支持79XX的CORE16_V2.13版内核开始内测

金鹏

很遗憾，是基于最新版的OpenMM，版本号2.13，AMD专门做过优化，配套测试包P10635，需要加beta内测标志才能在最新版的V7客户端下接到
PS: 搞不懂为啥不是基于OpenCL ，估计OpenMM下79XX还是一如既往的PPD杯具

Project 10635 (core16/ATI) will soon be released on beta to test a new core16 with v7 client.

Note: Beta testers please have the AMD Catalyst 11.7 or higher drivers installed.

Points and deadlines: 897 points, preferred deadline 4 days, final deadline 5 days

1. 01:28:54:Adding folding slot 00: READY gpu:0:"Tahiti XT [Radeon HD 7970]"
   
2. 01:28:55:WU00:FS00:Connecting to assign-GPU.stanford.edu:80
   
3. 01:28:55:Saving configuration to config.xml
   
4. 01:28:55:<config>
   
5. 01:28:55:  <!-- Folding Slot Configuration -->
   
6. 01:28:55:  <smp v='false'/>
   
7. 01:28:55:
   
8. 01:28:55:  <!-- Network -->
   
9. 01:28:55:  <proxy v=':8080'/>
   
10. 01:28:55:
    
11. 01:28:55:  <!-- User Information -->
    
12. 01:28:55:  <team v='33'/>
    
13. 01:28:55:  <user v='EvilPenguin'/>
    
14. 01:28:55:
    
15. 01:28:55:  <!-- Folding Slots -->
    
16. 01:28:55:  <slot id='0' type='GPU'>
    
17. 01:28:55:    <client-type v='beta'/>
    
18. 01:28:55:  </slot>
    
19. 01:28:55:</config>
    
20. 01:28:55:WU00:FS00:News: Welcome to Folding@Home
    
21. 01:28:55:WU00:FS00:Assigned to work server 171.64.65.56
    
22. 01:28:55:WU00:FS00:Requesting new work unit for slot 00: READY gpu:0:"Tahiti XT [Radeon HD 7970]" from 171.64.65.56
    
23. 01:28:55:WU00:FS00:Connecting to 171.64.65.56:8080
    
24. 01:28:56:WU00:FS00:Downloading 18.85KiB
    
25. 01:28:56:WU00:FS00:Download complete
    
26. 01:28:56:WU00:FS00:Received Unit: id:00 state:DOWNLOAD error:OK project:10635 run:2 clone:2 gen:16 core:0x16 unit:0x000000100a3b1e5c4f3031117f553d12
    
27. 01:28:56:WU00:FS00:Downloading core from http://www.stanford.edu/~pande/Win32/AMD64/ATI/R600/beta/Core_16.fah
    
28. 01:28:56:WU00:FS00:Connecting to www.stanford.edu:80
    
29. 01:28:56:WU00:FS00:FahCore 16: Downloading 1.13MiB
    
30. 01:28:59:WU00:FS00:FahCore 16: Download complete
    
31. 01:28:59:WU00:FS00:Valid core signature
    
32. 01:28:59:WU00:FS00:Unpacked 2.60MiB to cores/www.stanford.edu/~pande/Win32/AMD64/ATI/R600/beta/Core_16.fah/FahCore_16.exe
    
33. 01:28:59:WU00:FS00:Starting
    
34. 01:28:59:WU00:FS00:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:/ProgramData/FAHClient/cores/www.stanford.edu/~pande/Win32/AMD64/ATI/R600/beta/Core_16.fah/FahCore_16.exe -dir 00 -suffix 01 -version 701 -checkpoint 15 -gpu 0
    
35. 01:28:59:WU00:FS00:Started FahCore on PID 4876
    
36. 01:28:59:WU00:FS00:Core PID:1264
    
37. 01:28:59:WU00:FS00:FahCore 0x16 started
    
38. 01:28:59:WU00:FS00:Downloading project 10635 description
    
39. 01:28:59:WU00:FS00:Connecting to fah-web.stanford.edu:80
    
40. 01:29:00:WU00:FS00:0x16:
    
41. 01:29:00:WU00:FS00:0x16:*------------------------------*
    
42. 01:29:00:WU00:FS00:0x16:Folding@Home GPU Core
    
43. 01:29:00:WU00:FS00:0x16:Version                2.13 (Mon Jan 30 15:37:02 PST 2012)
    
44. 01:29:00:WU00:FS00:0x16:Build host             user-f6d030f24f
    
45. 01:29:00:WU00:FS00:0x16:Board Type             AMD/OpenCL
    
46. 01:29:00:WU00:FS00:0x16:Core                   16
    
47. 01:29:00:WU00:FS00:0x16:
    
48. 01:29:00:WU00:FS00:0x16:Window's signal control handler registered.
    
