請問一下～關於CPU

[tyrael]

我現在已經更新到
6.14.11.8585
這個版本了～～
不過～又發現一個問題～
cpu使用率是掉下來了沒錯～
但PPD值也跟著掉下來了耶！！

有掉很多嗎? 5~10%應該算正常範圍

若是不能忍受PDD下降.就換回舊驅動吧

 

[Nightstalker]

版本，我不是很清楚，但我從控制台裡看到的資訊如下：

驅動程式提供者：NVIDIA
驅動程式日期：2008/6/16
驅動程式版本：6.14.11.7739
數位簽署者：未經過數位簽署

我的版本太低嗎？？需要跟新一下驅動嗎？？

懇請大大教導。3Q

用GPU-Z就可以看到驅動版本

如果樓主有細心爬文,

就應該知道以前就算跑GPU,

還是會CPU 100% Full Load,

只是NV驅動到了180.XX(又好像是181.XX)之後才不會吃CPU資源,

還有就是FahMon看PPD到比較準確,

是需要設定一下的.

 

[swellfish]

FAHMON我有下載，監看～
PPD值大概都在6500-8000之間。

用顯卡跑GPU端效能有差？？這是說用顯卡跑比用CPU跑『成績』會比較差嗎？？

小弟愚昧，懇請大大教導。

不好意思我看錯了
看到標題以為您只用cpu client端跑...

要看產值還是要長時間觀察的,看一時的PPD值個人覺得意義不大
多跑幾個WU再看看fahmon的benchmarks吧
我自已跑5771的PPD也是飄來飄去的..

Project : 5771
Core : GPUv2 Gromacs
Frames : 100
Credit : 353


-- GPU_1 --

Min. Time / Frame : 1mn 13s - 4177.97 ppd
Avg. Time / Frame : 1mn 29s - 3426.88 ppd
No Cur. Time / Frame
No R3F. Time / Frame
No Eff. Time / Frame


-- GPU_3 --

Min. Time / Frame : 57s - 5350.74 ppd
Avg. Time / Frame : 57s - 5350.74 ppd
No Cur. Time / Frame
No R3F. Time / Frame
No Eff. Time / Frame


-- GPU_2 --

Min. Time / Frame : 59s - 5169.36 ppd
Avg. Time / Frame : 1mn 25s - 3588.14 ppd
No Cur. Time / Frame
No R3F. Time / Frame
No Eff. Time / Frame


-- GPU_5 --

Min. Time / Frame : 1mn 31s - 3351.56 ppd
Avg. Time / Frame : 1mn 31s - 3351.56 ppd
No Cur. Time / Frame
No R3F. Time / Frame
No Eff. Time / Frame

 

[sd123456]

感謝各位大大的教導～
讓我多瞭解好幾課！！
剛剛又遇到問題了～～
FAHMON出現了下列訊息

[08:11:31] Completed 1%
[08:13:32] Completed 2%
[08:15:34] Completed 3%
[08:17:08] Completed 4%
[08:17:08] mdrun_gpu returned
[08:17:08] NANs detected on GPU
[08:17:08]
[08:17:08] Folding@home Core Shutdown: UNSTABLE_MACHINE
[08:17:10] CoreStatus = 7A (122)
[08:17:10] Sending work to server
[08:17:10] Project: 5779 (Run 15, Clone 204, Gen 4)
[08:17:10] - Read packet limit of 540015616... Set to 524286976.
[08:17:10] - Error: Could not get length of results file work/wuresults_01.dat
[08:17:10] - Error: Could not read unit 01 file. Removing from queue.
[08:17:10] EUE limit exceeded. Pausing 24 hours.
[09:47:09] + Working...

然後就卡住了～～～
我英文很差～～能告訴我這是怎麼一回是嗎？？？

 

[sd123456]

用GPU-Z就可以看到驅動版本

如果樓主有細心爬文,

就應該知道以前就算跑GPU,

還是會CPU 100% Full Load,

只是NV驅動到了180.XX(又好像是181.XX)之後才不會吃CPU資源,

～恕刪～

我的GPU-Z

FAHMON～的設定我也改好了～感謝大大教導～

 

[swellfish]

感謝各位大大的教導～
讓我多瞭解好幾課！！
剛剛又遇到問題了～～
FAHMON出現了下列訊息

[08:11:31] Completed 1%
[08:13:32] Completed 2%
[08:15:34] Completed 3%
[08:17:08] Completed 4%
[08:17:08] mdrun_gpu returned
[08:17:08] NANs detected on GPU
[08:17:08]
[08:17:08] Folding@home Core Shutdown: UNSTABLE_MACHINE
[08:17:10] CoreStatus = 7A (122)
[08:17:10] Sending work to server
[08:17:10] Project: 5779 (Run 15, Clone 204, Gen 4)
[08:17:10] - Read packet limit of 540015616... Set to 524286976.
[08:17:10] - Error: Could not get length of results file work/wuresults_01.dat
[08:17:10] - Error: Could not read unit 01 file. Removing from queue.
[08:17:10] EUE limit exceeded. Pausing 24 hours.
[09:47:09] + Working...

