besides prime95, memtest, and intel burn test, what else to test stability?

[Maximus96]

what else should i use to test the stability of my new i7-920/p6td combo? right now i have it on 3.8 @ 1.27v with about 68 degree C under load.

 

[Udgnim]

can loop game benchmarks if you feel like it

Crysis, Resident Evil 5, and Street Fighter 4 have downloadable benchmark tools. can also include 3DMark 2006 / Vantage.

 

[brandonwh64]

intelburntest and LinX is about the best so let them run about 20 runs and you should be good.

Is that I7 920 a D0 or C1/C0?

 

[Maximus96]

I will try the 20 loops. Thanks. It's a d0.

 

[ShawnD1]

You can always try running the OCCT PSU test to make sure the system works under absolute full load. It's probably a 1 in a million type thing, but I guess it's theoretically possible for the CPU to start screwing up if a fully loaded GPU bogs down the PSU enough to cause a bunch of voltage ripple.


Fun thing about the PSU test is that I can cause massive ripples in the output graphs by turning a vacuum cleaner on and off. My dad's an electrical engineer and he jokingly said I should try doing that just to see what happens :awe:

 

[zerocool84]

Do those then try to run something demanding like an encoding program or an intensive game. Sometimes your oc will do fine in the tests but crash during a program or game.

 

[Maximus96]

at 3.8ghz (20x190, or 19x200) and 1.27 vcore, prime95 passes for 12+ hours with temps around 68 C. however, intel burntest fails randomly under the "maximum" test, sometimes after 15 runs, sometimes after 2 runs. "standard" test passes. temps never gets higher than 68-70 C.

i then tried to lower it to 3.4 ghz (19x180) and upping vcore to 1.325, burntest at maximum still randomly fails with the temp even lower than before.

any ideas? i've yet to run memtest because i don't have a dvd drive hooked up, will do that tonight.

 

[tweakboy]

Quit fiddling with your nice rig, no need to abuse the CPU with 10 hour runs,, thats abuse and CPU will live shorter,, gl,

 

[Billb2]

SuperPi 32M

 

[MrK6]

at 3.8ghz (20x190, or 19x200) and 1.27 vcore, prime95 passes for 12+ hours with temps around 68 C. however, intel burntest fails randomly under the "maximum" test, sometimes after 15 runs, sometimes after 2 runs. "standard" test passes. temps never gets higher than 68-70 C.

i then tried to lower it to 3.4 ghz (19x180) and upping vcore to 1.325, burntest at maximum still randomly fails with the temp even lower than before.

any ideas? i've yet to run memtest because i don't have a dvd drive hooked up, will do that tonight.

Interestingly enough, I have had great CPUs fail Intel's Burn Test at stock settings. I really don't trust the program, to be honest. Same with LinX. I'll use Prime95 for 10 hours or so on Large FFT's to verify any ballpark guesses I've made and access thermal management. Beyond that, I just use a rig as normal. I've had CPU's overclocked using this method run F@H without error for 120 days straight with no reboots.

 

[Kenmitch]

Well I guess after you have done all of those you could maybe switch to the more advanced testing

Maybe something like while stress testing shake your case violenly to simulate an earthquake, tornado, typhoon, etc....Would eliminate the possibility of lose internal parts.

Just kidding

Your system should be stable....Windows providing of course!

 

[EarthwormJim]

at 3.8ghz (20x190, or 19x200) and 1.27 vcore, prime95 passes for 12+ hours with temps around 68 C. however, intel burntest fails randomly under the "maximum" test, sometimes after 15 runs, sometimes after 2 runs. "standard" test passes. temps never gets higher than 68-70 C.

i then tried to lower it to 3.4 ghz (19x180) and upping vcore to 1.325, burntest at maximum still randomly fails with the temp even lower than before.

any ideas? i've yet to run memtest because i don't have a dvd drive hooked up, will do that tonight.

I'm curious what are your benchmark results in Intelburntest? Are you getting really low values? Like below 40gflops?

If you're getting a rather low gflops rating, that means the test is not really running at it's fullest. Could be why you can randomly pass or randomly fail.

 

[Maximus96]

i remember the maximum test tooks about 200+ seconds and about 50something gflops. i don't know how to read those results. i was just looking for the other number to be same as the rest.

 

[EarthwormJim]

i remember the maximum test tooks about 200+ seconds and about 50something gflops. i don't know how to read those results. i was just looking for the other number to be same as the rest.

50 something is in line with a good number. The results will vary depending on the work size.

 

[ear8dmg]

S&M: The best stress tester I know of. It's as intensive as IBT but breaks down the results by the components of a CPU. It makes it really easy to troubleshoot errors if you get them.

Just a warning - it can make your CPU a lot hotter than other stability tests. I'd recommend an eye on temps and a finger on the stop button for the first run (especially FPU test).

 

[kalrith]

Furmark or ATItool's artifact test