49. 01:29:00:WU00:FS00:0x16:Preparing to commence simulation
    
50. 01:29:00:WU00:FS00:0x16:- Looking at optimizations...
    
51. 01:29:00:WU00:FS00:0x16:DeleteFrameFiles: successfully deleted file=00/wudata_01.ckp
    
52. 01:29:00:WU00:FS00:0x16:- Created dyn
    
53. 01:29:00:WU00:FS00:0x16:- Files status OK
    
54. 01:29:00:WU00:FS00:0x16:sizeof(CORE_PACKET_HDR) = 512 file=<>
    
55. 01:29:00:WU00:FS00:Project 10635 description downloaded successfully
    
56. 01:29:00:WU00:FS00:0x16:- Expanded 18789 -> 75731 (decompressed 403.0 percent)
    
57. 01:29:00:WU00:FS00:0x16:Called DecompressByteArray: compressed_data_size=18789 data_size=75731, decompressed_data_size=75731 diff=0
    
58. 01:29:00:WU00:FS00:0x16:- Digital signature verified
    
59. 01:29:00:WU00:FS00:0x16:
    
60. 01:29:00:WU00:FS00:0x16:Project: 10635 (Run 2, Clone 2, Gen 16)
    
61. 01:29:00:WU00:FS00:0x16:
    
62. 01:29:00:WU00:FS00:0x16:Assembly optimizations on if available.
    
63. 01:29:00:WU00:FS00:0x16:Entering M.D.
    
64. 01:29:02:WU00:FS00:0x16:Tpr hash 00/wudata_01.tpr:  2218487377 1157777680 3988771318 2192315900 2731675856
    
65. 01:29:02:WU00:FS00:0x16:GPU device id=0
    
66. 01:29:02:WU00:FS00:0x16:Working on 264 Fs_coil
    
67. 01:29:02:WU00:FS00:0x16:Client config unavailable.
    
68. 01:29:02:WU00:FS00:0x16:Starting GUI Server
    
69. 01:29:04:WU00:FS00:0x16:Setting checkpoint frequency: 500000
    
70. 01:29:04:WU00:FS00:0x16:Completed         3 out of 50000000 steps (0%).
    
71. 01:31:39:WU00:FS00:0x16:Completed    500000 out of 50000000 steps (1%).
    
72. 01:34:29:WU00:FS00:0x16:Completed   1000000 out of 50000000 steps (2%).
    
73. 01:37:19:WU00:FS00:0x16:Completed   1500000 out of 50000000 steps (3%).

复制代码

hmaimi

openmm是一个三维分子模拟库
基于opencl和cuda的程序也可以通过这个库来进行优化

nekoko

同上，openMM和opencl没有什么关系……

zxj-iq180

amd又杯具了....

blah

俺下过openmm的源代码，支持a卡的是opencl，也只是opencl，opencl木有刻意优化能达到50%效能就可以偷着笑了...。N卡是用的cuda

wpf999

1、OpenMM里面调用了OpenCL

2、造成PPD偏低的原因可能是OpenMM代码没有针对7900新硬件进行专门优化

金鹏

GPU usage fluctuates between 99% and 76% every few seconds.
Memory usage fluctuates between 290k KB and 340k KB every few seconds as well.
Core 16 still uses up an entire CPU core.

金鹏

CORE16版本真多，开始支持CUDA了

kvip

N卡啊,快出来吧,回归N卡后打酱油的每天跑跑吧.呼呼.

opshang

N卡的PPD不低啊

wdya6

果断升级啊~~~~

wdya6

虽然还不一定稳定