CPU stress tester better than Orthos !!! (CoreDamage)

O

old skool

[H]ard|Gawd
Joined
Apr 5, 2000
Messages
1,694
Orthos and Prime95 are really good for checking the accuracy of the CPU but not so good at making it hot or testing your cooling.

This little proggy will make your CPU really fucking hot so monitor the temperature (use CoreTemp) or you might fry it.
http://damage.vigilantesoftware.com

hightemp.png
 
old skool said:
Orthos and Prime95 are really good for checking the accuracy of the CPU but not so good at making it hot or testing your cooling.
Click to expand...

I beg to differ ... I've been using Prime95 since the days of the Celeron300a for load testing. It gets you temps that are always a few 'C more than real world usage scenarios while also proving to be a great chipset/memory stress tester.

If your hardware has a problem Prime95 will tell you. ;)
 
I've always found that Intel's own TAT shows the hottest temps for Intel CPUs. It is consistently about 5C per core hotter than multi-threaded Orthos.
 
Could it be argued that it's not necessary to use synthetic means to get the cores as hot as you possibly can, since it's not going to happen in real life?

That's not to say that it's a bad idea checking that it's all stable at loads close to the theoretical maximum, just wondered what thoughts were.
 
TAT will give you the maximum you'd ever see to test your thermal solution and you don't have to leave your PC doing nothing useful for 24 hours just to see if your rig stays cool enough.

Testing for stability (ie orthos) is different than testing for adequate cooling. HTPC builders may not overclock at all but may want to run a fanless setup and need to test for thermal properties only.
 
I agree with the last two posters, you want to run something like TAT to check your thermal solution under worst case just to make sure you dont thermal trip the cpu/northbridge AND then run something like Orthos for stablity and making sure your OC is good.

As alluded too above the "torture" tests do not stress all parts/functions of the cpu so in theory (and often in practice) you could pass the thermal but fail in orthos, it would seldom, if ever, be the other way around.
 
BillParrish said:
As alluded too above the "torture" tests do not stress all parts/functions of the cpu so in theory (and often in practice) you could pass the thermal but fail in orthos, it would seldom, if ever, be the other way around.
Click to expand...


Correct. Its not possible to run the CPU at full-tilt and check the results, so its never running at the highest possible temperature when checking for accuracy. To get the maximum heat output, you have keep the pipeline full and disregard the results and just keep going and going... Thats why CoreDamage is good at making sure your cooling is good while Orthos and Prime 95 are good at checking for accuracy.


The problem is there are no publically available software to get maximum possible heat output until now. CoreDamage will do just that. Name one other program (publically available) that can match it???
 
your processor was 59C at idle? oh my god.
 
What's the purpose of pushing impossibly high temperatures through your CPU besides possibly damaging it? Real-world temps will never reach close to that and the purpose of the CPU is to generate accurate, quality data, not produce heat. I just don't see the purpose beyond testing the thermal limit of a heatsink, and I would not want to use a CPU as the guinea pig. FWIW, I use four instances of Prime x64 large FFT to test my max overclocks.
 
orthos always gave me far more heat then say F@H, as said, i want stability and accuracy and if small ftt's in orthos isnt enough heat,. i dont know what app i will ever use that will give off more heat then orthos would
 
old skool said:
The problem is there are no publically available software to get maximum possible heat output until now. CoreDamage will do just that. Name one other program (publically available) that can match it???
Click to expand...

You don't consider TAT publically available?
 
Optional87 said:
your processor was 59C at idle? oh my god.
Click to expand...


I have an ultra-quiet PC; tucked away inside a computer desk with the door closed, in a sound proofed ATX case with foam on all 4 sides. Video card and motherboard chipset are fanless. Using a big Zalman copper cooler with a variable speed fan adjustment; at 83 degrees and 3.5ghz I can barely hear my PC. I can hear the faint crackling of my PC speakers over my fan. I like it that way.
 
old skool said:
I have an ultra-quiet PC; tucked away inside a computer desk with the door closed, in a sound proofed ATX case with foam on all 4 sides. Video card and motherboard chipset are fanless. Using a big Zalman copper cooler with a variable speed fan adjustment; at 83 degrees and 3.5ghz I can barely hear my PC. I can hear the faint crackling of my PC speakers over my fan. I like it that way.
Click to expand...

Way too hot in my opinion, but everyone has a different feel for it I guess.

I'd just lower my overclock at that point, settle for 3ghz or something, at the very least 1333mhz fsb (333).
 
Every computer I build, overclocked or not, I run 24 hours of Orthos
 
Mr. K6 said:
What's the purpose of pushing impossibly high temperatures through your CPU besides possibly damaging it? Real-world temps will never reach close to that and the purpose of the CPU is to generate accurate, quality data, not produce heat. I just don't see the purpose beyond testing the thermal limit of a heatsink, and I would not want to use a CPU as the guinea pig. FWIW, I use four instances of Prime x64 large FFT to test my max overclocks.
Click to expand...



