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This page is outdated!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 UIDb13714fc4afe98d1
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:b13714fc4afe98d1
CompilerGCC 4.9
CPUSamsung EXYNOS5420
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 96e439b9a92621b0 1.16 1.00 -O3 -fbranch-target-load-optimize -fbranch-target-load-optimize2 -fcx-fortran-rules -fno-fast-math -fno-finite-math-only -fgcse-sm -fno-ivopts -fmodulo-sched -fmodulo-sched-allow-regmoves -fmove-loop-invariants -fno-sched-spec -fno-sched-last-insn-heuristic -fsel-sched-pipelining -fno-tree-reassoc -fno-variable-expansion-in-unroller -falign-labels=0 --param max-inline-insns-single=15 --param max-variable-expansions-in-unroller=1 --param max-delay-slot-insn-search=70 --param gcse-after-reload-partial-fraction=5 --param gcse-unrestricted-cost=1 --param max-peeled-insns=186 --param max-completely-peeled-insns=96 --param max-grow-copy-bb-insns=8 --param scev-max-expr-complexity=6 --param omega-eliminate-redundant-constraints=1 --param sink-frequency-threshold=8 --param max-pipeline-region-insns=61 --param sched-mem-true-dep-cost=2 --param ira-loop-reserved-regs=4 --param max-vartrack-expr-depth=3 --param lto-min-partition=1589 --param max-stores-to-sink=2 --param tree-reassoc-width=0 -O3 1 0 2 1 milepost-codelet-mibench-consumer-mad-src-layer3-codelet-5-1 default pcm-0001 data.pcm 1900, 1900, 1900, 1900 1 SAMSUNG SM-T900 Android 5.1.1

[ 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