep-attempts=703 -O3 1 0 2 1 milepost-codelet-mibench-security-pgp-e-src-mpilib-codelet-1-1 default 1900, 1900, 1900, 1900, 1900, 1900, 1900, 1900 1 SAMSUNG SM-N910C Android 5.0.1



[ 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