Now that we know P95 and .09nm AMD procs don't like each other, what app to use now?

[DevilsRejection]

I ask because prime95 has beeen the holy grail for me when it came to overclocking and testing for stability. All my systems I have ever owned have passed 72 hour prime95 priority 10.

Now that we can rule out this app working with whinny's then what is the other alternative? :confused:

[[ r2 ]]

Stresscpu/super_pi 32MB pass and memtest :p:

[TheReaper]

As said above, I'm using Super Pi 32MB and memtest (windoze version).

[DevilsRejection]

so i should just loop superpi 32m for a few hours and its considered a stable system?

hmm i can only run it once, anyway to make it loop?

I want a nice app, or atleast make prime95 work with this core.

[[ r2 ]]

so i should just loop superpi 32m for a few hours and its considered a stable system?

hmm i can only run it once, anyway to make it loop?

I want a nice app, or atleast make prime95 work with this core.

32Mb pass usually will show stability for cpu+ram, but nor enough so best thing is to run stresscpu for atleast 8hrs and also memtest test #5 or #8

[DevilsRejection]

where can I get this stresscpu?

EDIT: http://home.comcast.net/~wxdude1/emsite/download/stresscpu.zip

OK so now how do I make it report how long it has been running?

[agenda2005]

We cannot just rule out Prime95 like that. Prime95 is correct, but something is obviously wrong with AMD A64 memory controller if P95 fails at stock speeds. Prime 95 is the ultimate stability test and for those of us working with floating point computations P95 is definitely the holy grail and we can not just accept a failure just like that. Someone has proposed a petition to AMD and was told that something will be done about it. AMD better do something because I cannot run computations for days just for the computer to chunck out gabbage at the end. AMD must do something about it.

[HiJon89]

I think its actually Prime95 that has the flaw, from what I've heard they're working on a newer version that will test 90nm CPU's properly

[[ r2 ]]

I think its actually Prime95 that has the flaw, from what I've heard they're working on a newer version that will test 90nm CPU's properly

Yes they are E revision which will be availabe in Q1 2005.

[bachus_anonym]

i think that all should understand that this is most likely overheating issue... the only blame that you could put on this excellent piece of soft is that it heats up that poor Winnie a lot...
for the last couple days i've been trying to pass at least 12h of prime95 at 265x10... i could do only up to 3 hours of Blend. so i gave up on 265 and lowered it to 260x10.
and guess what? 9hrs so far running overnight...
you guys should try to run Small FFTs and Large FFTs seperately to figure out what the problem is. in my case i think that problem is somewhere within the processor that just can't hande 2650MHz prime stable, although it can do SuperPi 32M at 2700MHz. why? coz i can easly pass full loop of LargeFFTs but Small FFTs fail after about 2 hrs. both variant don't test RAM much but only cache and CPU itself.
just play with BIOSes and settings... keep in mind that 1.40v in bios IS NOT 1.4v REAL.... it's somewhere close to 1.34V LOAD ;)

[agenda2005]

i think that all should understand that this is most likely overheating issue... the only blame that you could put on this excellent piece of soft is that it heats up that poor Winnie a lot...
for the last couple days i've been trying to pass at least 12h of prime95 at 265x10... i could do only up to 3 hours of Blend. so i gave up on 265 and lowered it to 260x10.
and guess what? 9hrs so far running overnight...
you guys should try to run Small FFTs and Large FFTs seperately to figure out what the problem is. in my case i think that problem is somewhere within the processor that just can't hande 2650MHz prime stable, although it can do SuperPi 32M at 2700MHz. why? coz i can easly pass full loop of LargeFFTs but Small FFTs fail after about 2 hrs. both variant don't test RAM much but only cache and CPU itself.
just play with BIOSes and settings... keep in mind that 1.40v in bios IS NOT 1.4v REAL.... it's somewhere close to 1.34V LOAD ;)

I can't agree with you on this one. Even people using phase change and sub zero temperatures got error on Prime 95 after couple of hours. It cannot be prime itself or because it heat up the processor as other reports shows that Athlon XP and Intel P4 have not experienced similar problems even at higher temperatures. We need to do alot more research and let AMD do something to resolve the problem in their upcoming revision A64 CPU.

[bachus_anonym]

so how would you explain that SOME people can run Prime95 and some just can't pass more than few hours (as far as we're told). that totally rules out Prime95 as a faulty software.
so maybe we just should gather all the infio about people that can and can't run Prime... like mems, cpu stepping/week and voltages? i really don't think that it's Prime's fault so the only other reason would be just "something wrong with the overall config and enviroment" not just CPU itself...

