Prime errors

Prime is one of the best stability testers, it is not the best but for testing it is okay.

24/7 folding is the hardest test that a PC can do.

Currently at 6 weeks up-time @ 100% load, no W/Us have failed, NOTHING.


I first tested my 2.6ghz XP-M using Prime95 which was stable for 24 hours @ 2.6Ghz @ 1.8V, yet to get F@H stable I needed 1.9V. It crashed at 1.6 and 1.7V therefore it told me what if it was stable or not.

Cyber-mav, most of your posts have been giving rubbish information, I read XS and I have read about Prime95, but as a indication of stability it is good to be used.

CR.

EDIT: I used S&M on my XP-M too and that past, none of these programs test all of the CPU at once, none of them.

What program uses ALL the instruction sets, FPU, Memory controllers etc? F@H uses most of them and when you start up a game with F@H in the background is how I test for absolute stability.
 
Last edited:
when using free testing utilities why settle for less than the best? snm is the best. then again i notice a lot of noobs that get into overclocking start out with prime just like i did because i did not know better. then eventually when you have a system like my old one which was prime stable for 35 hours at 2500mhz but crashed when playing a game (dawn of war) which is a very cpu intensive game.

i wasted weeks doing a reformat of the pc, prime tests which took forever, if i remember right i ran a 46 hour prime on that old pc which passed prime at 2500mhz. so i ruled the cpu out. played around with gfx card drivers and card overclocks/underclocks/relocked pipes etc. all to no help.

then i found out about snm. on my 46 hour prime stable setup snm crashed it in 7mins 26 seconds. was only snm stable at 2435mhz.

now i have been using prime for many years and used to swear by it but not after that day i saw how crappy prime was.

that was the day i shed all my noobness and stepped up into the real world of overclocking knowledge.

prime is excellent for hunting for prime numbers since thats what it was designed to do.

snm has been engineered from the ground up for the sole purpose of cpu testing.

have a look at the chart here: http://testmem.nm.ru/snm.htm at the bottm of the page.

the highest tdp reached by prime is 85% of a p4 cpus theoretical maximum. the chart shows that the cpu tested using prime gave off less heat than a few other programs.

snm maxes it out at 100% tdp so the cpu is actually running as hard as it can and thats why its better at stability testing.

the facts don't lie, only an uninformed noob or prime worshippers are the ones who lie.

do what you will, i have shown you the facts and the charts i don;t understand why people don;t try snm out, all you have to loose is your hardware if its not up to the job.
 
As far as i'm concerned, i have never known a pc that passed prime to become unstable or to mess up folding work units or anything, and to 99% of people thats what they need.
So as far as i'm concerned, a system that passes prime is stable, and thats stable enough to play games, fold 24/7 or anything else you might do with it.

And for that reason i will continue to recommend and use prime to check and ensure the stability of a system.

Nothing you will ever run will stress a system more than prime, not even folding, so why bother using something that puts even more stress on it.
 
None of them are, because none of them test the whole CPU. It is used as an indication of stability.

In my view Prime is up with S&M etc because it can run continuously, im not sure about S&M as I haven't used it in ages.

When I had my XP-M @ 2.6ghz, Prime and S&M said I was fine at 1.8V. But I started folding and it needed 1.9V to be stable, yet games etc were fine?

Then I used both on my 3Ghz San Diego, 1.595V was fine for both of them, 24 hours prime and full test of S&M yet F@H crashed until I reduced it to 2800 @ 1.55V.

Prime95 shouldn not be used as a complete indication of stability, or any other "testing" program, because they can be wrong. As the OP is having problems shows he is having issues and he needs options to sort them, S&M would give the same indications. As CPU voltage isn't helping im thinking either CPU maxed, mem issues or mobo.


When I get my Conroe rig in September ill do some tests, using Prime 95 for 24 hours, S&M and OCCT. If they all pass ill see if F@H does the same.

We will then see which is best for Conroe, as im 100% sure it changes from CPU to CPU...
 
Joe42 said:
Nothing you will ever run will stress a system more than prime, not even folding, so why bother using something that puts even more stress on it.
Click to expand...


because i played a game that was more stressful than prime and it crashed my pc where prime was 35 hours stable.
i have posted a link to the stability table if i was on my own pc i would have uploaded a full on screenshot of the table to show you that prime only maxes a cpu up to 85% at best. since prime is a constant load unlike games were things change all the time and there is a mixture of fpu and integer execution the stress varies a lot so only in certain games dawn of war for one which places a strong load on the cpu's fpu for all its calculations will crash.

majority of games are gfx card limited so cpu will never be really maxed out.

do remember that 100% cpu utilisation is not the same as 100% cpu load.
its a lot for beginners to understand but you need indepth knowledge of cpu power displacement and it just boils down to experience.

thats why i stopped testing my pc with prime because i wanted a system that was 100% stable and not just 85% stable.
 
