new program for testing stability [Archive]

View Full Version : new program for testing stability

flexy

01-03-2005, 10:20 PM

http://www.ocbase.com/downocct.html

http://translate.google.com/translate?u=http%3A%2F%2Fwww.ocbase.com%2Ffaqocct. html&langpair=fr%7Cen&hl=en&safe=off&c2coff=1&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools

the page is in french, but you can always use google to translate.

The program itself is in english.

It LOOKS like a pretty good program similiar to prime95 since it calculates and compares calulations to an internal table. It works together with Motherboard monitor (mbm) and shows voltage rails and temps while its testing. Also, it allows uploading results to a database and has nice graphical statistics.

It might be worth looking into it.

I personally would be interested in how this program behaves for all those people with A64 winchesters which fail (eg. at stock). Does this program give problems too ? Play around with the options :)

Addendum: We REALLY, really need to differ between programs which only run and put heavy load on the CPU - and programs which calculate and compare the calculated results to fixed, known internal results.
ANY program which just "utilizes" the CPU but does not validate calculation results (IMHO) is not suitable for accurate testing of stabillty.

Thats why i think this program might be interesting - especially with all the recent problems with prime95 and some AMD winchesters.

---

Edit:

http://216.239.37.104/translate_c?hl=en&ie=UTF-8&oe=UTF-8&langpair=fr%7Cen&u=http://www.bel-hardware.be/cpustests.html&prev=/language_tools

has a nice review/listing of all the latest tools for stability testing,

- OCCT 0.91
- Prime 95
- Super Pi 1.1e
- Toast
- CPU Burn-In
- CPU Burner
- CPUStabTest
- Hot CPU To test Pro 4.2.3
- BurnInTest

as i mentioned above, ALL programs except Prime and OCCT are *worthless* since they only burn/bench the CPU.
Only prime and OCCT can give you a result whether your CPU is actually stable or not - i see TOO MANY people here on the forum using eg. SuperPI and making an assumption about stability - but they cant with those programs.

Thats why i am still against some recent opinions like "Prime is the most overhyped program, just dont use it" etc..... but Prime/OCCT is the ONLY program which allows us to make a somewhat reasonable statement about stability.

Read the review at above link. (page translated with google).
OCCT btw. comes out as big winner there with its many features and it was the only program (together with prime95) to detect instability in the tests there.

flexy

01-03-2005, 11:01 PM

quickstart for OCCT:

"test" starts a 30mins logged test with nice graphics at the end.
"torture" is an infinite test.

With options you can set max. temps....eg. my test run just aborted because my CPU reached 54 degrees (still on AMD XP 1900+).
Thats a nice safety feature if you're afraid to burn your CPU.

you can also set options like task-priority and amount of memory to test.

For the temp/voltage readings you should either have MBM, speedfan or the asus probe installed.

DevilsRejection

01-03-2005, 11:18 PM

so what you telling me is occt is better than prime 95.

but prime has been proven day in and day out for it's stability. the winnie thing just proved AMD has a botched batch of prcessors.

halcyon

01-03-2005, 11:21 PM

Why do we need a new program to cover what Prime95 does?

Prime95 is ok.

The computers failing it are not.

Don't blame the messenger.

Also, S&M 1.3 is also superior in heat generation and it calculates accuracy as well.

DevilsRejection

01-03-2005, 11:23 PM

why doesn't xs do there own review? i mean we have some pretty hardcore crazy members here.

stealthbomber

01-10-2005, 03:49 PM

Any rig that runs Prime95 24/7 (I'm talking weeks/months here) is verified good enough (in terms of non stop long floating precision calculations) to run scientific apps like folding/UD/SETI. It's the only way to account for ambient/system/CPU temp and mains/PSU/system voltage fluctuations, either one of those variable can cause the system to miscalculate.

S&M seems to be coded in such a way that deliberately heats up the CPU like no other program I've seen (including various scientific apps), any system that passes Prime95 24/7 but fails S&M is likely due to the extra heat created by S&M rather than the hardware configuration.

IMO, S&M is a pretty useful program for 'worst case scenario' testing, like the ambient room temps going through the roof which affect both air and water cooling... :D

nealh

01-18-2005, 09:06 AM

seems to me what needs to be done is compare Prime95 to OCCT..ie if a system fails OCCT does it fail Prime95 and vice versa

WiCKeD

01-29-2005, 06:08 PM

I've had IE windows and other programs crashing all around me while Prime was still running without problems, so the more tests the better. Other than that whole French thing, this program sounds more practical than Prime. Best news is that the results are uploaded to a database, so know one can BS and say their system is stable. :cool:

chile

01-29-2005, 11:20 PM

anyone using this? it says it's beta...just wondering what kind of results everyone's getting...

