Forget P95......

[THUGSROOK]

again IMHO:

you need to actually use the computer while running p95 or encoding divx.
play music, solitaire, whatever but use it.
results will not quite be the same as just letting those things loop/run.

my example:
while running prime95, other programs (notepad, netscape, winamp) are crashing left and right but p95 still chugs along ~ what does that tell you?

it tells me that i wouldve never known it was unstable if i didnt run those other programs while stress testing

thats right notepad crashed but p95 kept running! :Q lol

 

[Thor86]

Heh, notepad.

Ok, so looping P95 and using other software, and nothing crashes, no anomolies, does that mean stable?

 

[ChampionAtTufshop]

Originally posted by: Thor86
Heh, notepad.

Ok, so looping P95 and using other software, and nothing crashes, no anomolies, does that mean stable?

stable:
do whatever you normally do iwth all of those applications running in the background...and dont turn it off for like 2 weeks at least
let it stay on
even when you finish using it, let it stay on

if it crashes during that time period, something is not stable

 

[SinfulWeeper]

Originally posted by: ChampionAtTufshop

Originally posted by: Thor86
Heh, notepad.

Ok, so looping P95 and using other software, and nothing crashes, no anomolies, does that mean stable?

stable:
do whatever you normally do iwth all of those applications running in the background...and dont turn it off for like 2 weeks at least
let it stay on
even when you finish using it, let it stay on

if it crashes during that time period, something is not stable

Wouldn't that be the OS...

 

[THUGSROOK]

you are stable when:

your system can run anything you throw at it and not crash/error.
how far you want to take that, is upto you

 

[ChampionAtTufshop]

Originally posted by: SinfulWeeper

Originally posted by: ChampionAtTufshop

Originally posted by: Thor86
Heh, notepad.

Ok, so looping P95 and using other software, and nothing crashes, no anomolies, does that mean stable?

stable:
do whatever you normally do iwth all of those applications running in the background...and dont turn it off for like 2 weeks at least
let it stay on
even when you finish using it, let it stay on

if it crashes during that time period, something is not stable

Wouldn't that be the OS...

and the os is not stable becuase of either a hardware instability or a software instability
usually software instability is becuase of hardware instability

id rather throw a barrage of tests/programs at it and see if it crashs than run a few selective tests and deem it stable...
it would suck to be doing some work in ms word or whatever and have it crash andyou didnt save

 

[Mikki]

DAPUNISHER, yeah the -advmethods -forceasm switches definitely help get you the gromacs, I didn't get a Tinker in like 3 weeks with those switches. A lot of it depends on Stanford and what WU's they're releasing, but the switches are a must.

At some point a couple weeks ago I had to lower the fsb a couple mhz because my system kept locking up...the only thing I had running was the F@H console and EMIII, so I can verify what you're saying. Good info! I think Duvie has a good suggestion also about divx encoding, I've been seeing a lot of folks say that around the net. I'm still an amatuer with that stuff but I'm learning...

 

[Duvie]

As I am encoding and prime95 crunching I am usually on the internet talking to you ppl!!!!! Or I watch my TV and the war coverage.....So I think I am pretty much "mixing it up" to test stability....

I think we are touching on a really good topic and I think we are starting to realize some important issues that may revamp the way ppl test these cpus....Hardware prefetching as been discussed briefly above may actually reduce stresss on system if it does repetitive operations from what I am hearing. Any program like this may be looping becomes less stressful, right???

 

[DAPUNISHER]

Thanx Mikki! I was hoping someone else would jump in about the gromacs I also agree encoding video wether it be DIVX or DVD2SVCD will put a serious load on the whole system. I don't do it as a strees test at this point but I'm thinking it's time to change that based on Duvie's comments about it.

 

[THUGSROOK]

LMAO!
prime95 = 6 hours passed
memtest86 = 6 hours passed
Divx encode = crashed in 5 minutes
LMAO!

p95 is useless ~ i am removing it from my system

 

[DAPUNISHER]

Duvie's mantra seems to be the best stability suite going at the moment so that's what I'll be recommending for n00bs and folk what don't know yet. Unless of course I can sign them up to be a F@H soldier

 

[Duvie]

Originally posted by: THUGSROOK
LMAO!
prime95 = 6 hours passed
memtest86 = 6 hours passed
Divx encode = crashed in 5 minutes
LMAO!

p95 is useless ~ i am removing it from my system

I am sorry for you on that!!! But thank you for helping me prove my point ans my sanity...You know when I ran into this about 2 weeks back I was perplexed...I am glad to see some proof of this...


The real question is how many are running system (c1's) right now thinking they have stability on prime95 alone!!! It may only be an illusion....

 

[Shimmishim]

what's a good program to use for divx encoding?

yeah... memtest86 was running fine... then i tried playing warcraft iii thinking my system was stable... then BAM... game crashes... i see my desktop but can't click on anything...

this was an issue fixed by relaxing my memory timings...

 

[Duvie]

I currently know of 3 tests I ran on my system when it was prime95 stable for 5-6hours and 10+pass of memtest and it failed....

Autocadd 2002 rendering with all shadows and lights on.....failed in a 2 min test
Divx 5.02pro codec in gknot....failed not in 5 min test but failed within 15min into a 2pass 2hour encode test...
Besweet audio encoder.....failed as well near the end of the test...


I bumped vcore up a notch and it passed all test...

 

[Duvie]

Go to www.doom9.net and go into guides and pick Divx 5.0 guides and download the programs they use...

I use gknot which many think is quite difficult, but I like it cause everything is batched into one program...Maybe Thugs and I need to write a quick guide to set it up....It really only takes me a few minutes to get it ready to fly...