Cyber-Mav said:
because i played a game that was more stressful than prime and it crashed my pc where prime was 35 hours stable.
i have posted a link to the stability table if i was on my own pc i would have uploaded a full on screenshot of the table to show you that prime only maxes a cpu up to 85% at best. since prime is a constant load unlike games were things change all the time and there is a mixture of fpu and integer execution the stress varies a lot so only in certain games dawn of war for one which places a strong load on the cpu's fpu for all its calculations will crash.

majority of games are gfx card limited so cpu will never be really maxed out.

do remember that 100% cpu utilisation is not the same as 100% cpu load.
its a lot for beginners to understand but you need indepth knowledge of cpu power displacement and it just boils down to experience.

thats why i stopped testing my pc with prime because i wanted a system that was 100% stable and not just 85% stable.
Click to expand...

Erm, S&M doesn't use all the CPU, games don't its pretty much impossible to use 100% of a CPU.

The only way would be to use:

33% Game for 3Dnow etc and any sections of the CPU which games use.
33% of Prime95 for FPU/memory tests
33% for rest of instruction testing, SSE/2/3/4/whatever

That is how I try to test, thats probably 95%? CPU load.

Constant load with 5/10 min breaks to allow it too cool down like F@H is a very very very hard test.

So if you say your system is stable, download F@H, get two sets of it going and leave it for a month on 24/7, It will probably crash, or fail some Work units.

CR.
 
F@H is useless what it can tell me in 2 months snm can tell me in 2 hours. i have used all these programs before and the best i have ever come across is snm.

on a side note im getting confused here are there 2 threads running about this here?
 
Erm, have you actually run it for more than a week? If you think its S&M stable is stable then your have another thing coming LOL.

On every processor so far none of the tests have picked up the instability than F@H has shown.

I can do 2.4Ghz at 1.5V with my X2 which passed S&M, F@H needs 1.55V at the same speeds? Also to run BF2 + F@H without crashing needs 1.55V? I've experienced this with 7 different CPUs in 4 different setups.

Now what does that tell you?

NONE of the stability programs is 100% proof that it is unstable, these errors could be hours or days apart, something that F@H shows.

Use it to your hearts content to test overclocked settings, but im pretty sure it doesn't show complete stability.
 
girls girls.. I think you're both getting a bit uppity about this.. prime stability is a level of stability, as is f@h stability, and various other levels of stability. Its a matter of personal opinion so neither of you are going to be right.

If you want 100% stability then you really need to be running at or very near to the manufacturers specs but as it is the OP asked for how to make his system prime stable which I guess means help with overclocking. Not an argument about the virtues of various stress testers.
 
matt100 said:
girls girls.. I think you're both getting a bit uppity about this.. prime stability is a level of stability, as is f@h stability, and various other levels of stability. Its a matter of personal opinion so neither of you are going to be right.

If you want 100% stability then you really need to be running at or very near to the manufacturers specs but as it is the OP asked for how to make his system prime stable which I guess means help with overclocking. Not an argument about the virtues of various stress testers.
Click to expand...

Well said. We could argue this until we're blue in the face and I doubt we would come to an agreement.

I myself used Prime95 for stress testing and found it to be a great tool. I am a folder and I wanted an overclock that was absolute rock solid. I ran Prime95 for just over 24 hours to help me simulate what my PC would be doing once I had settled on an OC.

While I agree S&M might throw up an error quicker than Prime95 it is only a 2 hour test. As previously said an error could occur hours or days later.

I primed my system for 24 hours and then moved onto two instances of F@H. I am glad to say that since Monday I have had no errors in any of the WUs I have crunched this week. :)

Having said all that, the next time I attempt an overclock, be it trying to push my current 4400+ past 2.6GHz or a new CPU I will probably run S&M as a preliminary stability test as it is very quick and stresses the CPU quite a bit. However, even if it passes I would still run a 24 hour stint of Prime95.

Now let us all just agree to disagree and see if we can't actually get back on topic - ie helping the OP with their OC problem :)

SiriusB
 
Cyber-Mav said:
i pity the fool who uses prime as a cpu tester.

if you dont believe me then go read up on the XS forums, those guys really know their stuff, and if you think prime is a better cpu tester than snm then your a fully qualified professional fool. nuff said. :rolleyes:
Click to expand...

what is snm fullname plz?
 
If you system is failing after 24 hours of prime, I would say that you need to lower your temps.
 
Got to agree with Cyber-Mav on this one. When your clocking it is very time intensive, so you need to get results as fast as possible. Whithout any shadow of a doubt THE fastest and most accurate way to test for cpu stability is with S+M. If it passes S+M then you can be rest assured it will pass Prime95, it ALLWAYS has done for me. If it fails Prime, ill lay odds on it that the problem is memory and not the cpu. As a side note although everyone uses Memtest for memory, it will only give an indication of good or bad memory/settings. If you want a diffinitive test of memory then use Goldmemory. It can and will fail ram or ram settings that would normally pass Memtest, which if you accept the Memtest results leads you then to fail Prime. That then leads you in the wrong direction, because you think it's the cpu and of course the cpu has passed S+M.
 
I've had CPUs pass S&M but not Prime. I find the fastest and toughest test is OCCT; if it passes OCCT it will always pass Prime in my experience.
 
You must log in or register to reply here.
Back
Top Bottom