Forget P95......

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

DAPUNISHER

Super Moderator CPU Forum Mod and Elite Member
Super Moderator
Aug 22, 2001
31,142
29,540
146
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...
I've been wondering what the ramifications of that feature would extend to since reading oldfart's post about it and I'd love to hear what pm or wingz have to say on the matter.
 

THUGSROOK

Elite Member
Feb 3, 2001
11,847
0
0
divx is pretty complicated ~ but for a stress test we could simply tell them what settings to use.

actually now that i think about it ~ im pretty sure AT would not allow such a thread
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
Lets revisit oldfart's thread........ Prefetch

Thugs that is a grey area but we could use possibly gknots ability to use avis and even mpeg sources..I haven't done any yet, but I now any raw avi which could come from a DV camcorder or video capture using ATi's MMC program.....

Hey if you own the DVD who cares, right??? All the ones you and I have done recently Thugs have been ours....I told you about 4-5 weeks ago Divx was only good for stress testing cause if anyone wanted to copy dvd's they would just get a burner and use many of the commercial product available in stores today...
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
Well we know it doesn't take a big file (just a clip or piece of a movie) then we can set all settings too high with a high bitrate and use 5.03's new nth passes and set it up to basically like run it 5-10 times....we know after 3 passes the visual quality increase is neglible but this would be clearly a stress test...
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
I think you are right there...I like the trailer to Gone in 60 seconds...It rocks!!!! best clarity I have seen in many of the trailers I have done plus the sound is encredible....Ofcourse stress test side of it about any would do but longer ones like Armagedon of lord of the rings movies can buy us a bigger file for longer encoding time...

I have had it pass my 5 min test but fail 15minutes in on the full encode...So I think a good 30min to 1 hour plus test neeeded...

Edit: Thugs lets continue the Divx tsress test thing in the PM's for now!!!

Lets leave this thread to the issue at hand!! Prime95 and it current releavence...
 

DAPUNISHER

Super Moderator CPU Forum Mod and Elite Member
Super Moderator
Aug 22, 2001
31,142
29,540
146
Originally posted by: Duvie
I think you are right there...I like the trailer to Gone in 60 seconds...It rocks!!!! best clarity I have seen in many of the trailers I have done plus the sound is encredible....Ofcourse stress test side of it about any would do but longer ones like Armagedon of lord of the rings movies can buy us a bigger file for longer encoding time...

I have had it pass my 5 min test but fail 15minutes in on the full encode...So I think a good 30min to 1 hour plus test neeeded...

Edit: Thugs lets continue the Divx tsress test thing in the PM's for now!!!

Lets leave this thread to the issue at hand!! Prime95 and it current releavence...
You guys have an excellent brainchild there! We will all bug you till the finished stress testing product is available so get busy gentlemen

 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
Originally posted by: DAPUNISHER
Originally posted by: Duvie
I think you are right there...I like the trailer to Gone in 60 seconds...It rocks!!!! best clarity I have seen in many of the trailers I have done plus the sound is encredible....Ofcourse stress test side of it about any would do but longer ones like Armagedon of lord of the rings movies can buy us a bigger file for longer encoding time...

I have had it pass my 5 min test but fail 15minutes in on the full encode...So I think a good 30min to 1 hour plus test neeeded...

Edit: Thugs lets continue the Divx tsress test thing in the PM's for now!!!

Lets leave this thread to the issue at hand!! Prime95 and it current releavence...
You guys have an excellent brainchild there! We will all bug you till the finished stress testing product is available so get busy gentlemen


We have some real issues to deal with.....

1) size (to make it worthwhile for ppl to download and use)...if we only need a guide and not a standard testfile set for them that wont be an issue

2) ease of use

3) Conflict of using possibly sensitive reference to programs that are often used to rip DVDs and violating copyright laws....even if we give a total legal use and tell them they need a dvd they own it could still be touchy...


