have you had a computer DIE from Distributed Computing

[Niege]

I've been running my 1.2G Thunderbird OC'd to 1.35G 24/7 for over 3.5 years now. Same 768Mb of PC-133. No trouble so far. I've swapped out two HDs but only because I needed more space. I've had software thingees foul things up and routers and cable modems drive me crazy but the base rig keeps on crunchin'....

 

[ProviaFan]

I had two old (cyrix and AMD K6-2) systems fail, but that was due to crappy motherboard rather than anything else - and the first one didn't run DC because I was 12, hadn't discovered AnandTech yet, and by extension hadn't discovered DC either. Heck, it could have been caused by bad capacitors, but I had no idea if that problem existed back then. More recently, I had an Epox 8RDA fail due to bad caps, but that's hardly the fault of 100% CPU utilization by DC.

Other than that, I have nothing to contribute; others have done a great job putting tons of useful info in this thread - thanks guys! :thumbsup:

 

[Kravahn]

mondobyte,

you make a lot of great points that unfortunately will be lost...

Your last statement, paraphrased... components fail within the first 30 days. This is common knowledge among resellers, and if you consider the availability of 'extended warranty' options from your local retailer, you notice it becomes big business. If an electronic component is gonna fail (due to mfg flaw) it's within the first 30 days. Don't ask me for the studies, I read them a few years ago but don't have links... use your google 😉 .

If you're silly, you can buy an upgraded warranty for almost anything (I got offered a 2 year upgrade for $15 on a $30 saw that came with a 5 year warranty

Most people here (I think) know a good power supply is critical to a stable system... still, a good point reinforced.

Concerning dusting, I'm a big fan of suckage vs. blowage... blowing dust leads to impingement damage (damage caused by particles impacting components at high speed) as well as caking (dust layered onto existing grease/moisture) Suck it up and avoid the hassle...

UPS... required for a stable, lasting rig... If you don't have a line-conditioning UPS don't bitch to me about surge (or Serge, if that's what you wanna call him). You spend a couple grand on a computer and hesitate to spend a C-note or two on power? Shame on you.

Like Mondobyte said, cooling is crucial. A cool rig is a happy rig, and the cooler the better (barring condensate).

That's my take, thought I'd back-up Mondobyte, with such good advice

 

[Kravahn]

Wow, I clicked 'reply' and here I am at a generic no-frills page... hehe

Anyway, couple things I wanted to respond to... Electronic components have a break-in period... for the most part, if they're gonna fail, they're gonna fail in the first 30 days. If they haven't failed by then, expect them to meet or exceed their expected lifecycle. Just Components 101.

Don't buy extended warranties... see above...

 

[mondobyte]

Originally posted by: Kravahn
mondobyte,

you make a lot of great points that unfortunately will be lost...
... That's my take, thought I'd back-up Mondobyte, with such good advice

Oh ... I hope my comments will not be lost ...

Every one of them can be backed up with fact.

Thanks for your support Kravahn ... you are likely more sensible than most.

I've been running PC's since the mid '80's and although I cannot claim to be a "guru", my experiences certainly have paralleled those of many, many, others.

Sometimes, experience can be a very kind teacher.

mondo

 

[Mlani]

No never lost a PC for doing DC

Oldest P11 450 mhz It ran 42 hours for 1 WU I ran 700 or so on it. PC was on from 16:00 - 23:00 and in the wwekend fr - sun 08:30 - 23:00. It is not running seti any more but still work pefectly.

This PC is a AMD athlon 2600+ It runs 3,5 hour for 1 wu and is running 24/7 from WU 700 until now WU 2100. I keep running seti classic until stop an d then go for bionc seti. I ran also some time find a drug simoltainiosly but that is a bit much for this machine.

 

[lobadobadingdong]

the ones that run DC seem to run longer.

 

[BadThad]

I've been running various DC projects for over 5 years, they cannot "kill" a PC. Computers are designed to run under load, if not, then what's the point of having a computer, lol. The only thing that wears out are the moving parts, i.e fans and hard drives, but they wear out no matter what runs on the PC.

 

[BespinReactorShaft]

I'll always attribute the death of my Quantum Fireball 13.6GB HDD to SETI Classic.