Is this a rhetorical question?

The purpose is to find out if your PC is rock solid. Period. End of story. If you're computer can't handle the temps move over and get out of the way.
 
Dan_D said:
I'm partial to this utility myself. I also typically use Orthos but not if I'm working with quad core processors.
Click to expand...

I don't know if you were into the AOL scene back in the mid/late 90s, but OCCT reminds me of all of those AO-Proggies that VB scripters made to flood chatrooms with ASCII drawings.

To be truthful, I did use it on my latest QC build in my signature. I just don't feel 30 minutes is good enough, and I can see Prime95's 4 threads telling me exactly what is going on. I know OCCT can be run higher but I've been partial to Prime for at least 10 years now.
 
old skool said:
Is this a rhetorical question?

The purpose is to find out if your PC is rock solid. Period. End of story. If you're computer can't handle the temps move over and get out of the way.
Click to expand...
You missed the point: it has no use. Making your computer run under extremely high temperatures that will never be reached in normal computing, without a focus on testing its accuracy, is just a waste of time and an unnecessary stress on the CPU. As I said before, CPUs are made to produce accurate data, not heat. As an analogy, this is like making your car climb a 10,000ft. mountain before you decide to keep it when all you ever do is drive 30 miles to work on the highway; i.e. there's no point.
 
Mr. K6 said:
You missed the point: it has no use. Making your computer run under extremely high temperatures that will never be reached in normal computing, without a focus on testing its accuracy, is just a waste of time and an unnecessary stress on the CPU. As I said before, CPUs are made to produce accurate data, not heat. As an analogy, this is like making your car climb a 10,000ft. mountain before you decide to keep it when all you ever do is drive 30 miles to work on the highway; i.e. there's no point.
Click to expand...



Guess what? They do produce a lot of heat!! You better make sure you got the cooling taken care of!! So you have two choices; run Prime 95 for 24 hours and see how hot it gets, or run CoreDamage for 5 minutes.


Want another use? Try benchmarking heatsinks! The more heat you can subject a heatsink the better to test it out, eh?



Like I said, there are programs for testing the accuracy but barely anything that can test your heatsink.


Also, one thing you should remember is the heat generated by this program is all software; so this synthetic condition is actually nothing more than regular x68 optimizations that are present in many applications. In other words, you might see this kind of heat output in really small bursts during every day computing. So using your car analogy, this is like testing the car at its 7,000rpm redline for 5 minutes just to make sure the radiator doesn't blow even though you'll likely be cruising at 2,300 rpm most of the time and only ever hit redline when blasting down the freeway on-ramp once or twice a day.


Another thing; I fucking HATE car analogies. They are getting old.
 
I dunno what the app is doing but it did generate more heat than p95 v25.5 on my q6600 @ 8x333 and 1.15 V of vcore. I could reboot and do it @ 9x333 @ 1.2625 V but I suspect the results will be similar in order, just different in magnitude. There isn't any info on their website about what it's doing mathematically though.

P95 v25.5 (small ffts) does a better job equally applying stress on my machine though:

After 10 min of p95 v25.5: 51,52,51,51
After 10 min of core damage: 56,57,52,52

The other major knock I'll make against core damage is that it doesn't do anything helpful beyond just blindly stressing the cores. P95 for example will tell you if a rounding-error occurs during the tests which is very helpful when fine-tuning vcore settings as you well know. If stress testing with core damage, you'd have no idea beyond a reboot or blue screen.
 
CPU's don't produce a lot of heat, 100W is jackshit in the real world, and the heat produced is largely due to inefficiencies. It's always possible to run software to produce maximum heat, if you were using your computer purely as a space heater, then I could see a point of using this software to stress test. However, that'd be one VERY expensive space heater. Also, frying your CPU to test heatsinks is pointless. If you test heatsinks, you either test them to A) find the maximum stable overclock (which is indirectly related to heat output), in which you use a program that actually tests accuracy, or B) test maximum heat dissipation. Now one could say "oh, use this program for that." While you could, why would you risk a $200+ CPU when you can easily use a $10 heat block that does the same thing, and probably more accurately?

grasky brings brings up the same point I've been trying to make but evidently can't get across. I'm not trying to knock your program or new software, I think its always great to test new things, all I'm saying is if you're going to use it to stress test your CPU, there's better programs out there. I'd hate to see a threads show up saying "BSOD plz help!" because people thought their comps were stable by using CoreDamage, where it might have been thermally stable, but wasn't stable otherwise (accuracy etc.). One needs to use programs that test everything: heat, power, accuracy, etc.

Also, I used the car analogy to help get my point across but that didn't seem to work. Change it to boats/exercise phys/legos/whatever the hell you want if you care that much.
 
You must log in or register to reply here.
Back
Top