garbage collection - How can I make the old region resize adaptively in Java GC? -

Java 6 (1.6.0_23) for GC tuning, I use the -XX: + UseParallelOldGC Use

  -XX: + printing-distribution -XX: + use ambient policy -XX: + useAdaptive GCoundary   
 My goal is that the GC Calculating an optimal ratio between the old and new areas, however, the visual jvmstat 3.0 shows that while the proportion between Eden and the survivors is between my workload Notes that the revised date field on their initial size (strict) remains, which  newRatio  (2 in my case).  
 What's wrong with my settings or expectations?  
 After the 19th collection, the videographer shows the old General below 500 meters, while the old General's reserved and committed size is equal to 624 meters, I have set the pneumasse from 30m and the Max size size 300 The meter is set, but it does not change anything: Neurortho seems impossible ... The print advertive policy speaks only about the survivor  
 The issue is that VisualGc's "reserved space" Old does not show anything for the general space ... I -XX: The Use is made: + without any changes Upyogpr adaptive Jisibounderi.  
 Here are some logs:  -XX: + PrintTraining Distribution -XX: + PrintGCedit   
 is a complete GC, and the last print of the stack ... the desired distribution becomes distorted ...  
  desired survivor size 119275520 bytes, new range 1 (max 15) [PSYoungGen: 182096K-> 66576K (233024K)] 681978K- & gt; 631994K (932096K), 0.25,82,3 9 0 secs] [Times: user = 0.29 sys = 0.00, real = 0.26 secs] 116.431 [GC desired survivor Size 119,275,520 bytes, the new range 1 (max 15) [PSYoungGen: 183120K - & gt; 39952K (233024K)] 748538K- & gt; 670910K (932096K), 0.1578620 seconds] [Times: User = 0.29 sys = 0.00, Real = 0.16 sec] 116.589: [Full GC [PSYoungGen: 39952K-> 0K (233024K)] [ParOldGen: 630958K- & Gt; 310429 (6797-7)] 670 9 10-of-> 310429 (932096) [PSPR: 217-K-> 219 73A (3092/2, 1.3326720 seconds)] [Times: Users = 1.79 sys = 0.01, real = 1.33 sec] 118.050: [GC desired survivor size 119,275,520 bytes, new limit 1 (max 15) [PSYoungGen: 116544K-> 18448K (233024K)] 426973K-> 328877K (932096K), 0.0745320 secs] [Times: user = 0.14 sys = 0.00, real = 0.08 secs] 118.234 [GC desired survivor Size 119,275,520 bytes, the new range 1 (max 15) [PSYoungGen: 134992K- & gt; 24592K (233024K)] 445421K - & gt; The 35140 9 (9320 9 6), 12:12 9 9 60 seconds] [Times: user = 0.14 sys = 0.00, actual = 0.13 secs] 118.526 [GC required survivor size 114,884,608 bytes, new range 1 (max 15) [PSYoungGen: 141136K - & gt; 36880K (237312K)] 467953K- & gt; 380085K (936384K), 0.11,87,320 seconds] [Times: user = 0.21 sys = 0.00, real = 0.12 secs] 118.759 [GC desired survivor Size 108,593,152 bytes, the new range 1 (max 15) [PSYoungGen: 142446K - & gt; 40976K (233024K)] 485652K-> 416957K (932096K), 0.11,53,230 seconds] [Times: User = 0.23 sys = 0.00, real = 0.12 secs] ads PSYoungGen total 233024K, 97957K used (0xb3a40000, 0x9e4f0000, 0x9e4f0000] Eden Space 120 832 47% Used [0x9e4f0000, 0xa1c95668,0xa5af0000] Space 1121 9 2K, 36% [0xa5af0000,0xa82f4020,0xac880000] Space used in 106048K, 0% [0xad2b0000,0xad2b0000,0xb3a40000] ParOldGen total 699072K, 375981K [ 0x73a40000, 0x9e4f0000, 0x9e4f0000] Object Space 699072 K, 53 [0x73a40000,0x8a96b730,0x9e4f0000] Used PSPermGen Total 30592K, 21981k [0x6fa40000, 0x71820000, 0x73a40000] Object Space 30592 In the experiment, 71% [0x6fa40000,0x70fb7668,0x71820000] was used    
 
 
 
 How can I optimize the old area in Java GC?   
 I do not think this is possible. There is a limited set of GC's "tuning knob", and I can not see who does this for you. I realize that you think about "problem".  
 Reference:  
 
 Joseph D. If this answer is not enough, then your options are limited:  
 
 
 If you are using appropriate channels If you went through the medium and were ready to cut out some money, then you probably get some help from Oracle Tuning Expert.   
 
 If you have the time and patience, you can try to get your head around the GC source code in the OpenJDK code-base.