Does running RC5 ruin your CPU? (edited)

[sciencewhiz]

Below is a private message that I would like to send to Stefan in response to the thread linked in the message.

Please pick it apart and tell me what I can improve.

-------------------------------------------------------------------------------------
Stefan, I didn't want to reply to your message here because I didn't want to start another flame war.

You claim that running RC5 killed your Overclocked CPU. It was not RC5 that killed your CPU, it was the fact that you were running your chip out of spec. Just because your CPU runs fine for a little while at a higher speed doesn't mean that you aren't doing damage to it internally. Your CPU would have died anyway.

CPUs don't care what kind of load they are running at when they are run in spec. This is why Intel rates their chips to last 7 years no matter what the load on the chip. When you overclock, you are automatically decreasing the life on your chip.

In either case, running RC5 will not cause a problem with your chip, it may cause the problem to show itself earlier. You just happened to find the problem the hard way, with a burned out chip.

As far as I can tell, you are the only one complaing that RC5 killed your CPU. What you happen to be ignoring is the fact that there are hundreds of people running RC5 in their overclocked systems whom haven't had a problem.

In reality, it wasn't RC5 that killed your chip, it was the fact that you were overclocking it. If you had been running Quake 3 and your chip died, would you really be telling everyone to stop playing Quake 3?

 

[BoberFett]

I cannot say unequivocally that it will not damage your CPU. Only from my experience can I say that I have yet to see it cause problems. I have been running RC5 on roughly 35 CPUs for a year or more and have seen no loss of stability or other effects that might be attributed to CPU damage.

 

[Russ]

Ditto what BoberFett said. I'm running it on hordes of overclocked CPU's and haven't had even problem one. I would think, though, that if a piece of hardware is close to death, additional stress might cause it to happen sooner. But, RC5 as the underlying cause of the failure? No.

Russ, NCNE

 

[BurntKooshie]

I've been running it 24/7 on my Cyrix for the last 2 years. We all know that Cyrix chips are out of spec when they are manufactured . Its still running.

 

[sciencewhiz]

Ok, so RC5 may possibly cause a problem on a overclocked CPU. But the idea is, when you are running overclocked, you take what you get and if your CPU fails, it isn't a certain programs fault, but yours for not running at spec.

Is it OK to say that on a non-overclocked chip running RC5 will not cause a problem?

Edit: Is my speeling OK?

 

[Russ]

I think that in either case, it will not cause a problem, but it may accelerate a problem that already exists.

Russ, NCNE

 

[Robor]

I've got SETI (not rc5 but still pegs the CPU usage at 100%) running on 23 boxes. At least 6 of them are overclocked. They all run 24/7/365 and I've had no CPU failures at all. Do I think rc5/SETI cause CPU failures? Absolutely not. But as Russ stated above, it could highlight an existing problem.

Stefan is one of those with a 300A overclocking it to 450/464. Most of them (myself included) had to boost the CPU voltage and/or use better cooling to get it stable. My guess is Stefan was running his CPU at it's thermal limit and getting away with it. Bad contact with the heatsink, no thermal grease, too much voltage, a heatsink not up to the task. Any of those could be the source of his problem. When he put rc5 on it he put more stress on the CPU because it was being used all the time. I don't think it would make a difference if he looped Q3A/UT overnight instead of rc5. In fact, the additional heat generated by the 3D card would probably be worse for cooling.

Anyway, I've been doing SETI since it started and I've had no problems with the systems it's running on. In fact, my old Celeron 300A @ 450 is still pumping away using 2.3v and a dualie fan from ComputerNerd!

Rob

 

[sciencewhiz]

message updated with Russ' wording.

Do you think I should bother sending this? I begining to think that I might confuse him with the facts.

