RESOLVED WONTFIX 155338
bmalloc: scan by object when the object is bigger than a line 
https://bugs.webkit.org/show_bug.cgi?id=155338
Summary bmalloc: scan by object when the object is bigger than a line
Geoffrey Garen
Reported 2016-03-10 16:50:27 PST
bmalloc: scan by object when the object is bigger than a line
Attachments
Patch (9.31 KB, patch)
2016-03-10 16:57 PST, Geoffrey Garen 		no flags
DetailsFormatted DiffDiff
Patch (11.52 KB, patch)
2016-03-10 18:08 PST, Geoffrey Garen 		no flags
DetailsFormatted DiffDiff
Show Obsolete (1) View All Add attachment proposed patch, testcase, etc.
Geoffrey Garen
Comment 1 2016-03-10 16:57:07 PST
Created attachment 273656 [details] Patch
Geoffrey Garen
Comment 2 2016-03-10 16:58:01 PST
~/OpenSource/PerformanceTests/MallocBench> ./run-malloc-benchmarks Baseline:~/OpenSource/WebKitBuildBaseline/Release/ Patch:~/OpenSource/WebKitBuild/Release/ Baseline Patch Δ Execution Time: churn 82ms 82ms list_allocate 73ms 73ms tree_allocate 81ms 82ms ! 1.01x slower tree_churn 87ms 88ms ! 1.01x slower fragment 71ms 71ms fragment_iterate 61ms 64ms ! 1.05x slower medium 190ms 194ms ! 1.02x slower big 146ms 146ms facebook 239ms 240ms ! 1.0x slower reddit 116ms 114ms ^ 1.02x faster flickr 122ms 118ms ^ 1.03x faster theverge 150ms 148ms ^ 1.01x faster nimlang 130ms 132ms ! 1.02x slower message_one 214ms 210ms ^ 1.02x faster message_many 116ms 116ms churn --parallel 38ms 38ms list_allocate --parallel 71ms 66ms ^ 1.08x faster tree_allocate --parallel 84ms 83ms ^ 1.01x faster tree_churn --parallel 84ms 85ms ! 1.01x slower fragment --parallel 55ms 48ms ^ 1.15x faster fragment_iterate --parallel 31ms 32ms ! 1.03x slower medium --parallel 188ms 187ms ^ 1.01x faster big --parallel 135ms 134ms ^ 1.01x faster <geometric mean> 98ms 98ms ^ 1.01x faster <arithmetic mean> 111ms 111ms ^ 1.01x faster <harmonic mean> 86ms 85ms ^ 1.01x faster Peak Memory: churn 796kB 776kB ^ 1.03x smaller list_allocate 2,100kB 2,088kB ^ 1.01x smaller tree_allocate 5,452kB 5,448kB ^ 1.0x smaller tree_churn 4,764kB 4,756kB ^ 1.0x smaller fragment 7,000kB 6,996kB ^ 1.0x smaller fragment_iterate 25,580kB 25,576kB ^ 1.0x smaller medium 1,068,416kB 1,068,424kB ! 1.0x bigger big 1,060,460kB 1,060,468kB ! 1.0x bigger facebook 73,428kB 73,472kB ! 1.0x bigger reddit 13,340kB 13,420kB ! 1.01x bigger flickr 25,348kB 25,312kB ^ 1.0x smaller theverge 26,488kB 26,488kB nimlang 216,900kB 216,840kB ^ 1.0x smaller message_one 5,280kB 4,560kB ^ 1.16x smaller message_many 4,940kB 5,244kB ! 1.06x bigger churn --parallel 968kB 956kB ^ 1.01x smaller list_allocate --parallel 2,212kB 2,228kB ! 1.01x bigger tree_allocate --parallel 3,320kB 3,324kB ! 1.0x bigger tree_churn --parallel 2,996kB 2,980kB ^ 1.01x smaller fragment --parallel 7,136kB 7,120kB ^ 1.0x smaller fragment_iterate --parallel 25,716kB 25,700kB ^ 1.0x smaller medium --parallel 1,066,248kB 1,066,368kB ! 1.0x bigger big --parallel 1,052,452kB 1,057,496kB ! 1.0x bigger <geometric mean> 18,523kB 18,424kB ^ 1.01x smaller <arithmetic mean> 204,406kB 204,610kB ! 