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This page is outdated!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 UID89eb1bb72ca3a5e6
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 8289e0cf24346aa7:89eb1bb72ca3a5e6
CompilerGCC 4.9
CPUQualcomm Technologies, Inc MSM8992
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 b69dc3fc4135a1ef 1.35 1.06 -O3 -fno-btr-bb-exclusive -fconserve-stack -fno-cx-limited-range -fno-devirtualize-speculatively -fno-dse -fno-inline-small-functions -fno-ipa-cp -fno-ipa-reference -ffunction-cse -fno-sched-interblock -fno-sched-spec -fno-rerun-cse-after-loop -fsched2-use-superblocks -fno-single-precision-constant -ftree-copyrename -ftree-loop-optimize -ftree-loop-vectorize -fno-tree-pre -fno-tree-partial-pre -fno-tree-sra -ftree-tail-merge -fno-unit-at-a-time -fvpt --param max-delay-slot-live-search=73 --param inline-unit-growth=25 --param early-inlining-insns=3 --param gcse-after-reload-critical-fraction=2 --param max-completely-peel-loop-nest-depth=10 --param max-crossjump-edges=2 --param min-crossjump-insns=10 --param omega-max-geqs=89 --param omega-max-keys=344 --param vect-max-version-for-alignment-checks=4 --param max-cselib-memory-locations=909 --param max-jump-thread-duplication-stmts=30 --param l1-cache-size=17 --param graphite-max-nb-scop-params=16 --param max-vartrack-expr-depth=24 --param allow-store-data-races=1 --param asan-instrument-writes=0 -O3 1 0 2 1 cbench-telecom-crc32 default number-0001 633.6, 633.6, 633.6, 633.6, 633.6, 633.6 1 LGE NEXUS 5X Android 6.0.1

[ 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