 

[Assimilator1]

Ditto what Robor said (& the others)

If your cpu is on the limit clock speed/heat then loading 100% with any program would cause an earlier death than just using it for a light load.
Sounds to me like there was some kind of cooling problem.
BTW I run SETI 24/7 on my Cel 366 @ 550 & have done since about March ,previously 17/7 since November last year.The only problem I had was recently when the heatsink compound was beggining to go off giving me random crashes.Since replacing the compound it has been MUCH more stable .I've had the 366 @ 550 since August last year

 

[sciencewhiz]

Should I use this from the dnet overclocking faq?



<< Overclocking has been known to cause your machine to become more unstable, more prone to crashing, produce greater heat, and shorten the life of your processor. Furthermore, it is known to cause programs to unreliably execute code correctly, resulting in occasionally incorrect calculations. >>



I would like to avoid the use of anecdotal evidence that running RC5 will not harm a processor, because unfortunately he has anecdotal that running RC5 does.

 

[ElFenix]

i overclocked my pentium pro specifically for seti@home, which is probably harder on your processor than rc5, since it also makes use of the fpu (my understanding is that rc5 doesn't). its been running better than a year now like that, still fine (.35 micron was good to 266 or 300, so its well within tolerances. intel simply didn't rate processors faster than 200 back then, why bother, no one else had anything close.) its been on for more than a month right now, pretty much since i hooked it up at the new apartment.

 

[Russ]

<< because unfortunately he has anecdotal that running RC5 does. >>



Our anecdotal consists of a few hundred thousand systems. His consists of one. I'd say that's a smackdown.

Russ, NCNE

 

[sciencewhiz]

Ok, I added another paragraph. If there are no objections I will send it off to him and save it so I can send it to anyone else who wants to complain.

 

[Slacker]

I have seen this type of detractors attitude towards Team AnandTech and Rc5 before, it is a personal problem and has nothing to do with hardware.


I have a c366 that wont do win98 at 566 at less than 2.3 volts and cant do win2k at 2.3/2.4 , that cpu has been running at 566 24/7 for over a year and is still going strong.







its all about joining and acceptance

 

[ViRGE]

No objections from me.

 

[sciencewhiz]

Ok, I just sent the message. Hopefully he has the sense to at least stop crapping in our threads.

 

[dvch]

sciencewhiz: if he doesn't have &quot;the sense to at least stop crapping in our threads&quot; then your PM would make a nice public reply.

 

[dennilfloss]

One of the things I did when I started cracking was to bring my FSB down from 75MHz to the default 66MHz, bringing my PC from 562 to 500MHz. I was running Rain 1.0 on my overclocked PC and it helped bring the temperature down a lot. However, I could not continue to use Rain and do RC5 at the same time. I decided that running at default speed, with 2 case fans in my superminitower, all my cabling bundled neatly and lots of free space for air flow, and a large heatsink+fan+thermal grease combo on my Celeron, that it was probably safe to run RC5.

Choosing between overclocking and cracking, I decided to crack.

No regret.

 

[MWalkden]

I posted the following in that thread:

I think it has all been said already, but I will say it too.

If you don't like distributed computing then don't involve yourself.

A lot of us do this to give back to Anand for all he has provided us all these years, great information on hardware! Anand supports distributed computing (can you tell by the links on the home page?) And we show our support for him in DC projects.

It could very well be that a CPU running as far out of spec as you had it developed some real small problems and that running a distributed project added stress to those yet un-noticed problems. If that is true, then you just found out sooner then you would have otherwise. It would have happened sooner or later, maybe while playing an extended game, or simply when you opened a browser window! The damage was already there, and if using a CPU causes damage then we should all be losing CPUs left and right! If your CPU failed, it was not a project client that caused the failure, but it could be one to bring it to the forefront.

I too have OC'd CPU's running a DC client. 300a@450, 3-366's@550, Duron 650@850, P3-550@733. The 300a is my oldest chip, running strong after 2 years and 9 months! I also have about 40 other computers running it too! Tens of thousands of people running in spec and out of spec do distributed computing, with no problems!.

No, I'm afraid you are incorrect in thinking a DC project client damaged your CPU. Something else actually did that.

