新的bigbeta project 8104/8105已转正

金鹏

最新消息:BIGBETA包8105上线,属于8103和8104包系列
PPD略低于8103但是远远高于8101包,届时能大大缓解因690X下线造成的BA包缺粮的尴尬

1. by bollix47 » Thu Apr 18, 2013 6:00 pm
   
2. 
   
3. AMD 4P 6274(64 cores)@stock(2.2Ghz)
   
4. Ubuntu 11.10
   
5. 
   
6. 8101: Avg. Time / Frame : 00:13:53 - 335,793 PPD
   
7. 
   
8. 8102: Avg. Time / Frame : 00:10:52 - 484,917 PPD
   
9. 
   
10. 8103: Avg. Time / Frame : 00:10:59 - 477,211 PPD
    
11. 
    
12. 8104: Avg. Time / Frame : 00:08:04 - 501,058 PPD
    
13. 
    
14. 8105: Avg. Time / Frame : 00:10:51 - 452,079 PPD

复制代码

1. by Leonardo » Fri Apr 19, 2013 1:16 pm
   
2. 
   
3. 8103 and 8105 comparison on the same machine:
   
4. 
   
5. Machine: AMD Opteron, 4-Processor, 48-Core, 2.64GHz (overclocked)
   
6. OS: Linux/Ubuntu 10.10
   
7. 
   
8. Work Unit: 8105/0/81/0
   
9. TPF (average): 9:06
   
10. Production rate (PPD/HFM.NET): 588,600
    
11. 
    
12. Work Unit: 8103
    
13. TPF (average): 9:09
    
14. Production rate (PPD/HFM.NET): 627,600

复制代码

by kasson » Sat Apr 20, 2013 5:26 am

We have increased the base points value for 8105 slightly to 22606. We will continue to monitor. Thanks again for testing.

by kasson » Wed Apr 17, 2013 10:17 pm

Project 8105 is now officially released to beta. This is in the same series as 8103 and 8104. As indicated from early testing, points values are (now) very similar to 8103.
We may adjust this somewhat further, but here are the numbers for the moment.
Points: 21530
Deadline: 4 days
Preferred: 2.4 days
K-factor: 47.15
Thanks for testing!

by kasson » Fri Apr 05, 2013 2:28 am

We are pleased to announce a new bigbeta project coming online. 8104 is similar to 8103--it has a change to the molecules involved, so it runs slightly differently. We are continuing to look at how influenza virus gets into cells and how its proteins change shape and interact with membranes.

We have initially assigned values as follows. Please let us know how it compares to 8103 on your rig; we may adjust points and k-factors as we go.
17153 points
preferred deadline 1.8 days
final deadline 4 days
k-factor 50

by kasson » Fri Apr 05, 2013 6:48 am

The deadlines were by benchmark numbers, but again the standard benchmark machine has been less than predictive for bigadv performance. What I did was set initial numbers via standard benchmarking (with bigadv numbers) and then adjust PPD to match 8103 on a bigadv-capable machine. Since PPD don't scale perfectly, we may have to adjust a bit more in testing.

     

by kasson » Mon Apr 15, 2013 8:37 am

Please note the above comment about the assigns being a pre-benchmarking leak. 8105 is a clone of neither 8103 nor 8104 but a kind of hybrid between them. We expect 8105 to perform somewhere between 8103 and 8104. Our benchmarking setup has finally picked up a work unit, so we should be able to assign points within the next day or so (and there shouldn't be more bigbeta assigns between now and then).
If you have performance data, it would be good to know time to complete (or TPF) values compared to both 8103 and 8104.

今天开始新的BA测试包8104上线，个人猜测是替代上星期下线的6901和从未接到过的6905、6906

从原子数量上看8104和8101 8102 8103 一样，但是交梁日期为1.8天，明显比以前8101-8103的2.4天短了，难道是比8103还要甜蜜的甜包？
8104包PPD果然比8103的还高了1万余




PS：参考老贴新的bigbeta project 8103 和SMP2 beta project 8530-85XX 上线

1. 8104 folding progress report:
   
2. 
   
3. Machine: 32-Core AMD Opteron, 4-Processor, moderate overclock at 2.94GHz
   
4. OS: Linux/Ubuntu 10.10
   
5. Work Unit: 8104/0/11/0
   
6. TPF (average): 8:36
   
7. Production rate (PPD, as calculated by HFM.NET): 455179
   
8. 
   
