Pick my "burn-in" program!

[ShotgunSteven]

Like the title says. Pick your pet DC project and sway me to run it for a month to burn-in the quad-core system I am building. Will run it only for an AnandTech affiliated team, of course.

 

[Allio]

Seventeen or Bust is great because it's super sensitive to errors (maths related), logs any mistakes it finds and will happily run on all four cores without using too much memory

Edit: or any bandwidth!

 

[TallBill]

Seventeen or Bust because we're moving in on Free DC for the #3 spot and every bit helps.

 

[Rattledagger]

To "shake out the bugs", I would recommend Seasonal Attribution, since AFAIK it's the DC-project putting the highest simultaneous load on cpu and memory-systems.
Seasonal Attribution uses 450 MB memory, so for a quad-core you needs minimum 2 GB memory, if you're not going to share with other BOINC-projects... Depending on system-speed, it's likely each wu takes atleast 2 weeks to crunch...
To join team, http://attribution.cpdn.org/team_display.php?teamid=19

If memory-requirements is too high, the main CPDN-project is a good choise, but each wu will likely take 2+ months to crunch...
It's more than one "stable", often overclocked but not always, computer that has been running another DC-project for months, that craps-out then trying CPDN, showing computer wasn't really stable afterall...
To join team, http://climateapps2.oucs.ox.ac...m_display.php?teamid=4


Probably the 3rd. choise to give high system-load should in theory be Folding@Home's SMP-client, but atleast in my opinion this beta-client has too many bugs of it's own. Also, in case of buggy wu's crapping-out for "unknown" reasons, Folding@Home is AFAIK the only DC-project there same computer can be re-assigned the same wu again and again...
Still, you can be luckier than me with the SMP-client, so it's worth a try. Would recommend not to try running any other DC-projects at the same time...

If don't want to try the SMP-client, the "normal" FAH console-client is still a good choise to give high cpu-load but at the same time doing something useful.
Regardless of client, in Folding@Home, type-in 198 for team-id.


Another choise is to download and run an optimized SETI@home-application.

For the rest of the BOINC-projects, I don't know any that gives any exceptionally high load, so my recommendation would be to run a combination of them, to try to find the computers "weak" spots.


Now, after the "burn-in" on the other hand, my recommendations would be different, among other things I would not recommend any of FAH's cpu-clients as long as they've not got a BOINC-application...
With 40+ BOINC-projects, there's a good chance there's something for your taste.

 

[petrusbroder]

What Rattledagger said ... my preference is very similar.

 

[Amaroque]

What is the name of that program that runs the Prime95 torture test on multiple cores? I would reccomend running that for 24 Hrs. Just for stability testing only. It's very sensitive to errors.

IIRC Orthos, thats it...

 

[VirtualLarry]

There's a new Prime95 that will run on up to four (more?) cores, look for it at majorgeeks. I'm still unsure how well it works as a torture test, as some of the code has also been re-written.

 

[Assimilator1]

Originally posted by: Allio
Seventeen or Bust is great because it's super sensitive to errors (maths related), logs any mistakes it finds and will happily run on all four cores without using too much memory

Edit: or any bandwidth!

I thought SoB used lots of memory bandwidth? or did you mean net b/w?

Originally posted by: petrusbroder
What Rattledagger said ... my preference is very similar.

Except with the w's & t's swapped around

Amaroque
That's right

 

[Alyx]

I'll pipe in that I've had great experiences so far with the folding@Home SMP client. I've only had a couple work units end early in the few months I used it (and I don't think you'll care if they end early anyways, your stress testing ).

 

[Uppsala9496]

The F@H SMP program is nice. I've got it running on a quad core and an X2 and have only had a couple of faulty wu's with both systems. Nothing to worry about, and since you are just stress testing, it shouldn't matter. It will give all your cores a workout.

 

[Allio]

Originally posted by: Assimilator1
I thought SoB used lots of memory bandwidth? or did you mean net b/w?

Yeah, net bandwidth. The tests and results are a few bytes.

It does like tight memory timings and lots of memory bandwidth, you're right.

 

[Golgatha]

Originally posted by: Amaroque
What is the name of that program that runs the Prime95 torture test on multiple cores? I would reccomend running that for 24 Hrs. Just for stability testing only. It's very sensitive to errors.

IIRC Orthos, thats it...

Run four instances of Folding@Home and Orthos. Then open up your disk defragmenter, run a virus scan, burn a DVD, and fire up 3DMark looping. Do all of that for 24hrs.

Yes, I'm a bit excessive when it comes to stability.

 

[Golgatha]

Or just loop a high unit count Supreme Commander gameplay recording .

 

[Assimilator1]

Originally posted by: Golgatha

Originally posted by: Amaroque
What is the name of that program that runs the Prime95 torture test on multiple cores? I would reccomend running that for 24 Hrs. Just for stability testing only. It's very sensitive to errors.

IIRC Orthos, thats it...

Run four instances of Folding@Home and Orthos.

Err ,I think that would be pointless as one of the programs would simple stall