Now with that all said, you did have a CPU that did die. There are a lot of people in the distributed computing forum that may be willing to sell you a replacement really cheap, possibly even give it to you! We do try our best to take care of those who support this effort and invite you to that forum for any help we may be able to provide. We would love to have you back cracking too! So please join us!

 

[bigjon]

Some people will actually argue that running the CPU at 100% is actually better for it because it reduces thermal stress due to fluctuations in CPU temp.

Personally, I couldn't care less. The more crack the merrier!

 

[dawks]

SETI would bang your CPU harder.

 

[jinsonxu]

Well bigjon, i don't really believe that line but i do my best to convince myself of that when i'm belting out my speech to assimilate someone's computer.

 

[Eug]

Perhaps our overzealousness in trying to get new members is getting to some of us? Maybe our messages should not ignore issues such as these. It seems that some of the more recent recruiting drives could be interpreted as spam, especially since they don't always present a balanced view.

First off, even though the clients (OGR in my case) are set at idle priority, they CAN have a significant impact on the performance of certain software. We shouldn't deny that this exists. For instance, UT runs poorly on my machine if the RC5/OGR client is active. However, pausing the client or shutting it down solves the problem.

Second, it CAN affect some marginal computers. Some computers that work fine in Windows, surfing or emailing, etc. WILL choke if a distributed computing client is installed. Now, this probably means the computer is unstable to begin with, and it's not the client's fault, but the bottom line is that it can make Aunt Jessie's non-overclocked eMachines POS crash in the summer. And while a client may not directly cause a marginal CPU to die per se, I firmly believe it can hasten a bad CPU's demise.

Third, certain people have reported that they had compatibility issues with some of the clients, esp. if run as a screensaver.

Perhaps our FAQ threads should include these points.

Now get cracking!!!

 

[Robor]

Eug: In all honestly I must confess that you are 100% correct. Most if not all of use who participate in DC projects are computer &quot;experts&quot; (and I use that term loosely) . In other words, we've been working with 'puters and we know how to diagnose a problem and how to solve it.

In my Seti@Home experience I've found that if I install the SETI client (which version doesn't seem to matter) on weaker systems with <64MB of RAM it causes problems. Either the PC takes longer to return to &quot;normal&quot; condition or it causes other problems if left on for long periods without reboots. I'm not *positive* that SETI created the abnormal lockups/crashes/freezes but it's too much of a coincidence to think anything else. I probably shouldn't have bothered to load the client on these PC's because they are so slow that it would take a day or more to process a WU anyway but that's another topic (greed) .

Let me say that this *IS NOT* a hardware issue and has caused no hardware failures (even old and slow hardware doesn't fail). I think it's just a matter of overloading a PC with a program that wasn't designed to run on it in the first place. I would compare this to trying to load Win2000 Pro on a low end system and expecting it to run fine. It might install and run but it isn't going to be stable. That said I feel that most of &quot;us&quot; here know when and where to apply DC clients even if it takes some trial &amp; error to learn it!

Rob

PS - TA SETI is 15th in the world!!!

 

[Viztech]

Interesting discussion here.

Bigjon-
I happen to agree with the theory that thermal stress, ie expansion and contraction, is the real issue with CPU life. There are many examples of this scenario in other areas such as furnaces, air-cooled aircraft engines, exhaust valves on any engine, electric motors, and light bulbs.

All the above are examples that involve expansion/contraction caused by temperature differences or the inrush of electrical current causing physical damage.

Eug-
Good points my friend. Difficulties do exist, but are rare. Most of us here are indeed 'geek' enough to handle them. In my experience, most users don't know the difference. If there is a conflict, it is mostly with CD burner software and obscure programs.

Someone posted in one of the DanC server threads that it was bad to run the client on a server, particularly on an SQL Server. I have had the Dnet client running quite happily on an SQL server for months. It gives no loss of performance, or any more trouble than any other NT box.

Hmm, I have the Dnet client running on ~60 machines, and at least 7 or 8 are OCed.

viz