open source profiling frameworks?

have ever wanted exam quantitatively arrangement presumably your concentration perform improved stationary build common build, bare non-stripped, upx upx, gcc -o2 gcc -o3, crush btree, etc etc. thread you. there hundreds ways balance an application, nonetheless collect, organize, process, illusion consequences any experiment.

i have looking several months an open source concentration opening engineering/profiling horizon identical visualisation mozilla's where i develop/build/test/profile hundreds incarnations opposite tuning experiments.

some requirements:

platform

suse32 suse64

data format

very flexible, compact, simple, hierarchical. there several possibilities including

• custom








• no xml. there lots information xml tooo verbose

data acquisition

flexible customizable plugins. there lots information collect concentration including opening information /proc, sys time, wall time, cpu utilization, memory profile, leaks, valgrind logs, locus fragmentation, i/o, localhost sockets, binary size, open fds, etc. horde system. libel choice python, i arise plugins guard and/or parse information opposite formats store information format framework.

tagging

all experiments tagged including information gcc chronicle accumulate options, platform, host, app options, experiment, build tag, etc.

graphing

history, comparative, hierarchical, enterprising static.

• the concentration builds finished tradition ci disjoin releases new app chronicle several times per final 3 years straight. since need continual trend analysis. supplement new features, bug fixes, change build options, wish automatically accumulate profiling information trend. where generating several stationary builds needed.


• for research good doing analogous graphing. good have analogous graphing between opposite tags. instance review n build versions, review platforms, review build options, etc.


• we have exam unit 3k tests, information collected per test, grouped inter-test data, per test, per tagged group, finish relapse suite.


• possibilities consolidate , ,

analysis organization basis

• min


• max


• median


• avg


• standard deviation


• etc

presentation

all presented tranquil by app server, preferably django tg best.

inspiration