请问一下关于Prime95的出错信息问题

xjflyttp

[Fri Feb 04 11:38:40 2005]
Iteration: 1472637/20031667, ERROR: SUM(INPUTS) != SUM(OUTPUTS), 6.685791892860936e+017 != 6.685791892863557e+017
Possible hardware failure, consult the readme.txt file.
Continuing from last save file.


以上的是超频造成的吗？全部TEST都通过了。但是经常看到这样的ERROR发生。这台机子跑什么fah、3Dmark01 03 05 、superpi都没有发生任何错误。
请教.谢谢

碧城仙

在 readme.txt 文件中有如下文字：
=====================
The other two errors messages, SUMINP != SUMOUT and ROUND OFF > 0.40 are
caused by one of two things:
        1)  For reasons too complicated to go into here, the program's error
        checking is not        perfect.  Some errors can be missed and some correct
        results flagged as an error.  If you get the message "Disregard last
        error..." upon continuing from the last save file, then you may have
        found the rare case where a good result was flagged as an error.
        2)  A true hardware error.
=====================
楼主提到的是上面指出的第一个错误：“ SUMINP != SUMOUT ”，有上文可以看出有两种可能，一是“太多错综复杂的原因不在这里多说了，我们的程序检查错误的机制并不是很完美，有一些错误是可以忽略的，也有些正确的结果被视为错误，如果您得到的报错信息是“ Disregard last error...（忽略上一个错误...）”，从上一个错误之前的副本继续运算，那么您可能碰到程序将一个正确结果判断为错误的很为少见的情况。”；二是真正的硬件出错（可能是超频导致的硬件过热）。

也就是说从 readme.txt 文件并不能得知具体是什么错误，看来楼主得等几天，我到官方论坛上去问问看。

暂时请楼主在菜单栏上依次“Test”——“continue”，继续计算。

碧城仙

在 readme.txt 文件中亦有如下文字：
====================
If you do not get the "Disregard last error..." message or this happens
more than once, then your machine is a good candidate for a torture test.
See the stress.txt file for more information.
====================
意思是：如果您并没有得到“Disregard last error...（忽略上一个错误...）”的报错信息或者您的错误不止一次发生，那么说明您拥有一台非常适合作最残酷的测试的机器。更多详情请参见 stress.txt 文件。

碧城仙

在官方论坛上的相关回复：
======================
The SUMINP != SUMOUT error isn't a serious problem.
Prime95 just restarts from the last safefile. If the error doesn't recur, you shouldn't worry about it. There's also a thread on the forums dealing with recurring errors due to SoundCard-Drivers, but I'm not sure about its title.

Any driver can cause errors. I find them most in video cards. Just recently, some NVidia cards were causing errors. It happened that I moved the vid card to another machine and used old drivers. As soon as I updated the drivers, no more errors.

Good luck. Don't forget about Ram failure.
======================
意思是：
出现 SUMINP != SUMOUT 错误并不是一个严重的问题。Prime95 将从出错前最后一次保存的结果重新开始计算。如果该错误不复发，您就没有任何担心的必要。由于在我们的论坛里经常出现一些与声卡驱动有关的错误的讨论，所以我对您的问题在理解上不能完全把握。

很多驱动程序都会导致出错。我发现的最多的错误都是视频卡引起的。就像最近，一些 NVidia 核心的显卡经常引起错误，当时我把我的一张盘拿到另外一台机器上并使用了稍旧一点的驱动。后来我更新了它的驱动，就没有出现什么错误了。

祝你好运！不要忘记公羊失败！
======================
老外总是喜欢说很多与主题毫不相干的事情，顾左右而言他是他们的绝活！！！
B4ing................尤其是最后一句话，我也不知道该怎么翻译好，我估计是什么谚语之类的东西......

看来楼主还要再等几天。

碧城仙

wpolly  在 2005-2-9 22:52 发表:
Ram是内存.............................

RAM是内存，这个我知道的，只是老外大小写乱搞都把我搞糊涂了。

碧城仙

又有一老外回复了：
===============
Did you consult the README.TXT file?

Near the end of it is a section titled "POSSIBLE HARDWARE FAILURE". Since you got the "SUM(INPUTS) != SUM(OUTPUTS)" error, the section mentioning "SUMINP != SUMOUT" applies to your situation. Since you got this error more than once, the advice that applies is:

If ... this happens
more than once, then your machine is a good candidate for a torture test.
See the stress.txt file for more information.

Running the program on a computer with hardware problems may produce
incorrect results. This won't hurt the GIMPS project since all results
will be double-checked. However, you could be wasting your CPU time.
If you are getting several errors during each primality test, then I
recommend using your machine to factor Mersenne numbers or run less
strenuous distributed computing projects at
http://www.mersenne.org/projects.htm.

===============
他也只是说让看 README.TXT 文件，然后简单拷贝了 README.TXT 文件的最后一段话。

碧城仙

新回复来了：

Ding Cong:

Please be aware that prime95 pushes certain parts of your processor harder than they will ever be in Folding, 3dmark2003 etc

Therefore your computer may do those fine but report an error while doing prime95.

Also you may run the prime95 torture test for a while before one of these errors occurs.

Set it to torture test because it will compare results with the known correct answer. Leave it running prime95 in torture test for an extended period of several DAYS.

If you can pass that, then you can try using prime95 in production mode.

Are you saying you can run torture test mode without errors? If so for how long?

意思是：
请您意识到 Prime95 在测试您的处理器性能上比 Folding@home、3dmark2003 等等软件要更为严格！
因此，您的计算机可能在做那些测试时很正常，但是做 Prime95 的测试时就出现错误。
您仍然可以使用 Prime95 对您的计算机做一会会最为残酷的测试——当然，是在出现错误之前。
可以将测试结果作为最终的测试成绩与其他比较权威的测试结果做做比较。也可以由着 Prime95 自己运行一段时间。
如果能够通过最为残酷的测试，那么您可以正常的加入搜索梅森素数的队伍中。
不知道您的机器在做最为残酷的测试过程中是否还在出现错误？如果是的，这个错误持续出现多长时间了？

xjflyttp

测试是正常通过的，只是在正常运作的时候出现以上错误， 这个错误有时会出现，可能没2天会出现一次到两次

碧城仙

楼主是在同时运行 Folding@home 和 prime95 的吗？如果是的，停止 Folding@home ，只跑 prime95 试试看。