 

DAPUNISHER

Super Moderator CPU Forum Mod and Elite Member
Super Moderator
Aug 22, 2001
31,142
29,540
146
Yep, that's a real catch22 :frown: I have forums you could post the info in and I and others could host FTPs with whats need as well but how to link peeps to it without it technically violating the wares rule here???
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
We are looking at using programs and possibly test files that should be accepting...freeware and video that doesn't violate things...however the file and download could be big....

Hell if we have to we can come up with a guide and pass amongst us and make no mention of it in the threads...anyway we look at it, something new is needed and many may be floating around with unstable systems with a false sense of security with prime95 and memtest success...I admit I have advise ppl to use them but I have usually said to run some divx or other real world test....
 

RaynorWolfcastle

Diamond Member
Feb 8, 2001
8,968
16
81
We even bother with encoding a DVD? Anyone with a DV camera could just record a snippet with a lot of camera movement (difficult to compress well) and save it in the appropriate format. Then you couldn't possibly have any legal issues.

Just a thought
 

Shimmishim

Elite Member
Feb 19, 2001
7,504
0
76
that's an idea ...

or what duvie said...

we could just PM each other with a link to it and pass it amongst ourselves as to not break any rules...
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
Originally posted by: RaynorWolfcastle
We even bother with encoding a DVD? Anyone with a DV camera could just record a snippet with a lot of camera movement (difficult to compress well) and save it in the appropriate format. Then you couldn't possibly have any legal issues.

Just a thought


We are looking at it....Virtualdub program offers many types of input files but avi would be the best...AVI files can be large so that may be an issue...

 

DAPUNISHER

Super Moderator CPU Forum Mod and Elite Member
Super Moderator
Aug 22, 2001
31,142
29,540
146
Well if it gets the mods approval I have no doubt enough of us can host it or exchange it through IM for those with broadband, and the dial-up guys could get a CD in the mail??? Otherwise the underground is simple and easy and acceptance might follow later. I do also strongly believe a new standard is needed to help more people evaluate a system's stability and avoid the irritating and potentially fustrating experiences a crash causes especially when they thought it was stable 'cause P95&memtest said it was
 

THUGSROOK

Elite Member
Feb 3, 2001
11,847
0
0
obviously at this point certain programs will surface with complaints of instability, we should keep tabs on what those programs are.

in the mean time Duvie and i will attempt to make something "simple" and "legal" for divx encoding.

we will also be searching for other "simple" stress programs with good accuracy.
since our systems are pretty thoroughly tested it should be a piece of cake to evaluate new software.
 

pm

Elite Member Mobile Devices
Jan 25, 2000
7,419
22
81
Prime95 is far from useless as a stability test. The programmers have done a very good job of making sure that it makes the most of the available CPU resources and thus it is capable of raising the core temperature of the device to a very high level. Considering it was never intended to be a stability test, but is supposed to finding prime numbers, it is surprising how well they are able to stress the execution units of most CPUs.

That said, I don't think that Prime95 is the ultimate test of system or CPU stability. The fact is that there is no one program that is capable of being the best test of all CPUs to determine whether it will be stable under all circumstances. The best test for one CPU may not be for another of even the same family - let alone one from a different company or an older generation.

At Intel the way CPU's are tested involves a very large, extremely expensive machine such as an Agilent Advantest, or a Schlumberger S9000. You load the CPU into one of these machines and then run literally thousands of programs on the CPU at various temperatures and voltages. When one looks at the output of one of these testers, you find that CPUs from the same wafer, or even side-by-side on a wafer, fail on completely different programs.
 

stippix

Junior Member
Mar 28, 2003
4
0
0
Prime95 originally is a program to find Mersenne prime numbers.
In order to do so it has to compute absolutely correct values
using integer and floating point units, which is very stressfull for the entire CPU.
So there is an option called "torture test" which loads the CPU with heavy FFT tests
until you end the program - or it stops itself with an error message
giving you a hint what probably went wrong.
The results returned are checked for correctness.
If you run the program for for 12-48 hours without any errors then you can be sure
that there is hardly an other program that will be able to kill the CPU.

