找回密码
 新注册用户
搜索
查看: 13294|回复: 37

王者归来-GPU Bonus大甜包P8057在CORE15-2.22/2.25下的不同PPD表现

[复制链接]
发表于 2012-11-10 19:27:04 | 显示全部楼层 |阅读模式
本帖最后由 金鹏 于 2012-11-20 10:50 编辑

今天RP高涨,480 560TI 560 都接到了GPU奖励包8057 (加了 -betateam 参数,包大小设置取消了10M勾选的普通接收)、(V7客户端client-type   beta)
PPD那叫一个甜蜜蜜、蜜蜜甜啊

CORE15-2.22核心下:

850核心频率的GTX480跑出TPF=1:46  PPD=16.24万  
970核心频率的GTX560TI的TPF=1:43  PPD=16.83万
910核心频率的GTX560的  TPF=2:01  PPD=13.22万
960核心频率的GTS450的  TPF=2:37  PPD= 9.00万
SP频率1944的9600GSO 384M 取得了TPF=5:25  PPD达到30.3K

vennyfan发表
报告一下 最近的心得 2.22内核 8057包 G92 G200都是用乾坤大挪移算的
GTS250    702-1944   45K
GTX260    675-1458   50K
GTS450    900            70K
GTX460    870            110k
GTX560TI 970            176K


a.PNG


1.PNG


4.PNG
  1. Folding@Home Client Version 6.41r2

  2.                           http://folding.stanford.edu

  3. ###############################################################################
  4. ###############################################################################

  5. Launch directory: C:\Folding@home-gpu3-0
  6. Arguments: -betateam -verbosity 9

  7. [10:30:47] - Ask before connecting: No
  8. [10:30:47] - User name: husq (Team 3213)
  9. [10:30:47] - User ID: 4F430D202514EF95
  10. [10:30:47] - Machine ID: 1
  11. [10:30:47]
  12. [10:30:47] Gpu type=3 species=20.
  13. [10:30:47] Loaded queue successfully.
  14. [10:30:47] Initialization complete
  15. [10:30:47]
  16. [10:30:47] + Processing work unit
  17. [10:30:47] Core required: FahCore_15.exe
  18. [10:30:47] Core found.
  19. [10:30:47] - Autosending finished units... [November 10 10:30:47 UTC]
  20. [10:30:47] Trying to send all finished work units
  21. [10:30:47] + No unsent completed units remaining.
  22. [10:30:47] - Autosend completed
  23. [10:30:47] Working on queue slot 09 [November 10 10:30:47 UTC]
  24. [10:30:47] + Working ...
  25. [10:30:47] - Calling '.\FahCore_15.exe -dir work/ -suffix 09 -nice 19 -priority 96 -checkpoint 3 -verbose -lifeline 6756 -version 641'

  26. [10:30:47]
  27. [10:30:47] *------------------------------*
  28. [10:30:47] Folding@Home GPU Core
  29. [10:30:47] Version                2.22 (Thu Dec 8 17:08:05 PST 2011)
  30. [10:30:47] Build host             SimbiosNvdWin7
  31. [10:30:47] Board Type             NVIDIA/CUDA
  32. [10:30:47] Core                   15
  33. [10:30:47]
  34. [10:30:47] Window's signal control handler registered.
  35. [10:30:47] Preparing to commence simulation
  36. [10:30:47] - Ensuring status. Please wait.
  37. [10:30:57] - Looking at optimizations...
  38. [10:30:57] - Working with standard loops on this execution.
  39. [10:30:57] - Previous termination of core was improper.
  40. [10:30:57] - Files status OK
  41. [10:30:57] sizeof(CORE_PACKET_HDR) = 512 file=<>
  42. [10:30:57] - Expanded 66886 -> 261042 (decompressed 390.2 percent)
  43. [10:30:57] Called DecompressByteArray: compressed_data_size=66886 data_size=261042, decompressed_data_size=261042 diff=0
  44. [10:30:57] - Digital signature verified
  45. [10:30:57]
  46. [10:30:57] Project: 8057 (Run 0, Clone 118, Gen 0)
  47. [10:30:57]
  48. [10:30:57] Entering M.D.
  49. [10:30:59] Will resume from checkpoint file work/wudata_09.ckp
  50. [10:30:59] Tpr hash work/wudata_09.tpr:  5435426 3075663414 3146638198 3548376167 1443208750
  51. [10:30:59] GPU device info: vendor=0 device=0 name=<NA> match=0
  52. [10:30:59] Working on Giving Russians Opium May Alter Current Situation
  53. [10:30:59] Client config found, loading data.
  54. [10:30:59] Starting GUI Server
  55. [10:32:01] Resuming from checkpoint
  56. [10:32:01] fcCheckPointResume: retreived and current tpr file hash:
  57. [10:32:01]    0      5435426      5435426
  58. [10:32:01]    1   3075663414   3075663414
  59. [10:32:01]    2   3146638198   3146638198
  60. [10:32:01]    3   3548376167   3548376167
  61. [10:32:01]    4   1443208750   1443208750
  62. [10:32:01] fcCheckPointResume: file hashes same.
  63. [10:32:01] fcCheckPointResume: state restored.
  64. [10:32:01] fcCheckPointResume: name work/wudata_09.log Verified work/wudata_09.log
  65. [10:32:01] fcCheckPointResume: name work/wudata_09.trr Verified work/wudata_09.trr
  66. [10:32:01] fcCheckPointResume: name work/wudata_09.xtc Verified work/wudata_09.xtc
  67. [10:32:01] fcCheckPointResume: name work/wudata_09.edr Verified work/wudata_09.edr
  68. [10:32:01] fcCheckPointResume: state restored 2
  69. [10:32:01] Resumed from checkpoint
  70. [10:32:01] Setting checkpoint frequency: 500000
  71. [10:32:01] Completed  14500001 out of 50000000 steps (29%).
  72. [10:33:47] Completed  15000000 out of 50000000 steps (30%).
  73. [10:35:33] Completed  15500000 out of 50000000 steps (31%).
  74. [10:37:19] Completed  16000000 out of 50000000 steps (32%).
  75. [10:39:06] Completed  16500000 out of 50000000 steps (33%).
复制代码
捕获.PNG