1.0x bigger <harmonic mean> 4,446kB 4,391kB ^ 1.01x smaller Memory at End: churn 384kB 364kB ^ 1.05x smaller list_allocate 404kB 392kB ^ 1.03x smaller tree_allocate 432kB 428kB ^ 1.01x smaller tree_churn 424kB 416kB ^ 1.02x smaller fragment 448kB 444kB ^ 1.01x smaller fragment_iterate 628kB 624kB ^ 1.01x smaller medium 4,608kB 4,616kB ! 1.0x bigger big 4,592kB 4,600kB ! 1.0x bigger facebook 2,804kB 2,848kB ! 1.02x bigger reddit 1,712kB 1,792kB ! 1.05x bigger flickr 2,756kB 2,720kB ^ 1.01x smaller theverge 2,804kB 2,804kB nimlang 33,788kB 33,728kB ^ 1.0x smaller message_one 688kB 672kB ^ 1.02x smaller message_many 1,052kB 976kB ^ 1.08x smaller churn --parallel 556kB 544kB ^ 1.02x smaller list_allocate --parallel 584kB 576kB ^ 1.01x smaller tree_allocate --parallel 672kB 680kB ! 1.01x bigger tree_churn --parallel 1,096kB 1,124kB ! 1.03x bigger fragment --parallel 1,304kB 1,372kB ! 1.05x bigger fragment_iterate --parallel 828kB 816kB ^ 1.01x smaller medium --parallel 4,724kB 4,724kB big --parallel 4,668kB 4,708kB ! 1.01x bigger <geometric mean> 1,342kB 1,335kB ^ 1.01x smaller <arithmetic mean> 3,129kB 3,129kB ! 1.0x bigger <harmonic mean> 871kB 858kB ^ 1.02x smaller ===== ~/OpenSource/PerformanceTests/MallocBench> ./run-malloc-benchmarks Baseline:~/OpenSource/WebKitBuildBaseline/Release/ Patch:~/OpenSource/WebKitBuild/Release/ Baseline Patch Δ Peak Memory: reddit_memory_warning --runs 0 13,344kB 13,324kB ^ 1.0x smaller flickr_memory_warning --runs 0 25,588kB 25,556kB ^ 1.0x smaller theverge_memory_warning --runs 0 26,496kB 26,468kB ^ 1.0x smaller <geometric mean> 20,837kB 20,811kB ^ 1.0x smaller <arithmetic mean> 21,809kB 21,783kB ^ 1.0x smaller <harmonic mean> 19,768kB 19,742kB ^ 1.0x smaller Memory at End: reddit_memory_warning --runs 0 8,812kB 8,792kB ^ 1.0x smaller flickr_memory_warning --runs 0 11,876kB 11,844kB ^ 1.0x smaller theverge_memory_warning --runs 0 20,768kB 20,740kB ^ 1.0x smaller <geometric mean> 12,953kB 12,926kB ^ 1.0x smaller <arithmetic mean> 13,819kB 13,792kB ^ 1.0x smaller <harmonic mean> 12,203kB 12,176kB ^ 1.0x smaller
WebKit Commit Bot
Comment 3 2016-03-10 16:59:50 PST
Attachment 273656 [details] did not pass style-queue: ERROR: Source/bmalloc/bmalloc/Heap.cpp:135: Place brace on its own line for function definitions. [whitespace/braces] [4] ERROR: Source/bmalloc/bmalloc/Heap.cpp:194: Place brace on its own line for function definitions. [whitespace/braces] [4] Total errors found: 2 in 3 files If any of these errors are false positives, please file a bug against check-webkit-style.
Geoffrey Garen
Comment 4 2016-03-10 18:08:06 PST
Created attachment 273662 [details] Patch
WebKit Commit Bot
Comment 5 2016-03-10 18:10:17 PST
Attachment 273662 [details] did not pass style-queue: ERROR: Source/bmalloc/bmalloc/Heap.cpp:135: Place brace on its own line for function definitions. [whitespace/braces] [4] ERROR: Source/bmalloc/bmalloc/Heap.cpp:194: Place brace on its own line for function definitions. [whitespace/braces] [4] Total errors found: 2 in 5 files If any of these errors are false positives, please file a bug against check-webkit-style.
Note You need to log in before you can comment on or make changes to this bug.