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This page is outdated!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 UID7fa824a6cbf8a30e
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 8289e0cf24346aa7:7fa824a6cbf8a30e
CompilerGCC 4.9
CPUQualcomm Technologies, Inc SDM636
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 da46ebaf6bd446c5 1.11 0.99 -O3 -fno-auto-inc-dec -fbranch-target-load-optimize2 -fno-dce -ffast-math -fgcse-after-reload -fno-gcse-sm -fno-hoist-adjacent-loads -fif-conversion -finline-small-functions -fno-ipa-cp-clone -fira-hoist-pressure -fno-isolate-erroneous-paths-dereference -fno-function-cse -fno-sched-interblock -fno-trapping-math -fprefetch-loop-arrays -fno-reorder-blocks -fsched-critical-path-heuristic -fstrict-aliasing -fstrict-overflow -fno-tracer -ftree-slsr -fno-tree-ter -fno-use-linker-plugin -falign-functions=0 -ffp-contract=off -fsched-stalled-insns=0 --param max-inline-insns-recursive-auto=128 --param partial-inlining-entry-probability=107 --param large-stack-frame-growth=1118 --param gcse-cost-distance-ratio=10 --param max-unrolled-insns=321 --param omega-max-vars=213 --param omega-max-eqs=143 --param max-sched-region-insns=144 --param max-sched-extend-regions-iters=0 --param max-last-value-rtl=4497 --param switch-conversion-max-branch-ratio=3 --param graphite-max-bbs-per-function=90 --param cxx-max-namespaces-for-diagnostic-help=745 --param tree-reassoc-width=0 --param max-slsr-cand-scan=774679 --param asan-memintrin=0 -O3 1 0 2 1 milepost-codelet-mibench-consumer-lame-src-fft-codelet-2-1 default 1113.6, 1113.6, 1113.6, 1113.6, 1113.6, 1113.6, 1113.6, 1113.6 1 XIAOMI REDMI NOTE 5 Android 9



[ 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