CORE15-2.25核心下:

850核心频率的GTX480跑出TPF=2:20  PPD=10.70万  
970核心频率的GTX560TI的TPF=2:07  PPD=12.39万
910核心频率的GTX560的  TPF=2:27  PPD=9.95万
960核心频率的GTS450的  TPF=3:10  PPD= 6.77万


2.PNG
  1. Folding@Home Client Version 6.41r2

  2.                           http://folding.stanford.edu

  3. ###############################################################################
  4. ###############################################################################

  5. Launch directory: C:\Folding@home-gpu3-0
  6. Arguments: -betateam -verbosity 9

  7. [10:56:14] - Ask before connecting: No
  8. [10:56:14] - User name: husq (Team 3213)
  9. [10:56:14] - User ID: 4F430D202514EF95
  10. [10:56:14] - Machine ID: 1
  11. [10:56:14]
  12. [10:56:14] Gpu type=3 species=20.
  13. [10:56:14] Loaded queue successfully.
  14. [10:56:14] Initialization complete
  15. [10:56:14]
  16. [10:56:14] - Autosending finished units... [November 10 10:56:14 UTC][10:56:14] + Processing work unit

  17. [10:56:14] Trying to send all finished work units
  18. [10:56:14] + No unsent completed units remaining.
  19. [10:56:14] - Autosend completed
  20. [10:56:14] Core required: FahCore_15.exe
  21. [10:56:14] Core found.
  22. [10:56:14] Working on queue slot 09 [November 10 10:56:14 UTC]
  23. [10:56:14] + Working ...
  24. [10:56:14] - Calling '.\FahCore_15.exe -dir work/ -suffix 09 -nice 19 -priority 96 -checkpoint 3 -verbose -lifeline 7208 -version 641'

