prime 95

Overlag · 29 Jan 2006 at 12:17

YooEntSinMeROYT said:
I've found S&M doesn't seem to agree with having 2gb RAM.

Stress Prime works fine for me, along with multiple passes of memtest from within my BIOS.

works ok here

w3bbo · 30 Jan 2006 at 07:07

I always found S&M to be more 'intense' (higher heat) but P95 to be more thorough i.e S&M passed P95 failed. I usually run S&M first...if it passes then I leave P95(Large fft) running overnight.

chopchop · 30 Jan 2006 at 11:06

w3bbo said:
I always found S&M to be more 'intense' (higher heat) but P95 to be more thorough i.e S&M passed P95 failed. I usually run S&M first...if it passes then I leave P95(Large fft) running overnight.

large ftt doesnt stress the cpu as much as small ftt's. large is for ram really.

w3bbo · 30 Jan 2006 at 18:45

In place large FFT - maximum heat, power consumption some ram tested.

I will have already tested the CPU with S&M that stress the cpu more than prime does so no need to do it again...might aswell test 'system stability' with large FFT's.

Reason I do that is that if S&M passes I usually get a small fft pass aswell yet Large FFT throws up some errors that small fft doesn't seem to catch as its proably memory orientated as you say. If the CPU was unstable it would also fail large FFTs aswell as small fft.

Just my thoughts.

chopchop · 31 Jan 2006 at 03:21

w3bbo said:
In place large FFT - maximum heat, power consumption some ram tested.

I will have already tested the CPU with S&M that stress the cpu more than prime does so no need to do it again...might aswell test 'system stability' with large FFT's.

Reason I do that is that if S&M passes I usually get a small fft pass aswell yet Large FFT throws up some errors that small fft doesn't seem to catch as its proably memory orientated as you say. If the CPU was unstable it would also fail large FFTs aswell as small fft.

Just my thoughts.

well, with memory set too high, i can still pass small ftt's. but with memory set normal and cpu unstable i can pass large ftt's.

so u really have to set it to what your testing, ie ram or cpu. blend does that, but doesnt run each test for long enough. so i normally use 1024-4096 30mins 900mb for ram and 8 - 512 30mins 900mb for cpu.

w3bbo · 31 Jan 2006 at 07:14

I tested this out again to satisfy my own curiosity.

The settings I know to be stable (24hrs p95 small&largefft) using standard torture settings:

Opty 165 [email protected]
2gb Corsair 3500ll 3-3-3-8 1T

Test 1 small fft (dual Prime)

Opty 165 [email protected]
2gb Corsair 3500ll 2-3-2-6 1T

Failed after 40 mins (one core). The only settings I changed were to tighten the memory to 'stock' settings.

Test 2 Large fft (dual prime)

Opty 165 [email protected]
2gb Corsair 3500ll 3-3-3-8 1T

Failed after 3 iterations on one core, settings changed this time were cpu only (raised fsb by 11).

So it looks as though BOTH tests test both components (CPU&memory) to a certain extent otherwise it would not fail. Unless the chipset (K8->NB/SB-NB) have some bearing on the matter (these were not altered btw). Could also be a memory controller fault on the cpu...bah too many 'possibles' :confused:

Do you really think 30 minutes is long enough to test your components? I have had prime fail after 14 hours although I personally consider my PC stable if it passes 8 hours of both tests. Never really tried altering the test parameters.

Matt82 · 31 Jan 2006 at 07:25

i find it best to open prime, then click on to test 4 (custom) and reduce the amount of ram used).

if you have a gig of ram it will allow a gig to be used (but of course windows is running plus a million other things) so i reduce the amount of ram used by 200mb or so.

test performs exactly the same, but it allows me to run my comp

w3bbo · 31 Jan 2006 at 07:59

Yeah when I test some 'new' settings I just run one version of P95 (although standard settings) to test the weaker '0' core as that always fails first which leaves the other core to run windows. Wanna go home now and see if altering the amount of memory to use makes a difference to stability

chopchop · 31 Jan 2006 at 10:58

w3bbo said:
I tested this out again to satisfy my own curiosity.

The settings I know to be stable (24hrs p95 small&largefft) using standard torture settings:

Opty 165 [email protected]
2gb Corsair 3500ll 3-3-3-8 1T

Test 1 small fft (dual Prime)

Opty 165 [email protected]
2gb Corsair 3500ll 2-3-2-6 1T

Failed after 40 mins (one core). The only settings I changed were to tighten the memory to 'stock' settings.

Test 2 Large fft (dual prime)

Opty 165 [email protected]
2gb Corsair 3500ll 3-3-3-8 1T

Failed after 3 iterations on one core, settings changed this time were cpu only (raised fsb by 11).

So it looks as though BOTH tests test both components (CPU&memory) to a certain extent otherwise it would not fail. Unless the chipset (K8->NB/SB-NB) have some bearing on the matter (these were not altered btw). Could also be a memory controller fault on the cpu...bah too many 'possibles'

Do you really think 30 minutes is long enough to test your components? I have had prime fail after 14 hours although I personally consider my PC stable if it passes 8 hours of both tests. Never really tried altering the test parameters.

yea they will both test it to some extent, but im on about when your so close to being stable, eg my axp will do 12x200 1.85 actual, prime pass on both small and large. 12x202 fails on large only, because memory wont do 202. 11x202 fails too! 12x200 1.825v only fails on small ftt's, large will pass!