A very general question (stability test/heat issues etc. ?)

[flexy]

ok,

let's say someone does the usual memtest, OCCT, prime95 etc. tests.

I have various situations where i get the first error, say, after 30mins or so.

QUESTION (example: memtest)

If i run memtest #5 and i get the first error in run 20, cant i ASSUME that my errors are heat related ?

Assumption:
* if i have a very basic setting wrong, errors would come up in the first few runs ?

* *IF* i get errors after 30mins OCCT/memtest #5 etc...i probably might RATHER try to experiment w/ voltages (eg. decrease one notch) instead of fiddling with other settings like TRCD, TRFC etc...

Would like to hear some thoughts regarding this :)

[EMC2]

QUESTION (example: memtest)

If i run memtest #5 and i get the first error in run 20, cant i ASSUME that my errors are heat related ?

NO...

CPU errors (mem underclocked, testing max CPU clock with OCCT) will usually show up later in time when caused by temps, but memory errors can be any of a number of things that might only show up once every 50 to 100 passes and have nothing to do with temps ;) Could be noise on power, mem controller, timing, Vtt, or signal quality on memory, not just heat.

Best way I've found for finding best Vmem spot is raise till no further speed scaling is found using extra tight timings... then test at max error-free/infrequent-error speed with the tight timings, lower and raise Vmem by 0.1 to see if errors appear/increase till best 0.1 setting found. On DFI or with pot control of Vmem, then adjust 0.05 up/down same way. After that, then scale up till you find max FSB with timings you can live with/want. This way you eliminate a lot of other possible causes for errors other than Vmem.

Example - for TCCx find Vmem sweet spot with 1.5-2-2-5 timings. Then loosen timings to find max FSB with other settings adjusted.

Peace :toast:

[timpanogos]

Being new to this, one thing that somewhat surprised me was the speed/time it took for temps to jump from idle to load. Seems like about 2 seconds is about how long it takes my CPU temp to settle from idle to load levels (34C to 42C shortly after enter/ok pressed).

So, with temps stable after just a few seconds of load, not so likely it's simply temp after say 1/2hr of stable loaded temp

[flexy]

NO...

CPU errors (mem underclocked, testing max CPU clock with OCCT) will usually show up later in time when caused by temps, but memory errors can be any of a number of things that might only show up once every 50 to 100 passes and have nothing to do with temps ;) Could be noise on power, mem controller, timing, Vtt, or signal quality on memory, not just heat.

Best way I've found for finding best Vmem spot is raise till no further speed scaling is found using extra tight timings... then test at max error-free/infrequent-error speed with the tight timings, lower and raise Vmem by 0.1 to see if errors appear/increase till best 0.1 setting found. On DFI or with pot control of Vmem, then adjust 0.05 up/down same way. After that, then scale up till you find max FSB with timings you can live with/want. This way you eliminate a lot of other possible causes for errors other than Vmem.

Example - for TCCx find Vmem sweet spot with 1.5-2-2-5 timings. Then loosen timings to find max FSB with other settings adjusted.

Peace :toast:

yeah thats a good testing method - although i am almost 100% positive that my week 37 TCCD right now need 3.0V. (at my current settings). I must have one of the chips which like voltage - otherwise i cant explain.

I did some testing 302 FSB (a lilttle higher than my 289)

2.7/2.8 are a total NO-GO ! (cant even run memtest in bios)
2.9 flakey/many errors
3.0 less errors

The funny thing is, i wouldnt even be surprized if i get even less errors at 3.1/3.2Volts....but i really dont want to since my whole system is hot enough already and 3.0 is the max which i want to pump in those chips.

(The above findings also are in sync with my problems that i get BSODS and general flakyness if i use less than 3.0V VDimm ! If i use 2.9 i am OCCT stable for a while..but i can have BSODs out of the blue while i am not even doing anything...so i am very, very sure that my mem loves 3.0V, also didnt get any of those BSODs w/ 3.0V now)