EDIT: btw, this is what i got so far... mbm was started together with prime95 so it's 9hrs...

http://www.xtremesystems.org/forums/attachment.php?attachmentid=20711

[agenda2005]

so how would you explain that SOME people can run Prime95 and some just can't pass more than few hours (as far as we're told). that totally rules out Prime95 as a faulty software.
so maybe we just should gather all the infio about people that can and can't run Prime... like mems, cpu stepping/week and voltages? i really don't think that it's Prime's fault so the only other reason would be just "something wrong with the overall config and enviroment" not just CPU itself...

EDIT: btw, this is what i got so far... mbm was started together with prime95 so it's 9hrs...

http://www.xtremesystems.org/forums/attachment.php?attachmentid=20711

bachus_anonym,
Thanks for that invaluable thought. Running MBM5 simultaneously with P95 for 24Hours should give us an idea of voltage/temperature variations in the system during the test. A very good thought indeed! I'm currently getting my system set to run similar test. My PC power & cooling PSU 510W deluxe will arrive today and with that I should have a stable constant voltage with minimal variations (~1%) for my system. With that, I want to rule out system instability due to PSU or something like that. This is what my system configuration will look like by 4:30pm today

A64 3500+ 130nm at 200*11 = 2200MHz (Stock Speed) HSF Removed!
A Reliable 512MBx 2 Mushkin lvl2 PC 3500 running at DDR400 PC3200 2-2-2-10
MSI K8N Neo 2 Platinum 1.3 Bios(default that arrive with the MB)
PNY Verto GeForce 6800
Thermalright XP-120 with 110cfm 120mm Panaflo fan at full speed
510W PCP & C PSU
74GB WD Raptor
Win XP SP2 (No Anti virus, no other program running in the background except MBM5)

TEST --> Prime95 toture test(small and large FFT blend) for 48hours
with no other program running except MBM5 in the background.

RESULT: We can urge other people in the forum to submit their results with the system configs to know what is really going on.
That's just my opinion.

[dnottis]

Personally, I think it's an issue with the voltage fluctuations...

[DevilsRejection]

well I got my silenx hooked up to a UPS w/ voltage regulation built in, so IDK, first I have to get the parts then I will test everything out.

today I got my ocz ddr booster :)

[agenda2005]

well I got my silenx hooked up to a UPS w/ voltage regulation built in, so IDK, first I have to get the parts then I will test everything out.

today I got my ocz ddr booster :)

Nice idea once again. I think I will get the UPS voltage regulation too and plug it into my PC Power & Cooling 510W PSU to further rule out the voltage regulation problem. Hope to see what this will bring to the table. I will held out to COMPUSA to get one soon.

[bachus_anonym]

guys, also i would not trust too much MBM.... soft is NEVER accurate. e.g BIOS shows pretty much right 12v rail where MBM has some weird reading as you can see.
also, on the screen i posted there's some fluctuation on all of the voltages. i actually measured vcore using multimeter for 10 mins and it just moved SLIGHTLY under LOAD - 1.81v - 1.83v. vcore under load goes up comparing to IDLE.

@agenda2005.... i don't know about your 2x512kit of LvL II but mine LvL would never be prime stable at only 2.8v (max vdimm without Booster or mods)!!!

[Beenthere]

Current version of Prime 95 works fine. Good quality PSU is required on all new CPUs since Athlon/P4.

http://www.houseofhelp.com/forums/showthread.php?t=22249

[dnottis]

']Yes they are E revision which will be availabe in Q1 2005.


This is just stupid. First of all I have issues with Prime95 too. It will run once for 10 hours, then not for 10 mins. I understand something is a miss. But you are going to hope AMD releases a new CPU that will "FIX" Prime95? Or you think Prime will be patched to "work correctly with 90nm CPUs"? Umm, yea. Have you ever considered that the voltage dips is causing Prime to fail - this is for the stock folk anyways.

For the overclockers like myself - I ran stable Super PI 32M, 3dmark 2001 SE and primed for 10 hours stable (sure, the next run crapped out at 10 mins) - but I still BSOD'd randomly the next day. So regardless of what you think of Prime95 - its still showing that there is some instability within the system. Whether its a slight voltage dip, issue with memory controller being run way out of spec, cooling or something else I dont think AMD needs to update their product or prime needs to write a patch for us running the hardware above the intended specifications.

As for the people having issues with prime at stock speeds - again it's probably not the CPU or the APP, you have a problem (PSU, Mobo, etc).

I have an Asus A8v coming in to see if I can get any better results. I've had 2 Neo2's and the voltage dips really concern me with this motherboard.

Just my 2 cents, not looking to start a war -but some of these posts are getting ridiculous...



"Now that we can rule out this app working with whinny's then what is the other alternative? "

Well lets see - prime95 doesn't give your OC the "OK" it must be the app - so "what will make my unstable OC seem ok now, what apps should I use to fool myself?"

