O/Cing Gurus..why does Prime95 fail but Hot Cpu tester runs

[nealh]

In the FAQ at the top of the page..it recommends looping 3DMark2001,Prime95,HotCputester etc for stability

Just built Monday: running about 15hrs

P4 2.4@3.0 retail HSF and ASII at 166fsb on IT7 MAx2 Rev2(fsb is actually 166.8) at 1.7v board undervolts by 0.25...idle at 1.67/1.68...load 1.65v
Crucial PC2100 2x256mg ram 166fsb cas 2.5/3/3/6..at 2.8v...I am trying 2.7v today to see if this ok
Just so you know temps seem hot but I a mpretty sure this board reports high temps(my old KR7A in same case with same fans would give a case temp of 27-30)

Case38/40
Idle 47/48
load 69

I think these temps are 5-10 c (more like 8C) too high


Hotcpu tester ran last night for 1 1/2 hrs..I stopped it...and indicated there were failed tests with cpu 99.9% load..no crashes
3DMark2001SE seems fine as well

Prime crashes immediately???

Why would one cpu intensive program seem ok..but another not..is Prime 95 that more sensitive

also is 2.8v for crucial ddr ram too high

Thanks

 

[SupermanCK]

it could be the ram...you should go ahead and get some pc2700 at least just so you can eliminate the ram as ur problem

to answer you main question: prime95 can be sensitive to certain part of you system, ie Memory, not just the CPU. But there can be a number of things affecting it...but since HotCputester doesn't give you error, i believe the reason for prime to fail is ur memory. But try adjusting ur vcore a notch and see if that help pass prime95...if it does...i think you should knock ur fsb down so you don't go pass 1.7v

 

[Duvie]

I agree with supermanck....Prime will weed out certain memory and memory subsystem errors. Try memtest86 and see if the ram is running fine at that speed. Also may want to for sake of testing drop memory speed, cas settings, or what not to test if the ram is the issue....

 

[nealh]

thanks for the info...Mem86 is run in DOS correct....

Very interesting I just ran 3dMArk2001SE for about 2 hrs without a hitch max temps seem to be around 60C

On IT7 Max2 Rev 2..max voltage is 1.725 but since the undervolts by 0.25v at idle..you never get over 1.7v....mobo will not allow it...this undervoltaging is reported in both WinBond hardware monitor and MBM5

the problem with temps is interesting too..the temp will 69/70(this is as alarm temp by winbond hardware monitor) but the cpu never crashes...also I added 2 fans one exhaust and one front fan this lowered cpu temp a bit...

I am disappointed by the temps since this gives I believe a false sense of a concern though I get the diode reading the temp on this Abit board maybe closer to a real cpu temp(how the hell do I know)

BTW is 2.8v too high for DDR ram??

 

[nealh]

Ok ran Memtest86..but just 2 pass and stopped it..is this another long loop test???..No errors

For anyone that care..I checked Memtest86 website and it says one pass should be fine to check for errors..although in my case I think it only ran 7 tests not all they listed

Update...decreased vcore to 1.675 in bios...1.65 idel/1.63 load
HotCPU tester after 1hr 15min was ok all tests passed

Well I will run WinXP Pro for awhile and see what happens

 

[TimisoaraKill]

This is strange ,..

 

[nealh]

I do not understand this at all...I would figure the other tests would crap out as well after a few minutes if prime95 fails that quickly

As I said I did not run the ram for 5 hrs via Memtest86..I did there default settings and looped twice...when i was doing I had no idea therre were more than the 7 tests....

I am not sure I have all the tests as I have the shareware version

I am at a loss...I will try some video editing stuff next and see if it crashes if not ... It is stable enough for me

 

[nealh]

My fsb is 166 on IT7 Max2 Rev2...the ram seems fine

Crucial PC2100 DDR 2.7v....at cas2.5/3/3/6 it will run at 166fsb....it does 160 fsb at 2.5/2/2/6 maybe 5



As I have stated in previous posts...this setup seems very stable but prime95 crashes...but 3D MarkSE 2001 and hot cpu tester have run at least 1.5hrs without a hitch...have not gone longer as my wife keeps using the computer
I have even be able to lower vcore...but the retail HSF temps are warm and I am sure it is mobo reading high..as the same case with 5 fans and athlon xp case temp was at least 5C cooler

P2.4@3.0ghz at 1.65 in bios actual is 1.63 load drops to 1.60v...
IT7Max2 Rev2..undervolts
Crucial PC2100 DDR 2x 256 at 2.7v at cas2.5/3/3/6..run memtest 86 default 7 tests for 2 loops no issues but need to do it longer as the extra tests stress the memory for hrs
Radeon 9700
3HDD 80gb 7200rpm DX740 Maxtor on serial ATA via adapter..master
80 gb WD 7200 rpmSE master
120gb Maxtor 5400 rpm master....all 3 HDD are on HPT 374 controller
LTD 16x DVD
Liteon 52246S
Aureal Vortex 2 works in winxp....snds very nice
SCSI adaptec 2906 for my scanner
Onboard realtek NIC
Onboard IEEE 1394
USB2
5 fans..3 sunon at full speed, 2 antec on fan control by PS
Antec True Power 480 w
WinXP

I have asked this before..why in the world will hot cpu tester and 3dMark2001SE run without crash but prime 95 dies right away

Also load temp is 64-67C with ASII and retail HSF...it clipped down hard...I really believe this mobo reports higher temps...also the HSF doesnot get very warm to touch and would guess if the HSF was not in good contact the temps would continue straight up under ful load and not pleateau

 

[Antos]

I've heard many reports that Prime95 is bombing out on overclocked CPUs' faster than 2.8 ghz, even though the CPU's will run anything else for hours.
This seems to be either a prime95 issue, or a strange FSB problem. Especially if NO OTHER program is having problems, including BURN BX, BURNP6, Super PI, and others. Perhaps Prime95 is having difficulties with very high FSB speeds?