R.I.P.
1998-2003

I'm not the least bitter over it. Being written over and over on the same sectors, day in day out for five long years is no mean feat. Your sacrifice shall not be forgotten. 😀

(Still, lesson learnt, in my new rig I've decided to move DC away from my main HDD's and onto an external enclosure with an older WD 40GB.)

 

[kranky]

Had about three dozen computers running RC5 24/7 and never had a problem. I cannot imagine how any DC project could damage a PC.

 

[trevinom]

Originally posted by: ming2020
I'll always attribute the death of my Quantum Fireball 13.6GB HDD to SETI Classic.

R.I.P.
1998-2003

I'm not the least bitter over it. Being written over and over on the same sectors, day in day out for five long years is no mean feat. Your sacrifice shall not be forgotten. 😀

(Still, lesson learnt, in my new rig I've decided to move DC away from my main HDD's and onto an external enclosure with an older WD 40GB.)

How do you know it was writing to the same sectors?

 

[BespinReactorShaft]

Originally posted by: trevinom

How do you know it was writing to the same sectors?

Speculation. SETI results are stored in files which are regularly updated (i.e. written over and over), right?

 

[Hurricane Andrew]

Originally posted by: BlackMountainCow
I think that one can't answer that question right away. Did the PSU in my PentiumIII 600 fail due to DC or would it have failed anyway? Would it have failed during gaming as well? Who knows?!

Aside that general problem, I never lost a cruncher specifically due to DC and right now a PIII600 is my oldest cruncher.

How's that PIII perform? I have an old PIII-550 that my kids use. I haven't set it up as a cruncher since crunching really takes a toll on it when the kids are playing games and such. When it was my main rig, it seemed to take a day or two per WU, and frankly, that doesn't seem to justify the electricity...

 

[dmcowen674]

Originally posted by: ming2020
I'll always attribute the death of my Quantum Fireball 13.6GB HDD to SETI Classic.

R.I.P.
1998-2003

I'm not the least bitter over it. Being written over and over on the same sectors, day in day out for five long years is no mean feat. Your sacrifice shall not be forgotten. 😀

(Still, lesson learnt, in my new rig I've decided to move DC away from my main HDD's and onto an external enclosure with an older WD 40GB.)

That's a pretty amazing feat considering Windows does not write to the same sectors.

Were you running Linux perhaps?

 

[RaySun2Be]

I've got a PII450 running, does a WU in about 14 hours, no other use, 3.03 CLI version. 🙂

So your rig should do better than that, depending on how much it's used. 🙂

(If it wasn't a dually setup, I probably wouldn't run it. )

 

[ShotgunSteven]

Originally posted by: narzy
What is your oldest rig?
Have you lost a rig due to distributed computing?

I had a harddrive that was failing. Turns out the only thing keeping it going was SETI@Home. If I turned off SETI, the drive failed, the machine crashed, et cetera. The drive was long overdue for death anyway, but as long as I ran SETI@Home it kept chugging away. I think it was something to do with the constant writes to the drive, kept it from locking up.

 

[trevinom]

Originally posted by: ming2020

Originally posted by: trevinom

How do you know it was writing to the same sectors?

Speculation. SETI results are stored in files which are regularly updated (i.e. written over and over), right?

There are many items at play here. If it's anything like F@H, the results files have a size limit at which point the current file is renamed to xxx.old (or words to that effect) and a new one gets started.
At the hardware level, I believe that the chances of a file having a contiguous chunk of address space, especially if it's a large results file, is pretty nil. More than likely, when the file gets opened for update access, chunks (layman's term) are appropriated that can be in any of a number of unused sectors. The same thing happens when the file grows past it's size limit. When another chunk of adress space is assigned to the file, it will come from a pool of available, addresses that the system has designated as overwrittable, so the chances of the same sector being grabbed is not very high, especially if you have a large HD.

 

[BGod]

My oldest rig is an Asus P2B-S with a Katami core P3 500 @ 560 MHz. I shut if off due to the utility bills about 6 months ago. It ran 24/7 for 5 years. It did puke out a SCSI Plextor CD burner, does that count?