9. This is within 2% of the production rate of a typical 8103 on the same machine.
   
10. 
    
11. UPDATE: Another machine is now Folding an 8104:
    
12. 
    
13. machine: 48-Core AMD Opteron, 4-Processor, 2.64GHz (overclocked)
    
14. OS: Linux/Ubuntu 10.10
    
15. Work Unit: 8104/0/27/0
    
16. TPF (average): 6.55
    
17. Production rate (PPD/HFM.NET): 631080

复制代码

1. AMD 4P 6274(64 cores)@stock(2.2Ghz)
   
2. Ubuntu 11.10
   
3. 
   
4. 8103: Avg. Time / Frame : 00:10:59 - 477,211 PPD
   
5. 
   
6. 8104: Avg. Time / Frame : 00:08:04 - 501,058 PPD

复制代码

1. Intel 2P E5-2660 (32 "cores")@stock (2.2GHz), Ubuntu 11.10
   
2. 8101: avg time/frame 21:55
   
3. 8103: avg time/frame 16:36
   
4. 8104: avg time/frame 12:25

复制代码

1. 4P 6272 stock, Ubuntu Natty Narwhal
   
2. p8104
   
3. 
   
4. [14:29:09] Completed 237500 out of 250000 steps  (95%)
   
5. [14:37:21] Completed 240000 out of 250000 steps  (96%)
   
6. [14:45:38] Completed 242500 out of 250000 steps  (97%)
   
7. [14:53:51] Completed 245000 out of 250000 steps  (98%)
   
8. [15:02:08] Completed 247500 out of 250000 steps  (99%)
   
9. [15:10:22] Completed 250000 out of 250000 steps  (100%)

复制代码

by kasson » Thu Apr 11, 2013 10:13 am

We have released 8104 to bigadv, as that will assist in bigadv work unit availability. Stay tuned for additional projects in this series. Please continue to report any problems. Thanks for testing!

by kasson » Sat Apr 13, 2013 10:14 am

Benchmarking on this project has not been completed yet. As you gathered, it was inadvertently released to beta (due to some technical issues). It is indeed similar to 8104; my guess is that it will run somewhere in between 8103 and 8104 and have k-factors and deadlines aligned accordingly.

xiaoren6412

哎，看到新BA，心痒痒，可以没火力了，跑不了。能跑的都去跑吧

0dragon

我都是随机收wu的。。。

vennyfan

期待ppd爆发

xiaoren6412

7、8月会有次PPD爆发，之后就没了，呵呵

金鹏

xiaoren6412 发表于 2013-4-5 23:25
7、8月会有次PPD爆发，之后就没了，呵呵

爆2个月威力相当于2年

金鹏

同频下8104比8103高了大约1万余PPD,目前为止算是最甜的甜包

zj0511

老大，这包怎样设置才能收到？？

金鹏

zj0511 发表于 2013-4-8 15:41
老大，这包怎样设置才能收到？？

16T+机器加参数: -bigbeta      

superTWT

木有16T的小水管泪崩飘过。。。。。

vennyfan

金鹏 发表于 2013-4-8 16:22
16T+机器加参数: -bigbeta

12T的3930K能接到吗？？

金鹏

vennyfan 发表于 2013-4-8 21:41
12T的3930K能接到吗？？

模拟16T草到4.5+G后可以试试
参考http://www.equn.com/forum/thread-34800-1-7.html

vennyfan

金鹏 发表于 2013-4-9 07:13
模拟16T草到4.5+G后可以试试
参考http://www.equn.com/forum/thread-34800-1-7.html

OK 3Q 有空了试试看

金鹏

8105甜包也快来了

by kasson » Sat Apr 13, 2013 10:14 am

Benchmarking on this project has not been completed yet. As you gathered, it was inadvertently released to beta (due to some technical issues). It is indeed similar to 8104; my guess is that it will run somewhere in between 8103 and 8104 and have k-factors and deadlines aligned accordingly

金鹏

vennyfan 发表于 2013-4-9 22:20
OK 3Q 有空了试试看

目前8103 8104包你的3930K草到4.1G+就能妥妥的跑完，遇到8101包也只要4.5G就行

最好使用原生LINUX+神医3.5.4内核，
怕麻烦使用1.5.0那个VB镜像单跑BA也可以

9-12万的PPD你无法拒绝