[dnottis]

Current version of Prime 95 works fine. Good quality PSU is required on all new CPUs since Athlon/P4.

http://www.houseofhelp.com/forums/showthread.php?t=22249


Prime 95 238 has been out for awhile.

[bachus_anonym]

very true, dnottis ;)

[agenda2005]

guys, also i would not trust too much MBM.... soft is NEVER accurate. e.g BIOS shows pretty much right 12v rail where MBM has some weird reading as you can see.

.


also, on the screen i posted there's some fluctuation on all of the voltages. i actually measured vcore using multimeter for 10 mins and it just moved SLIGHTLY under LOAD - 1.81v - 1.83v. vcore under load goes up comparing to IDLE.

@agenda2005.... i don't know about your 2x512kit of LvL II but mine LvL would never be prime stable at only 2.8v (max vdimm without Booster or mods)!!!

I agree with you on that. My 12V rail is showing something lile 2.98V too on MBM5 with MSI K8N Neo2 MB. We can't trust MBM5 too but it should give us some idea of what is going on. Oscilloscope will be the only true test, but how many people can afford it. I don't have one either, but I have a multimilter which you can only use at random, but that would not help over a long period of testing.

Thanks! My Mushkin lv2 ran memtest86 stable at DDR400 2-2-2-10 for 24 hours at 2.75V. I have OCZ DDR Booster and can feed it 2.85-2.9V if that will make you feel comfortable. AMD A64 CPU memory controller are spec to run fine upto 2.9V Vdimm.

[agenda2005]

dnottis,
That's right. I can't say it better myself. Some people like to fool themselves. It's better to investigate what the problem is than just ignoring it. It's even more important to those running D2OL or Folding@HOME to have a stable machine for they might be doing disservice to those projects rather than helping it.

[G H Z]

http://www.mersenneforum.org/showthread.php?t=3177

That leads me to believe it is a problem with the code running on A64 architecture, not that AMD need's to 'fix' something.

[cuddles]

Doesn't StressCPU use the same computation as Prime95? Or does it go about it differently, so we won't get errors? If so, I'll be uninstalling Prime and loading up StressCPU.

[dnottis]

:rolleyes: Prime95 isn't crashing because it's not compatible with AMD64s, it's crashing because there is an instability within your system. If you don't desire 100% stability don't bother running anything to stress it. Uninstalling Prime and installing some app that will tell you your OC is ok is just stupid, you guys are missing the point entirely - there isn't a problem with Prime. :rolleyes:

[cuddles]

The entire point of this thread is that there is an issue with the memory controller in the A64's compatibility with Prime. If that's the case, I was wondering what program would we be using otherwise. We still don't know 100% sure where the problem is, as bigtoe has shown by his A64 failing at stock speeds. I never said anything was wrong with Prime, I'm going by what people have said in this thread, and asked quite a simple question. Clearly you missed the point of my post.

I forgive you.

EDIT : :rolleyes:

[quicksilverXP]

Nothing is wrong with prime.... but Winchester's still fail at stock speeds. My winnie has been at 2.775 for the past two weeks... and it hasn't shown any instability whatsoever.

[dnottis]

I've primed fine at stock speeds and mild OC (240 x 10) for hours. That why I pointed out, failures are stock are something else...

[bachus_anonym]

has anyone of you guys with prime95 failing at stock checked voltages after LOAD DEFAULTS from BIOS? i don't know about you guys but raising vdimm to 2.85 from within BIOS (before i got the Booster) was NOT working!!! i had to use CoreCenter to do that.
just get yourselfs a multimeter and poke around at stock speeds... that's what i always do...

[dnottis]

has anyone of you guys with prime95 failing at stock checked voltages after LOAD DEFAULTS from BIOS? i don't know about you guys but raising vdimm to 2.85 from within BIOS (before i got the Booster) was NOT working!!! i had to use CoreCenter to do that.
just get yourselfs a multimeter and poke around at stock speeds... that's what i always do...

Thats what I'm saying with the stock issues, there is something going on different than the people with 600 MHZ oc's. If you are failing at stock it's a board issue, power issue, etc, etc...

[agenda2005]

http://www.mersenneforum.org/showthread.php?t=3177

That leads me to believe it is a problem with the code running on A64 architecture, not that AMD need's to 'fix' something.

ABSOLUTELY NOT. The Aurthor is talking about OPTIMIZATION OF THE CODE FOR K8 ARCHITECTURE AND NOT COMPATIBILITY ISSUE OR ANY BUG. READ CAREFULLY.

[Rabbi_NZ]

I think Prime95 is fine... if anything is causing problems it is either the actual memory controller hardware failing, OR, the AMD64 drivers for Windows may need an update to use the new stepping properly... jus my opinion.