  25. [10:56:14]
  26. [10:56:14] *------------------------------*
  27. [10:56:14] Folding@Home GPU Core
  28. [10:56:14] Version                2.25 (Wed May 9 17:03:01 EDT 2012)
  29. [10:56:14] Build host             AmoebaRemote
  30. [10:56:14] Board Type             NVIDIA/CUDA
  31. [10:56:14] Core                   15
  32. [10:56:14]
  33. [10:56:14] Window's signal control handler registered.
  34. [10:56:14] Preparing to commence simulation
  35. [10:56:14] - Ensuring status. Please wait.
  36. [10:56:24] - Looking at optimizations...
  37. [10:56:24] - Working with standard loops on this execution.
  38. [10:56:24] - Previous termination of core was improper.
  39. [10:56:24] - Files status OK
  40. [10:56:24] sizeof(CORE_PACKET_HDR) = 512 file=<>
  41. [10:56:24] - Expanded 66886 -> 261042 (decompressed 390.2 percent)
  42. [10:56:24] Called DecompressByteArray: compressed_data_size=66886 data_size=261042, decompressed_data_size=261042 diff=0
  43. [10:56:24] - Digital signature verified
  44. [10:56:24]
  45. [10:56:24] Project: 8057 (Run 0, Clone 118, Gen 0)
  46. [10:56:24]
  47. [10:56:24] Entering M.D.
  48. [10:56:26] Will resume from checkpoint file work/wudata_09.ckp
  49. [10:56:26] Tpr hash work/wudata_09.tpr:  5435426 3075663414 3146638198 3548376167 1443208750
  50. [10:56:26] GPU device id=0
  51. [10:56:26] Working on Giving Russians Opium May Alter Current Situation
  52. [10:56:26] Client config found, loading data.
  53. [10:56:26] Starting GUI Server
  54. [10:57:26] Resuming from checkpoint
  55. [10:57:26] fcCheckPointResume: retreived and current tpr file hash:
  56. [10:57:26]    0      5435426      5435426
  57. [10:57:26]    1   3075663414   3075663414
  58. [10:57:26]    2   3146638198   3146638198
  59. [10:57:26]    3   3548376167   3548376167
  60. [10:57:26]    4   1443208750   1443208750
  61. [10:57:26] fcCheckPointResume: file hashes same.
  62. [10:57:26] fcCheckPointResume: state restored.
  63. [10:57:26] fcCheckPointResume: name work/wudata_09.log Verified work/wudata_09.log
  64. [10:57:26] fcCheckPointResume: name work/wudata_09.trr Verified work/wudata_09.trr
  65. [10:57:26] fcCheckPointResume: name work/wudata_09.xtc Verified work/wudata_09.xtc
  66. [10:57:26] fcCheckPointResume: name work/wudata_09.edr Verified work/wudata_09.edr
  67. [10:57:26] fcCheckPointResume: state restored 2
  68. [10:57:26] Resumed from checkpoint
  69. [10:57:26] Setting checkpoint frequency: 500000
  70. [10:57:26] Completed  14500001 out of 50000000 steps (29%).
  71. [10:59:46] Completed  15000000 out of 50000000 steps (30%).
  72. [11:02:07] Completed  15500000 out of 50000000 steps (31%).
  73. [11:04:28] Completed  16000000 out of 50000000 steps (32%).
  74. [11:06:48] Completed  16500000 out of 50000000 steps (33%).
复制代码
3.PNG


从以上两组数据对比,看来未来GPU奖励分数全面开花时应以2.25核心下的PPD为参考,2.22核心下的PPD有点忒变态了
2.25核心下的PPD要比2.22核心下低了25-35%(这个从目前没有奖励分的包的PPD对比下也能看出)
从目前一接到7626高温毒包客户端就强制下载2.25版的CORE15核心来看也能得到未来是2.25核心天下的佐证。

结论:GPU Bonus包很甜 ,以后的显卡PPD参考2.25的核心下PPD为准,斯坦福此举符合吸引更广泛参与的初衷
(如此一来BIGADV在高配置门槛下,相比GPU的PPD无优势,苦逼了俺们一些BA党
      GPU王者归来了



###############################################################################
###############################################################################
  1. Configuration: FX-8120@4500 1.5V, 8Gb, W7x64, GTX580@865/1000, +SMP 8
  2. Project number: 8057 (0-16-3)
  3. Work unit: p8057
  4. WU size: 56,3 КБ
  5. WU result: ~ 345,24 КБ
  6. Credit: 22341,73
  7. Frames: 100
  8. Core: OPENMMGPU
  9. Server IP: 171.67.108.144
  10. PPH (Points Per Hour): 9613,17
  11. PPD (Points per day): 230716
  12. Avg time per step: 0:01:23
  13. Bonus factor: 8,7649
  14. Client.cfg: bigpackets=big
  15. Comleted: 19%
  16. FahSpy 2.0.1
复制代码
  1. Configuration: i7-2600K@4500, 8Gb, W7x64, GTX570@825/1000 1.075V
  2. Project number: 8057 (0-6-4)
  3. Work unit: p8057
  4. WU size: 56,3 КБ
  5. WU result: ~ 345,24 КБ
  6. Credit: 19599,77
  7. Frames: 100
  8. Core: OPENMMGPU
  9. Server IP: 171.67.108.144
  10. PPH (Points Per Hour): 6473,32
  11. PPD (Points per day): 155360
  12. Avg time per step: 0:01:49
  13. Bonus factor: 7,6892
  14. Client.cfg: bigpackets=big
  15. Comleted: 26%
  16. FahSpy 2.0.1
复制代码
  1. Project ID: 8057 Core: OPENMMGPU Credit: 2549

