bmalloc: Bump the size of the deallocator log to 512
Created attachment 310186 [details] Patch
Benchmark results from Mac Pro: ~/OpenSource/Source/bmalloc> ~/OpenSource/PerformanceTests/MallocBench/run-malloc-benchmarks Baseline:~/OpenSource/WebKitBuildBaseline/Release/ Patch:~/OpenSource/WebKitBuild/Release/ Baseline Patch Δ Execution Time: churn 71ms 71ms list_allocate 68ms 68ms tree_allocate 65ms 65ms tree_churn 76ms 76ms fragment 61ms 61ms fragment_iterate 50ms 49ms ^ 1.02x faster medium 147ms 145ms ^ 1.01x faster big 122ms 124ms ! 1.02x slower facebook 182ms 180ms ^ 1.01x faster reddit 103ms 100ms ^ 1.03x faster flickr 106ms 104ms ^ 1.02x faster theverge 133ms 135ms ! 1.02x slower nimlang 118ms 117ms ^ 1.01x faster message_one 172ms 174ms ! 1.01x slower message_many 935ms 920ms ^ 1.02x faster churn --parallel 42ms 32ms ^ 1.31x faster list_allocate --parallel 145ms 143ms ^ 1.01x faster tree_allocate --parallel 751ms 770ms ! 1.03x slower tree_churn --parallel 917ms 904ms ^ 1.01x faster fragment --parallel 78ms 77ms ^ 1.01x faster fragment_iterate --parallel 12ms 11ms ^ 1.09x faster medium --parallel 131ms 127ms ^ 1.03x faster big --parallel 114ms 115ms ! 1.01x slower facebook --parallel 4,850ms 4,760ms ^ 1.02x faster reddit --parallel 3,472ms 3,495ms ! 1.01x slower flickr --parallel 3,497ms 3,460ms ^ 1.01x faster theverge --parallel 4,709ms 4,684ms ^ 1.01x faster <geometric mean> 198ms 195ms ^ 1.02x faster <arithmetic mean> 782ms 777ms ^ 1.01x faster <harmonic mean> 89ms 85ms ^ 1.05x faster Peak Memory: churn 1,032kB 1,020kB ^ 1.01x smaller list_allocate 2,312kB 2,312kB tree_allocate 5,672kB 5,672kB tree_churn 4,944kB 4,944kB fragment 7,212kB 7,224kB ! 1.0x bigger fragment_iterate 25,804kB 25,812kB ! 1.0x bigger medium 1,182,668kB 1,182,680kB ! 1.0x bigger big 1,083,860kB 1,083,872kB ! 1.0x bigger facebook 73,112kB 73,036kB ^ 1.0x smaller reddit 14,192kB 14,040kB ^ 1.01x smaller flickr 26,716kB 26,568kB ^ 1.01x smaller theverge 27,068kB 27,024kB ^ 1.0x smaller nimlang 173,944kB 173,804kB ^ 1.0x smaller message_one 4,288kB 4,448kB ! 1.04x bigger message_many 2,716kB 3,040kB ! 1.12x bigger churn --parallel 1,652kB 1,688kB ! 1.02x bigger list_allocate --parallel 2,972kB 3,076kB ! 1.03x bigger tree_allocate --parallel 12,704kB 13,396kB ! 1.05x bigger tree_churn --parallel 12,968kB 12,956kB ^ 1.0x smaller fragment --parallel 7,232kB 7,392kB ! 1.02x bigger fragment_iterate --parallel 26,452kB 26,264kB ^ 1.01x smaller medium --parallel 1,147,656kB 1,124,716kB ^ 1.02x smaller big --parallel 1,043,872kB 1,037,744kB ^ 1.01x smaller facebook --parallel 1,594,592kB 1,596,080kB ! 1.0x bigger reddit --parallel 289,932kB 292,100kB ! 1.01x bigger flickr --parallel 553,680kB 553,156kB ^ 1.0x smaller theverge --parallel 604,364kB 599,460kB ^ 1.01x smaller <geometric mean> 35,459kB 35,743kB ! 1.01x bigger <arithmetic mean> 293,838kB 292,723kB ^ 1.0x smaller <harmonic mean> 6,737kB 6,850kB ! 1.02x bigger Memory at End: churn 592kB 580kB ^ 1.02x smaller list_allocate 596kB 596kB tree_allocate 620kB 620kB tree_churn 616kB 616kB fragment 632kB 644kB ! 1.02x bigger fragment_iterate 792kB 800kB ! 1.01x bigger medium 12,188kB 12,200kB ! 1.0x bigger big 1,376kB 1,388kB ! 1.01x bigger facebook 2,972kB 2,880kB ^ 1.03x smaller reddit 2,028kB 1,876kB ^ 1.08x smaller flickr 3,032kB 2,884kB ^ 1.05x smaller theverge 2,984kB 2,940kB ^ 1.01x smaller nimlang 58,348kB 58,320kB ^ 1.0x smaller message_one 932kB 1,020kB ! 1.09x bigger message_many 1,320kB 1,740kB ! 1.32x bigger churn --parallel 1,216kB 1,252kB ! 1.03x bigger list_allocate --parallel 1,508kB 1,528kB ! 1.01x bigger tree_allocate --parallel 2,196kB 2,520kB ! 1.15x bigger tree_churn --parallel 2,780kB 2,836kB ! 1.02x bigger fragment --parallel 2,328kB 2,836kB ! 1.22x bigger fragment_iterate --parallel 1,668kB 1,808kB ! 1.08x bigger medium --parallel 31,124kB 38,864kB ! 1.25x bigger big --parallel 78,712kB 77,840kB ^ 1.01x smaller facebook --parallel 38,840kB 36,432kB ^ 1.07x smaller reddit --parallel 30,104kB 30,552kB ! 1.01x bigger flickr --parallel 33,588kB 35,976kB ! 1.07x bigger theverge --parallel 31,916kB 29,700kB ^ 1.07x smaller <geometric mean> 3,524kB 3,637kB ! 1.03x bigger <arithmetic mean> 12,778kB 13,009kB ! 1.02x bigger <harmonic mean> 1,580kB 1,621kB ! 1.03x bigger ===== ~/OpenSource/Source/bmalloc> ~/OpenSource/PerformanceTests/MallocBench/run-malloc-benchmarks --memory_warning Baseline:~/OpenSource/WebKitBuildBaseline/Release/ Patch:~/OpenSource/WebKitBuild/Release/ Baseline Patch Δ Peak Memory: reddit_memory_warning --runs 0 14,036kB 14,020kB ^ 1.0x smaller flickr_memory_warning --runs 0 26,732kB 26,700kB ^ 1.0x smaller theverge_memory_warning --runs 0 27,016kB 26,976kB ^ 1.0x smaller <geometric mean> 21,642kB 21,615kB ^ 1.0x smaller <arithmetic mean> 22,595kB 22,565kB ^ 1.0x smaller <harmonic mean> 20,595kB 20,569kB ^ 1.0x smaller Memory at End: reddit_memory_warning --runs 0 8,980kB 8,964kB ^ 1.0x smaller flickr_memory_warning --runs 0 12,280kB 12,248kB ^ 1.0x smaller theverge_memory_warning --runs 0 21,048kB 21,008kB ^ 1.0x smaller <geometric mean> 13,240kB 13,212kB ^ 1.0x smaller <arithmetic mean> 14,103kB 14,073kB ^ 1.0x smaller <harmonic mean> 12,484kB 12,458kB ^ 1.0x smaller
Comment on attachment 310186 [details] Patch r=me
Committed r216895: <http://trac.webkit.org/changeset/216895>