[jlccarv]

or a BIOS update is nessesary for 90nm CPUs

[arj]

My 3000+ Winchester is 12 hours Prime stable at 2.6Ghz :banana:

[[ r2 ]]

My 3000+ Winchester is 12 hours Prime stable at 2.6Ghz :banana:

What week is your 3000+?

[arj]

Week 37. The 3200+ that I had which is from the same week has no problem priming as well.

[[ r2 ]]

Week 37. The 3200+ that I had which is from the same week has no problem priming as well.

I havent had much trouble priming, i was having issues priming with 1:1 o/c 3700 EBs and XL PROs but now using OCZ 3200 Plat rev2s there arnt many problems priming.

These days i just stresscpu/memtest/super_pi 32MB pass and 3D.

I noticed in your sig that you using OCZ 4200EL Plats, how are they performing for you? Im getting them real soon from the US as im in Aus and we dont have them anywhere.

[G|-|oST]

Mine primes like a champ at anything upto 2.5G. Any combination of HTT and dividers etc. I still belive there is nothing wrong with using prime to test stability.

If it fails, just remember this - it compares results against KNOWN results, thats all. Imagine if it was doing something simple like 2 x 2 = 4 and the like.
Well if there is ever an occasion when your CPU comes up with 2 x 2 = 5, there is a problem with your machine. No two ways about it.
You can ignore it if you like, but the fact remains - your machine IS somewhat unstable.

[[ r2 ]]

Mine primes like a champ at anything upto 2.5G. Any combination of HTT and dividers etc. I still belive there is nothing wrong with using prime to test stability.

If it fails, just remember this - it compares results against KNOWN results, thats all. Imagine if it was doing something simple like 2 x 2 = 4 and the like.
Well if there is ever an occasion when your CPU comes up with 2 x 2 = 5, there is a problem with your machine. No two ways about it.
You can ignore it if you like, but the fact remains - your machine IS somewhat unstable.

Well there still might be a problem with mem controller for some with winnies but im my case i havent had any problems priming with 3200+ Winny.

[LightSpeed]

or a BIOS update is nessesary for 90nm CPUs

i would strongly think that, for some motherboards is required to support the new winnies properly.

I think its the BIOS, cause some ppl have them working properly and some do not.

However i dont think its the software, to be honest.

[dnottis]

']Well there still might be a problem with mem controller for some with winnies but im my case i havent had any problems priming with 3200+ Winny.

I can prime stable with my 3200 @ 240 x 10 and 267 x 9 at stock volts for 12+ hours. However over 2.5G I can prime for 6 hours, then crap out at 10 mins. If it's the memory controller why can I consistently prime at 267 x 9? As long as my CPU doesn't get over 2.5G it seems to Prime just fine, it has more to do with overall clock speed than the HT. It's once I hit 2.5G that things become unstable, (and it doesn't matter how 250 x 10, 278 x 9) it has nothing to do with the memory controller, the CPU despite what people are saying, is just generating errors once it's clocked too high.

..or it's the chipset generating errors, voltage dips, power issues, memory issues, timings too tight, etc, etc if you are stock clocked and experiencing issues with Prime - or believe it or not maybe it's just a bad CPU. Of the 4 winchester CPUs I've had one had issues priming at stock speeds - the other 3 have not. So was it Primes fault that one CPU had problems, hell no, it was a bad CPU that got RMA'd and the replacement worked fine.

[[ r2 ]]

The best thing to do is to double check using another torture program such as stresscpu just to be certain.

[jikdoc]

my experience with prime95 and winnies pretty much echoes what dnottis wrote. i was prime stable at 2.5GhZ on a 0433 3200+ winnie at either the 9x or 10x multi. anything above this i'd get errors.

[first]

cpu burn in.... really good for stress test

http://users.bigpond.net.au/cpuburn/ ---------> download from here

Why CPU Burn-in is better:

In the past overclocking stability was tested by running intensive software such as Distributed.Net or SETI@home. Running either piece of software for 24 hours would generally show any possible instability. A looping Quake3 timedemo was also a good choice.

However, there are inherent limitations in these tests:

Not every error caused by overclocking causes a program to crash or the system to hang. Some errors may be more subtle, such as a slight miscalculation. If such an event occurs and causes a pixel to render a slightly different colour in Quake3 for example, the user is unlikely to notice and overall this is no big deal. However such small errors can have a potentially devastating on distributed projects such as SETI@home, which rely on the reliable processing of data.

As shown in the graphs below, CPU Burn-in consistently delivers a higher CPU operating temperature than the above mentioned applications, and other competeing CPU stress test programs. This allows CPU Burn-in to be particularly effective at testing stability and cooling effectiveness.