Libraryless. Click here for Pure Java version (5772L/32K).
// This is not that great. 31 mill/s cache lookups. // bench_concurrentHashMapBasedMethodCache_tweaked is still the best // (48 mill/s), at least with concurrency level 1. // // The advantage of ClassValue though is that it doesn't have to be // cleaned regularly if I understand this correctly. scope bench_classValueBasedMethodCache static long #cachesMade; static final ClassValue<_MethodCache> methodCacheClassValue = new { @Override protected _MethodCache computeValue(Class<?> type) { ++cachesMade; ret new _MethodCache(type); } }; static _MethodCache getMethodCache_cv(Class c) { ret methodCacheClassValue.get(c); } svoid bench_classValueBasedMethodCache(int nClasses default 1000) { print("Loading " + nClasses(nClasses)); L<Class> classes = map classForName(takeFirst(nClasses, classNamesInJigsawModule("java.base"))); print("Have " + nClasses(classes) + ". Filling method caches"); print(+cachesMade); time { for (c : classes) assertNotNull(getMethodCache_cv(c)); } print(+cachesMade); print("Looking them up again."); benchFor10(-> { for (c : classes) assertNotNull(getMethodCache_cv(c)); }); print(+cachesMade); }
Began life as a copy of #1034270
download show line numbers debug dex old transpilations
Travelled to 3 computer(s): bhatertpkbcr, mowyntqkapby, mqqgnosmbjvj
No comments. add comment
Snippet ID: | #1034271 |
Snippet name: | bench_classValueBasedMethodCache |
Eternal ID of this version: | #1034271/10 |
Text MD5: | 7bafdd41c3c5391551961036cb903630 |
Transpilation MD5: | 6d7245baa891cab392a2c4ba87fb0689 |
Author: | stefan |
Category: | javax / performance |
Type: | JavaX fragment (include) |
Public (visible to everyone): | Yes |
Archived (hidden from active list): | No |
Created/modified: | 2022-01-31 18:45:36 |
Source code size: | 1229 bytes / 42 lines |
Pitched / IR pitched: | No / No |
Views / Downloads: | 141 / 242 |
Version history: | 9 change(s) |
Referenced in: | [show references] |