FAH一遇到GPU重置就进度丢失,都到91%了,白跑了几小时

hunterr007

FAH一遇到GPU重置就进度丢失,都到91%了,白跑了几小时

蓝宝4870什么都好,就是满载温度高
第一次16%GPU重置了,重新来,好吧,我忍了
这次91%挂了,无语
另外我的4870是公版频率(750/3600),PPD2500多是不是太低了?WHY?

以下是第一次挂时候FahMon的记录:
[07:55:24] Completed 15%
[07:58:17] Completed 16%
[08:01:08] mdrun_gpu returned
[08:01:08] NANs detected on GPU
[08:01:08]
[08:01:08] Folding@home Core Shutdown: UNSTABLE_MACHINE
[08:01:11] CoreStatus = 7A (122)
[08:01:11] Sending work to server
[08:01:11] Project: 5732 (Run 4, Clone 18, Gen 327)
[08:01:11] - Read packet limit of 540015616... Set to 524286976.
[08:01:11] - Error: Could not get length of results file work/wuresults_03.dat
[08:01:11] - Error: Could not read unit 03 file. Removing from queue.
[08:01:11] - Preparing to get new work unit...
[08:01:11] + Attempting to get work packet
[08:01:11] - Connecting to assignment server
[08:01:13] - Successful: assigned to (171.64.65.102).
[08:01:13] + News From Folding@Home: Welcome to Folding@Home
[08:01:13] Loaded queue successfully.
[08:01:15] + Closed connections
[08:01:20]
[08:01:20] + Processing work unit
[08:01:20] Core required: FahCore_11.exe
[08:01:20] Core found.
[08:01:20] Working on queue slot 04 [September 1 08:01:20 UTC]
[08:01:20] + Working ...
[08:01:20]
[08:01:20] *------------------------------*
[08:01:20] Folding@Home GPU Core - Beta
[08:01:20] Version 1.24 (Mon Feb 9 11:00:12 PST 2009)
[08:01:20]
[08:01:20] Compiler  : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86
[08:01:20] Build host: amoeba
[08:01:20] Board Type: AMD
[08:01:20] Core      :
[08:01:20] Preparing to commence simulation
[08:01:20] - Looking at optimizations...
[08:01:20] - Created dyn
[08:01:20] - Files status OK
[08:01:20] - Expanded 98485 -> 492188 (decompressed 499.7 percent)
[08:01:20] Called DecompressByteArray: compressed_data_size=98485 data_size=492188, decompressed_data_size=492188 diff=0
[08:01:20] - Digital signature verified
[08:01:20]
[08:01:20] Project: 5732 (Run 4, Clone 18, Gen 327)
[08:01:20]
[08:01:20] Assembly optimizations on if available.
[08:01:20] Entering M.D.
[08:01:26] Tpr hash work/wudata_04.tpr:  203324067 1894399566 1126698296 3150467650 2005325704
[08:01:27] Working on Protein
[08:01:27] Client config found, loading data.
[08:01:27] Starting GUI Server
[08:06:45] Completed 1%
[08:12:25] Completed 2%

金鹏

产生此项现象是显卡不稳定造成，机箱风道不好导致显卡温度过高会产生产生此现象，超频过度也会产生（你的750/3600应该是默认频率应该排除此原因）。
还有就是系统软件冲突，
兄弟试试安装一下DirectX 2009.3，下载地址http://tools.mydrivers.com/soft/811.htm
安装Microsoft .NET Framework ，并打上所有更新补丁，http://tools.mydrivers.com/soft/672.htm
完全卸载原驱动，安装一下催情剂9.8或者以下低版本试试看

4870在催情剂8.12后计算FAH效率有所提升，你目前2500PPD的确低了些，看看是不是设置GPU占用100％，另外关闭其他3D程序，

具体参考下我原来4870毒药的PPD值



[ 本帖最后由 金鹏 于 2009-9-2 17:07 编辑 ]

bugfix

最近天气凉快了,电压上来了,所以不稳定开始了,我这也下降严重.

hunterr007

谢谢斑竹的回答.可能是蓝宝4870黄金版的满载温度过高,目前我吧占用率调到了80%,至今未出现问题

另外我的2550PPD是跑511包的,参考750MHz的4830,我的显卡PPD应该还算正常吧
目前配置是:WIN7 RC7100,X64
DX10,DX9都有
FrameWork 4.0
催化剂9.8 For X64

4870毒药真有那么高?能不能给个4850的值做参考

Bit_Rab

还是稳定重要……
我今天96GT跑到81%的时候因为超频稍微高了点，出错了……
只好重来……