 

[ChampionAtTufshop]

id do it, but like 50 viruses decided to infect me so im out of commission for the time being lol
(oy vay, and all this hw to do too.....hmmm....)

 

[DAPUNISHER]

Originally posted by: ChampionAtTufshop
id do it, but like 50 viruses decided to infect me so im out of commission for the time being lol
(oy vay, and all this hw to do too.....hmmm....)

:Q who'd you piss off?

 

[zephyrprime]

IT DOES NOT TEST PREFETCH!

That's the least of it's problems. Besides, prefetching is auomatic on the newest processors (but not on the older ones like the P3/P2). I doubt prime95 uses SSE and it seems to me that SSE has the potential to produce the most heat becuase the data it works on is big. And does prime95 even use floating point code? Prime95 can't heat up processors nearly as much as synthetic tools like CPUBurn.

 

[Duvie]

Originally posted by: zephyrprime

IT DOES NOT TEST PREFETCH!

That's the least of it's problems. Besides, prefetching is auomatic on the newest processors (but not on the older ones like the P3/P2). I doubt prime95 uses SSE and it seems to me that SSE has the potential to produce the most heat becuase the data it works on is big. And does prime95 even use floating point code? Prime95 can't heat up processors nearly as much as synthetic tools like CPUBurn.

the only problem with that is the fact the B0 stepping chips had SSE, and prime95 was the best stability tool we could use....I could run divx for 4 hours on a system that would fail prime95 in 15minutes....With the B0 stepping chips the highest temp I would ever encounter would occur in prime95 and not divx...they would be close but prime95 could be 1-2c higher. the c1 stepping has been reversed on all fronts...the hottest temps have been in Divx and often 4-5c higher...The system could be prime stable for 12 hours and fail divx...

So sse was there then and c1 prefetching was not...that is the only change I know of so I think it has to be somewhere in there...

 

[Shimmishim]

Originally posted by: Duvie
I currently know of 3 tests I ran on my system when it was prime95 stable for 5-6hours and 10+pass of memtest and it failed....

Autocadd 2002 rendering with all shadows and lights on.....failed in a 2 min test
Divx 5.02pro codec in gknot....failed not in 5 min test but failed within 15min into a 2pass 2hour encode test...
Besweet audio encoder.....failed as well near the end of the test...


I bumped vcore up a notch and it passed all test...

could you???

thanks!

 

[THUGSROOK]

Duvie ~ since ive been lucky (?) enough to own many different c1s running at various speeds....

i can tell you this "bug" gets worse the higher up we go in mhz.

3.0ghz 150fsb ~ was pretty typical, all stress tests agreed with p95
3.2ghz 178fsb ~ things got flakey here, DivX forced an extra .025v outta me :| p95 was wrong
3.4ghz 179fsb ~ things are even more flakey, DivX wants an extra .075v :|

im absolutely sure its not a memory issue ~ ive thoroughly ruled that out.

<EDIT> and yes i am multitasking while encoding ~ right now in fact

 

[Duvie]

Originally posted by: zephyrprime

IT DOES NOT TEST PREFETCH!

That's the least of it's problems. Besides, prefetching is auomatic on the newest processors (but not on the older ones like the P3/P2). I doubt prime95 uses SSE and it seems to me that SSE has the potential to produce the most heat becuase the data it works on is big. And does prime95 even use floating point code? Prime95 can't heat up processors nearly as much as synthetic tools like CPUBurn.

I found this in the prime95 help section:

New features in Version 21.2 of prime95.exe



1. The program now uses the SSE2 instructions introduced on the Pentium 4 CPU. This version is about 3 times faster than the previous version on a P4.



2. The program now uses the prefetch instructions on the Celeron 2, Pentium 3, and all Athlon CPUs. This results in about a 20% performance boost on these machines.

So it appears to use sse2 instruction so I assume sse is also used...


New features in Version 21.4 of prime95.exe



1. Exiting the Torture Test now prints out how long it ran.



2. P4 error checking was relaxed slightly to reduce false alarms.



#2 is interesting!!!!


I run 22.1x version and under whatsnew.txt I see nothing important mentioned...

 

[THUGSROOK]

#2 is very interesting!

ive had programs crash all over the place WHILE p95 was running ~ p95 didnt crash tho :Q

 

[Duvie]

Originally posted by: THUGSROOK
Duvie ~ since ive been lucky (?) enough to own many different c1s running at various speeds....

i can tell you this "bug" gets worse the higher up we go in mhz.

3.0ghz 150fsb ~ was pretty typical, all stress tests agreed with p95
3.2ghz 178fsb ~ things got flakey here, DivX forced an extra .025v outta me :| p95 was wrong
3.4ghz 179fsb ~ things are even more flakey, DivX wants an extra .075v :|

im absolutely sure its not a memory issue ~ ive thoroughly ruled that out.

<EDIT> and yes i am multitasking while encoding ~ right now in fact

I totally concur with all that info...At 3.24ghz I needed 0.25v as well..At 3.33ghz I needed .05v...I agree and have concluded memory is not issue as I ran it at lowest timings and 1:1 ratio versus 4:5 ratio and same problem existed...

 

[Duvie]

Originally posted by: Shimmishim

Originally posted by: Duvie
I currently know of 3 tests I ran on my system when it was prime95 stable for 5-6hours and 10+pass of memtest and it failed....

Autocadd 2002 rendering with all shadows and lights on.....failed in a 2 min test
Divx 5.02pro codec in gknot....failed not in 5 min test but failed within 15min into a 2pass 2hour encode test...
Besweet audio encoder.....failed as well near the end of the test...


I bumped vcore up a notch and it passed all test...

could you???

thanks!

You actually meant about coming up with a quick guide for stability testing with divx, right???