You can restart Prime95 every 8-12 hours because sometimes errors only occur
when programs are started on a heavy loaded (and hot) CPU -
e.g. a good test is to execute TaskInfo 2002 or 3dmark2001SE
just to see if it starts without errors - after that you can close it again (TaskInfo 2002)
or go for a 2 hours 3dmark2001SE bench if you like.
But remember if you start other programs Prime95 will pause because of its low priority.

CPU Burn includes six different benchmarks which are really going to toast you CPU -
especially BurnP5/P6, resp. BurnK6/K7. This programs created the maximum
ever measured temperature on my duallie system.
BurnMMX/BX are good memory stressers.
BurnPx/Kx are not checking the results - so you will not get error messages
if anything goes wrong (dito for memory tests).
In case of failure these programs simply stop running - if has not stopped after
12-48 hours everything should be okay with your system.

CPU Burn-In is a benchmark program which runs for a given time stressing your CPU.
You can enable or disable result checking - disable will toast your CPU a little bit more,
because the checking routines seem to cool down the CPU.
The stress put on the CPU seems to be less than reached with Prime95 or BurnPx/Kx.
In case of success the program gives a statement -
in case of failure the program simply stops executing, too.

It can/will be a good choice to start several instances of the stress tests,
e.g. running Prime95, CPU Burn/Burn-In twice can maximize the stress put on the CPU.
It is recommended to do so if you have a multiprocessor system, because every CPU
in your system should be stressed to the maximum ever possible.

Sandra is a program giving you all the information about you system you ever wanted to know.
It contains stress tests for the CPU, memory-, AGP/PCI-busses, harddisk, CD/DVD drives
and much much more ...
While each test is very good to configure so that everything you want to be tested can be tested
it simply lacks of real stress while running these tests - the CPU for example stays really cold
(up to 10-15 degrees) in contrast to Prime95 or CPU Burn/Burn-In.
If the tests succeed you get graphical results - if something goes wrong
the test executing simply freezes (sometimes the complete Sandra package will do).

Memtest86 is a simple but good memory tester - but it is only a tester.
It detects RAM errors, but it will not put any specific stress on you memory modules.
Run it if you think that you RAM may cause problems - but do not expect Memtest86
to find problems related to overclocking your memory.
For this case the are better programs (BurnMMX/BX or specific Sandra cache/memory tests).
You should run Memtest86 for at least 6-18 hours, so that it can do several rounds.

Very popular stress testers are programs like seti@home, folding@home, genome@home etc.
These programs heat up your CPU very well, but you have to remember that these programs
do not do any result checking - so your CPU may work and not crash, but you do not know
if the results computed are correct.

If you want to test your (overclocked) system you have to use programs checking the results
(for example Prime95).

That is also why 3d action games like UT2003 or BF1942 are not the best stress testers available -
of course they really heat up your CPU and toast your graphics card -
but are you sure you will see incorrect computed results or pixels for example?

Nearly the same holds for DVD encoding/decoding. It is really stressfull for your CPU,
but if the programs used do not quit because of an heavy error occuring during the coding,
you will only see that something went wrong afterwards if your video behavior/quality
is not what you expected.

Of course there are so much more stress testers out there and probably there will ever be
a program crashing your entire system, but you have to remember the most important thing
about stress testing:

If the programs you want to run every day for your normal work or fun run correctly
without any glitches, errors or unexpected results - then everything should be okay -
for you and your system!
 

THUGSROOK

Elite Member
Feb 3, 2001
11,847
0
0
If the programs you want to run every day for your normal work or fun run correctly
without any glitches, errors or unexpected results - then everything should be okay -
for you and your system!
....until you try a new piece of software or game and it cr@ps right out on ya
also, overclocking beyond actually stability is like asking for SNDS. (if you overvolt)

You can enable or disable result checking - disable will toast your CPU a little bit more,
because the checking routines seem to cool down the CPU.
interesting ~ could this be the key? (no error checking)
 

jhites

Golden Member
Mar 19, 2000
1,854
0
0
Well now - how about these new icons
IMHO - Thought I would throw in my 2 cents.
You can enable or disable result checking - disable will toast your CPU a little bit more,
Interesting Quote - makes one think about alternatives.
I still consider Prime95 to be an exceptional program for testing the stability of the processor due to the fact of comparing known results to calculated results. I think stippix and pm covered the reasons why I value its determination.