  2. Name: EVGA 690

  3. Avg. Time / Frame : 00:01:54
  4. Est credit 18980
  5. Est credit / day = 144,000 ppd per core

  6. Running two cores - 288,000 ppd total!!
复制代码
  1. v7 259 2x670ftw(stock) core15 v2.25

  2. 8057wu's

  3. (0,195,0) tpf 2m2s est credit 18264 est ppd 128287
  4. (0,279,0) tpf 2m5s est credit: 18100 est PPD 125174
  5. (0,146,7) tpf 2m5s est Credit 18198 est ppd 125782
复制代码
  1. These work units hardly tax my GTX 660ti at all.
  2. P8057
  3. 2 min 11 sec tpf
  4. 69% power
  5. 57 c
  6. 117k ppd

  7. p8018
  8. 91% power
  9. 64 c
  10. 19.5 k ppd
复制代码
  1. Project ID: 8057
  2. Core: OPENMMGPU
  3. Credit: 2549
  4. Frames: 100

  5. Name: F57-T5XE-i860-3.5+2xGT430-XP64-V729 Slot 00 (GT430, core 15 v222)
  6. Path: F57-T5XE-860-36330
  7. Number of Frames Observed: 300
  8. Min. Time / Frame : 00:07:37 - 18,147 PPD
  9. Avg. Time / Frame : 00:07:37 - 18,147 PPD


  10. Name: Hayes-P5Kpro-Q9450-X32--V7 Slot 00 (GTS450, core 15 v222)
  11. Path: P5K-PRO-HAYES-36330
  12. Number of Frames Observed: 300
  13. Min. Time / Frame : 00:03:00 - 73,411 PPD
  14. Avg. Time / Frame : 00:03:00 - 73,411 PPD

  15. Name: Hayes-P5Kpro-Q9450-X32--V7 Slot 02  (GTX550Ti, core 15 v222)
  16. Path: P5K-PRO-HAYES-36330
  17. Number of Frames Observed: 300
  18. Min. Time / Frame : 00:02:46 - 82,891 PPD
  19. Avg. Time / Frame : 00:02:46 - 82,891 PPD

  20. Name: Pam-P5Kpro-Q9450-X32--V7 Slot 00  (GTS450, core 15 v222)
  21. Path: P5K-PRO-PAM-36330
  22. Number of Frames Observed: 300
  23. Min. Time / Frame : 00:03:06 - 69,888 PPD
  24. Avg. Time / Frame : 00:03:07 - 69,328 PPD

  25. Name: Pam-P5Kpro-Q9450-X32--V7 Slot 02 (GT440, core 15 v222)
  26. Path: P5K-PRO-PAM-36330
  27. Number of Frames Observed: 300
  28. Min. Time / Frame : 00:06:39 - 22,244 PPD
  29. Avg. Time / Frame : 00:06:39 - 22,244 PPD
复制代码
  1. Project ID: 8057
  2. Core: OPENMMGPU
  3. Credit: 2549
  4. Frames: 100

  5. Name: GTX 460
  6. Path: \\10.10.0.2\flatlinegpu\
  7. Number of Frames Observed: 300

  8. Min. Time / Frame : 00:03:36 - 55,846 PPD
  9. Avg. Time / Frame : 00:03:37 - 55,460 PPD
  10. Cur. Time / Frame : 00:03:37 - 55,121 PPD
  11. R3F. Time / Frame : 00:03:37 - 55,121 PPD
  12. All  Time / Frame : 00:03:37 - 55,121 PPD
  13. Eff. Time / Frame : 00:03:41 - 53,927 PPD
复制代码
  1. F53-slot1 GTX650Ti (core 15, v2.25) Avg. Time / Frame : 00:03:12 - 66,638 PPD
复制代码
  1. F53-slot1 GT520 (core 15, v2.25) Avg. Time / Frame : 00:18:11 - 4,920 PPD
  2. F53-slot1 GT520 (core 15, v2.22) Avg. Time / Frame : 00:14:35 - 6,849 PPD

  3. F54-slot1 GT430 (core 15, v2.22) Avg. Time / Frame : 00:08:51 - 14,489 PPD
  4. F54-slot2 GT430 (core 15, v2.22) Avg. Time / Frame : 00:08:53 - 14,407 PPD
  5. F54-slot3 GT430 (core 15, v2.22) Avg. Time / Frame : 00:09:29 - 13,062 PPD

  6. F57-slot0 GT430 (core 15, v2.22) Avg. Time / Frame : 00:07:37 - 18,147 PPD
  7. F57-slot1 GT430 (core 15, v2.22) Avg. Time / Frame : 00:07:41 - 17,911 PPD

  8. Hayes-slot0 GTS450 (core 15, v2.22) Avg. Time / Frame :  00:03:01 - 72,246 PPD
  9. Hayes-slot2 GTX550Ti (core 15, v2.22) Avg. Time / Frame : 00:02:47 - 81,995 PPD

  10. Pam-slot0 GTS450 (core 15, v2.22) Avg. Time / Frame : 00:03:07 - 68,824 PPD
  11. Pam-slot0 GT440 (core 15, v2.22) Avg. Time / Frame :  00:06:39 - 22,217 PPD
复制代码

评分

参与人数 1基本分 +8 收起 理由
ONLY + 8 值得参考

查看全部评分

回复

使用道具 举报

 楼主| 发表于 2012-11-10 19:54:49 | 显示全部楼层
本帖最后由 金鹏 于 2012-11-11 22:20 编辑

奉上GPU3客户端下8057包供有兴趣的兄弟测试TPF,然后用http://www.linuxforge.net/bonuscalc2.php  估算你的显卡PPD吧--兄弟们欢呼吧

注意:此包对应俺的PASSKEY,俺已经上交了,千万别跑完100%后上传斯坦福,会坏了俺的成功率

8057.rar (226.36 KB, 下载次数: 3834)



FahCore_15_v2.22.rar (1.03 MB, 下载次数: 4060)




回复

使用道具 举报

发表于 2012-11-10 20:33:09 | 显示全部楼层
看来最有性价比的卡仍然归GTS450莫属了~
回复

使用道具 举报

发表于 2012-11-10 20:49:52 | 显示全部楼层
GPU奖励分政策目前存在一些争议,下面会如何发展还有待观察。
回复

使用道具 举报

发表于 2012-11-10 22:46:54 | 显示全部楼层
金版V5
回复

使用道具 举报

发表于 2012-11-10 23:32:24 | 显示全部楼层
本帖最后由 wpf999 于 2012-11-10 23:45 编辑

托金版的福啊,我也接到一个

screenshot.png
回复

使用道具 举报

发表于 2012-11-11 00:20:01 | 显示全部楼层
金版V5~~~~~~~
最近GPU加分这些好像没有新的情况
回复

使用道具 举报

 楼主| 发表于 2012-11-11 11:50:30 | 显示全部楼层
回复 3# chinalianxin

感觉性价比依次是GTS450  GTX560TI  GTX550TI  GTX560
回复

使用道具 举报

 楼主| 发表于 2012-11-11 11:51:43 | 显示全部楼层
回复 4# cuda

无非是分值的平衡了,奖励计划三年前就开始探讨了,到今天应该是水到渠成了
回复

使用道具 举报

 楼主| 发表于 2012-11-11 11:53:28 | 显示全部楼层
回复 6# wpf999

430也焕发春情了 PPD提高了3.3倍
回复

使用道具 举报

 楼主| 发表于 2012-11-11 11:54:38 | 显示全部楼层
回复 7# muclemanxb

目前只有8057包进行内测,等开始公测了GPU的PPD就开始王者归来了
回复

使用道具 举报

发表于 2012-11-11 12:11:41 | 显示全部楼层
本帖最后由 wpf999 于 2012-11-11 12:17 编辑

回复 4# cuda


    equal points for equal work,我觉得这样才能促进fah的效益最大化
回复

使用道具 举报

发表于 2012-11-11 12:20:51 | 显示全部楼层
本帖最后由 wpf999 于 2012-11-11 12:23 编辑

回复 11# 金鹏

we have started a GPU3 project (Project ID: 8057) and released it for beta testing. Once the benchmarking scheme has been tested, all the current GPU projects will be re-benchmarked to reflect the changes in the benchmarking scheme.



    8057就是公测项目,beta testing(软件工程术语)就是公测的意思。

一旦8057公测完成,GPU奖励分制度就全面施行。
回复

使用道具 举报

发表于 2012-11-11 12:22:01 | 显示全部楼层
回复 12# wpf999

有人说从这个前提推导下去会出现一些矛盾,详见这篇文章(很长):
http://hardforum.com/showthread.php?t=1725242
回复

使用道具 举报

发表于 2012-11-11 12:24:43 | 显示全部楼层
回复 14# cuda


好的,我观摩一下
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 新注册用户

本版积分规则

论坛官方淘宝店开业啦~

Archiver|手机版|小黑屋|中国分布式计算总站 ( 沪ICP备05042587号 )

GMT+8, 2024-4-27 11:30

Powered by Discuz! X3.5

© 2001-2024 Discuz! Team.

快速回复 返回顶部 返回列表