Mastakilla

03-06-2005, 02:50 PM

trying it now

prime failed at these settings after hour, curious what occt will do

edit:
hehe

it just failed

after 4 mins allready

guess ill try this again ;)

charlie

03-16-2005, 09:10 AM

i hate p95, I'll give this a shot!

C

flexy

03-22-2005, 08:37 PM

hey people, every day i love OCCT more and more...

I just run a few tests with prime and prime (large ffts since i always want to test some ram) failed after about 55mins. Under the same config OCCT already complained after 2 mins 47secs !

In other words: OCCT is MUCH more sensitive than prime....if OCCT runs in the normal "test" mode for 30mins or say 1hr or 2hr without complaining you can assume you have a pretty everyday stable machine.

If you want more assurance you can of course always run "torture" in OCCT and run as long as you want. Btw. i set it to test 512MB ram of my 1gig.

Carlos Henrique

03-23-2005, 06:21 AM

redgoo

03-28-2005, 11:22 PM

This is my new favorite stress tester - much more sensitive than prime for immediate testing.

Cossey

04-15-2005, 02:38 AM

occt is great for stress testing in a hurry, i really cba to wait 24+hours when 1-2 will do.

deltarealm

05-03-2005, 02:12 PM

Is it ture this will do the work in faster time.
Like running prime for 24 hours+ is a pain in the ass, I don't have that kind of time to be loading up on my pc when I have other things to do.

With OCCT I noticed with my P4 HT cpu, its says 100% load but task manager reports 50%.
Do I have to run to instants of this program just like prime95

vaio

05-03-2005, 03:23 PM

Is that not normal for a hyperthreading chip?
Me not a pentium user.

vaio

05-03-2005, 03:59 PM

Stress testing now......makes a change from prime :D

THunDA

05-10-2005, 08:32 AM

Another vote for OCCT .. :)

Ive also noticed that Prime will fail in 2-3 hours but OCCT will fail in 5-8 min's. Also I like the 30min test very much.. its a great quick way of testing stability and then following up with prime and other stuff..

deltarealm

05-10-2005, 01:04 PM

This program is no good for intel right? only get 50% load on one logical unit

oasked

05-10-2005, 11:36 PM

Wow, this is a great program.

Very quick and no Winchester bug! Nice. :)

Kosmic71

05-13-2005, 03:43 AM

I will give it a shot, thanks for the info Flexy. :toast:

Overlag

07-24-2005, 07:39 PM

would be nice if they edit this to allow it to use both cores or HT

what i do is run S&M burn on one core, and OCCT on the other for a few hours, then switch it...

eddieate

07-25-2005, 12:37 PM

CPU-Burn in claims to check results too;
The resultant calculations are constantly checked for data integrity. If the program detects erroneous data the user is immediately informed
Ed

[XC] moddolicous

07-25-2005, 01:08 PM

Are there any DOS based burn in utils other than memtest.

W0nderb0y

07-28-2005, 07:24 PM

yea, they have p95 and cpu-burn in on the ultimate boot cd, they use freedos i think.

IluvIntel

07-30-2005, 08:37 AM

I'd be careful of using a beta version for OCCT, the link posted in the first part of this thread says it is the beta version and v0.91, use with caution.

I'll stick with Sp2004 for now untill the final release of OCCT comes out.

IYP

07-30-2005, 03:53 PM

hows it work?

metro.cl

07-30-2005, 04:11 PM

might try occt, prime is not doing very good for me

Mr_Slinky

08-10-2005, 09:45 PM

i would also like to know what ocz or g.skill reps think about this?

Zeus

09-18-2005, 12:14 AM

Is there a way to get this nice prog to display the temps and voltages in conjunction w/ smartguardian?

Marlowe

09-25-2005, 07:07 AM

I think this prog is great. Prime95 stable with dothan is no guarantee for general stability, so you must run prime and 3dmark at the same time for really testing. This prog detects errors much faster. Had it supported Smartguardian as mention above it would be even better.

However I have a problem.. I don't get the graphs and stuff after running the 30 min test.. The prog doesn't hang, I just get this displayed for eternity:

http://img.photobucket.com/albums/v205/graastein/occt.png

winguy

12-11-2005, 12:48 AM

Here's the corrected url:
http://216.239.39.104/translate_c?hl=en&ie=UTF-8&oe=UTF-8&langpair=fr%7Cen&u=http://www.bel-hardware.be/Cpustests.html&prev=/language_tools

So do you guys just run the 30min Stability Test at default/unchanged settings or the infinite Torture Test (and for how long)?

catkicker

12-17-2005, 10:25 PM