I will say that I don't turn everything else on my system off and just run it wishing no-one will breath on the system. I also do not consider that my system is stable by just running prime95 for 3-5 hours. Once I think I have found the sweet spot, I leave all the background crap runnning and continue to use my system as always for a couple of days. Since my system is on 24/7, I will not reboot (unless it does it for me ) which of course means something is amiss.

I have and do run other programs while Prime95 is running its torture test. From one of the posts above maybe this could be allowing the processor to take a break but I don't think it would impact the end calculation. I check prime every so often while doing other tasks to ensure that it is still running. After dropping the fsb on my chip this weekend to 181 from 185 and dropping the vcore, I did capture some 8mm analog tape using Pennacle Studio 8. Created an mpeg and prime was still calculating numbers. I did originally start this using default 1.525v and got prime error. Kicked it up to 1.575v (1.616v actual) and all was fine again.

We all know that Sandra is a nice little bench compare and information program but that is about it.
How many times has Sandra run fine and then we could not complete 3DMark2001?
How many times have we completed 3DMark2001 and could not complete PCMark2002?
And last but not least. How many times have we run all of the above and got Prime95 error?

I don't run DivX so can't justify it one way or another. I can say that I have never had a problem once I had determined to be Prime95 stable. Prime95 seems to be all that I need to feel safe and stable. I do run some finacial programs that calculate tons of numbers (Metastock). Probably, for many users and gamers they might not even need to be 100% Prime stable for their system to serve their purpose.

You just gotta love overclocking for the shear pleasure of self satisfaction
Here everyone - :beer: - Have one on me.
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
The fact is something is different!!!!! prime95 ran fopr 6-12hours back with my old athlon, and b0 stepping p4's was the defacto stability test and then some...If it passed then I never had a crash (not related to ati driver updates )....


Differences I have noticed.....B0 stepping versus c1 stepping....

1) Prime 95 passes 6-12 hours no issues ever.......now.....12+ hours pass could still crash 2 out of 3 of my test programs with 5-10min....

2) Prime95 used to achieve hottest temp on average of all programs....now....it is about 4-5c lower then hottest temp achieved with Divx encoding and besweet audio encoding....


3) Divx errors in an unstable (prime would fail in 15min) system could pass but may have a few slight anamolies (nothing big) but no errors.....now.....Divx when it passes has no anamolies whatsoever yet will just error out and stop when at speed and vcore prime95 says is ok.....



Sidenote:

Prime95 at default is checking for errors and should have no relaxed checking going on.....I can get rounding errors at 3.24ghz with 1.55v so I know it is checking versus known numbers......



3.24ghz w/ 1.55v = prime95 passes 1 hour then fails (rounding) = besweet crashes = Divx crashes = autocadd rendering crashes = 3dmark passes = 3dmark2k1se loop 1hour with prime fails

3.24ghz w/ 1.58v = prime95 passes 12+ hours = besweet passes = Divx crashes = autocadd rendering crashes = 3damrk2k1se passes = 3dmark2k1se loop 1 hour with prime passes

3.24ghz w/ 16v actual = prime passes 12+ hours = besweet passes = Divx passes (mult test and reruns) = autocadd passes...3dmark2k1se passes = 3dmark2k1se loop 1 hour with prime passes


That is all that has to be said...No one can claim autocadd is not a stable program cause I have never had the renderer fail -in the last 2-3 years I have oc'd and worked with autocadd 2000-2002....
 

THUGSROOK

Elite Member
Feb 3, 2001
11,847
0
0
exactly the same here Duvie.

ive been encoding DivX for over a year now ~ and something has definately changed while using these C1s @ 3.1ghz+.

in the mean time im gonna be cheesy and try to find other encoding software.
as for you, you have multiple programs crashing ~ so the "problem" is definately real.

but we still need better ways to test this "new problem" since P95 doesnt cut it anymore.

 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |