aboutsummaryrefslogtreecommitdiff
path: root/src/benchmarks
Commit message (Collapse)AuthorAgeFilesLines
* fix fd.prepare()Florian Fischer2019-11-221-3/+4
| | | | always call provide of the linux sources to know where they were provided to
* keep not modified artifact data in cacheFlorian Fischer2019-11-213-7/+8
|
* don't use fixed sizes to aquire life dataFlorian Fischer2019-11-211-21/+46
|
* cleanup database dumps after redis benchmarkFlorian Fischer2019-11-131-0/+5
|
* use fixed sizes in loop benchmark and remove legacy source codeFlorian Fischer2019-11-131-51/+9
|
* don't always use jemalloc for redisFlorian Fischer2019-11-061-1/+1
|
* add tex tables to cfrac and espressoFlorian Fischer2019-10-302-0/+16
|
* add and use write_tex_table functionFlorian Fischer2019-10-306-3/+46
|
* introduce a global summary file extensionFlorian Fischer2019-10-303-10/+15
|
* use new dj_workload locationFlorian Fischer2019-10-231-2/+2
|
* fix location of dj_workload artifactFlorian Fischer2019-10-231-1/+1
|
* fix lld documentationFlorian Fischer2019-10-231-3/+2
|
* fix redis directoryFlorian Fischer2019-10-221-7/+8
|
* remove normalized plots from loop and t-test1Florian Fischer2019-10-212-17/+1
|
* run larson for 5 instead of 1 secondFlorian Fischer2019-10-211-1/+1
|
* fix y-label for fd plotsFlorian Fischer2019-10-211-1/+1
|
* add larson documentationFlorian Fischer2019-10-211-7/+31
|
* add lld and mysql documentationFlorian Fischer2019-10-212-14/+243
|
* improve analysis: add cycles portions and fix top10 percentagesFlorian Fischer2019-10-162-10/+21
|
* fix dj_traceFlorian Fischer2019-10-161-1/+1
|
* improve summary of the blowup testFlorian Fischer2019-10-161-2/+2
|
* increase workload size for bigger max speedupFlorian Fischer2019-10-162-4/+6
| | | | | | | | | | | with the current workload size of 100K allocations the parallel part (worker) takes on my development laptop 96% of the single threaded execution time resulting in a maximal speedup of 25. (Amdahl: 1/(1-0.96) ~= 25) Increasing the the workload to 1M allocations ups the parallel portion to ~99% which gives us a max speedup of ~295 enough for all our test machines. With the new workload the single threaded execution time is ~100s on my laptop and ~30s on one of the test servers which is tolerable in my opinion.
* use fixed xkcd color for ideal rssFlorian Fischer2019-10-161-1/+1
|
* don't store sysbench version twiceFlorian Fischer2019-10-161-3/+0
|
* add documentation for loop, cfrac, espressoFlorian Fischer2019-10-143-6/+75
|
* store mysql files persistent in build/benchmarks/mysqlFlorian Fischer2019-10-131-23/+13
|
* fix redisFlorian Fischer2019-10-131-2/+3
| | | | | don't symlink redis binaries if they already exist. Make shutdown_cmd work atexit by using its abspath.
* don't prepare fd if it was already preparedFlorian Fischer2019-09-231-0/+3
|
* use ArchiveArtifacts for all benchmark resourcesFlorian Fischer2019-09-235-167/+86
|
* add artifactsFlorian Fischer2019-09-211-9/+7
| | | | | | | | | Artifacts are downloadable external ressources. They provide their ressource (vcs commit, extracted archive) using Artifact.provide. GitArtifact download a bare git repository and provides specific commits using worktrees. Use GitArtifact for the linux kernel source
* add fd benchmarkFlorian Fischer2019-09-191-0/+104
| | | | | benchmark inspired by this blog post. https://dev.to/sharkdp/an-unexpected-performance-regression-11ai
* add blowup testFlorian Fischer2019-09-193-0/+177
| | | | | Blowup tests the described blowup problem of ptmalloc described in the TCMalloc post.
* add raxml-ng benchmarkFlorian Fischer2019-09-141-0/+120
|
* fix barplots for Ideal_RSSFlorian Fischer2019-09-141-1/+2
|
* fix mysld server after server changeFlorian Fischer2019-09-141-1/+1
|
* don't exit if we fail to prepare a benchmarkFlorian Fischer2019-09-113-17/+5
| | | | | benchmark.prepare should raise an Exception if it encounters an unrecoverable, intolerable error.
* explicitly collect binary versionsFlorian Fischer2019-09-114-0/+18
| | | | | | | There is no generic way to retrieve the version of a binary. Not everyone followes GNU cli guidelines and supports "--version". The larson benchmark for example reads input from stdin when started with "larson --version" blocking the Benchmark.prepare() call.
* port building of redis to pythonFlorian Fischer2019-09-112-20/+52
|
* add redis benchmark inspired by mimalloc-benchFlorian Fischer2019-09-113-0/+114
| | | | | | redis is build using a bash script because I wasn't able to build with make alone. TODO: build redis using only a Makefile
* improve server benchmarksFlorian Fischer2019-09-112-3/+7
| | | | | | | servers are now dictionary with two mandatory keys: "name" and "cmd". Optional keys are: * "prepare_cmds": commands are run after a server is started * "shutdown_cmds": commands are run before a server is terminated
* add "time" suffix to realloc time plotFlorian Fischer2019-09-041-1/+2
|
* improve benchmark code qualityFlorian Fischer2019-08-291-1/+1
|
* move urlretrieve report hook to util.py and use archive in dj_traceFlorian Fischer2019-08-292-60/+22
|
* require a name in Benchmark.__init__Florian Fischer2019-08-2911-26/+23
|
* readd missing numpy import for mysql summaryFlorian Fischer2019-08-281-0/+2
|
* revert accidentally added changesFlorian Fischer2019-08-281-2/+2
|
* add license header comments to benchmark definitionsFlorian Fischer2019-08-2812-0/+206
|
* fix IDEAL_RSS_RE regexFlorian Fischer2019-08-271-1/+1
|
* remove verbosity argument from hooksFlorian Fischer2019-08-275-6/+11
|
* improve benchmark code quality using pylintFlorian Fischer2019-08-2712-160/+190
| | | | The description Benchmark member is replaced with class docstrings
generated by cgit v1.2.3 (git 2.25.1) at 2026-04-25 13:48:03 +0000