然後就卡住了～～～
我英文很差～～能告訴我這是怎麼一回是嗎？？？

有超頻嗎?
有可能超過頭變UNSTABLE_MACHINE
用原來的設定值跑看看

 

[Nightstalker]

感謝各位大大的教導～
讓我多瞭解好幾課！！
剛剛又遇到問題了～～
FAHMON出現了下列訊息

[08:11:31] Completed 1%
[08:13:32] Completed 2%
[08:15:34] Completed 3%
[08:17:08] Completed 4%
[08:17:08] mdrun_gpu returned
[08:17:08] NANs detected on GPU
[08:17:08]
[08:17:08] Folding@home Core Shutdown: UNSTABLE_MACHINE
[08:17:10] CoreStatus = 7A (122)
[08:17:10] Sending work to server
[08:17:10] Project: 5779 (Run 15, Clone 204, Gen 4)
[08:17:10] - Read packet limit of 540015616... Set to 524286976.
[08:17:10] - Error: Could not get length of results file work/wuresults_01.dat
[08:17:10] - Error: Could not read unit 01 file. Removing from queue.
[08:17:10] EUE limit exceeded. Pausing 24 hours.
[09:47:09] + Working...

然後就卡住了～～～
我英文很差～～能告訴我這是怎麼一回是嗎？？？

一般都是三個問題,

一:GPU(CPU)有超頻而導致系統不穩定,
又或者Display Card本身開始有毛病,

二:溫度過高,散熬不理想,

三:Power Supply不夠力,

至於是那個就要大大自己去排除.

 

[Kuster]

一:GPU(CPU)有超頻而導致系統不穩定,
又或者Display Card本身開始有毛病,
二:溫度過高,散熬不理想,
三:Power Supply不夠力,

我以為是顯示卡壞了說.....
我的顯示卡以前跑GPU穩穩的，跑了幾個月後就變成這樣.....

 

[tyrael]

～恕刪～

我的GPU-Z

由圖來看似乎有超頻?

要不要降回原始頻率來跑看看

或者加強散熱

 

[sd123456]

我把shader降下來1950就ok了～

但是～～～～～～～～～～～～～～
我想開始『加』跑cpu～
我下載了x86的程式～
但怎麼跑了兩個小時了～～還是0％
我關掉重開～還是不會動～
以下是fahnon跑出來的訊息

###############################################################################

Launch directory: C:\Documents and Settings\Administrator\Application Data\Folding@home-x86


[00:42:33] - Ask before connecting: No
[00:42:33] - User name: coolaler.com_sd123456 (Team 31403)
[00:42:33] - User ID: 638A215561FC9ADB
[00:42:33] - Machine ID: 2
[00:42:33]
[00:42:33] Loaded queue successfully.
[00:42:33] Initialization complete
[00:42:33]
[00:42:33] + Processing work unit
[00:42:33] Core required: FahCore_78.exe
[00:42:33] Core found.
[00:42:33] Working on queue slot 03 [June 5 00:42:33 UTC]
[00:42:33] + Working ...
[00:42:33]
[00:42:33] *------------------------------*
[00:42:33] Folding@Home Gromacs Core
[00:42:33] Version 1.90 (March 8, 2006)
[00:42:33]
[00:42:33] Preparing to commence simulation
[00:42:33] - Looking at optimizations...
[00:42:33] - Files status OK
[00:42:33] - Expanded 238146 -> 1164961 (decompressed 489.1 percent)
[00:42:33]
[00:42:33] Project: 4460 (Run 411, Clone 0, Gen 28)
[00:42:33]
[00:42:33] Assembly optimizations on if available.
[00:42:33] Entering M.D.
[00:42:54] (Starting from checkpoint)
[00:42:54] Protein: p4460_Seq52_Amber03
[00:42:54]
[00:42:54] Writing local files
[00:44:15] Extra SSE boost OK.
[00:44:15] Writing local files
[00:44:15] Completed 0 out of 1500000 steps (0%)