Of course, I've seen even stranger reports, that people can run Prime95 24 hours a day, at those high frequencies, while superPI crashes within 10 minutes....

 

[Tates]

Same deal here.

My 2.4b "C1" @2880MHz runs everything (3DMark, Games, etc.) fine for hours on end, without a hitch. Try running Prime95, instant crash. That suggests to me that Prime95 is not the end-all, know-all benchmark.

I think Antos is on to something.

 

[nealh]

I have never been able to use prime95...but in the past my systems after awhile needed to be dropped back because other programs crashed...but I have run these other programs for somewhat short times with no apparant problems....even did video editing which crashed my Athlon XP at 150fsb but was ok at 148fsb ...on my new P4 and seemed ok but I do need more testing to know for sure...

My other problems are the temp not sure what to make of them and I am sure the HSF is in good contact...

I guess time will tell

What HSF would you guys recommend

 

[THUGSROOK]

prime95 runs fine for me.

chances are these are not bugs but in fact reality ~ prime95 is THE program to run for stability.
dont blame the program.
hot cpu? wth is that?

prime95 and memtest86 are still the kings of stability

 

[Mikki]

...just as a confirmation, no problems here either...

 

[RalfHutter]

Same here. If Prime95 and Memtest86 run w/o errors you are stable. If they give errors, sooner or later you'll see the results of it in Windows.

Trust them, they are an OCers friend.

 

[Tates]

This article posted 1/1/03 at Overclockers.com suggests there is merit to the reports from "C1" owners regarding Prime95:

"The Prime Culprit?"
Ed Stroligo - 1/1/03
--------------------------------------------------------------------------------
A very common complaint among PIV C1 users has been that Prime95 collapses on them while other stability checks go merrily on its way at the same speed, and that Prime only stabilizes at speeds much lower than the other tests.

While Prime has often been more sensitive than others in the past, it seems to be more so than usual here.

The problem might not be your machine, but an older version of Prime 95.

If you go to Prime 95's revision history, you'll see that a number of changes were made in versions 22.2-22.6 to correct bugs that created erroneous errors. In particular, versions 22.2 and 22.3 have revisions made because "version 21 was too aggressive in choosing FFT sizes for the P4."

This may/might be the reason why at least some machines are failing Prime 95 at speeds a good deal lower than other stability tests. Then again, it may well not; for instance, many of you may have well downloaded the last version when you tested your C1 in the first place.

But that's what experimenting is for, to look into all the possibilities, and see whether or not any of them is the cause. My suspicion is that this isn't THE reason, but may be the case in a few cases, especially when the error occurs well into testing.

If you want to see if that's the case for your machine, download the latest version of Prime, and see if it stays stable at a speed higher than the earlier version.

Tell me if you does or doesn't (or if you used the latest version in your initial test, and the machine still died sooner rather than later, tell me that, too).

If it does we'll report on it; if it doesn't, we'll report on that, too.

 

[LazyBastard]

I noticed that there is a new version of Prime95 out. It was at version 21 for a long time, but now it's at version '22.12'.

Edit: I guess I should have read all the other messages in this thread. Specifically, the one right before this one.

 

[nealh]

Just an FYI update..I have run this P4 2.4@3.0 ghz at 1.65v (actual is 1.625) with crucial pc2100 ddr ram at 166fsb and have encoded about 4 different dvd movies..they average 1 1/2 hrs of video with encoding times of 2-3 hrs ..I used DVD2SCVD which uses tmpgenc as the video encoder, also used DVDx with P4 version(this is finicky program to get the P4 to work the best I have used in input settings iDCT either FPU or SSE2/P4 and output need to set DCT/iDCT to P4 otherwise I get a crash with this version)

So now 3DMark 2001SE and HOT cpu tester work fine but Prime 95 still crashes.....

Also just so you know the VCD I created look awesome on a TV much better than I ever expected given dropping resolution almostinhalf from DVD quality to VCD..so the overclock did not produce a video with dropped frames and poor quality

I just hope as temps in FL go up this summer I do not have to drop speed as I did with my Athlon XP last yr

 

[ShoNuff]

I just got a C1 and I was searching the forum b/c Prime95 won't run. Memtest86 ran for 4 hrs not a single problem. My B0 2.4 that I ran at 2.9 would work with P95 without out a hitch. My 2.66 (oc'd to 3.2) will not run P95. I am using the same system and the same memory from before. The only difference is the processor.

I blame the program!:| Ultimately time will tell.

Anybody with a C1 overclocked higher than 2.8GHz that can successfully run P95 please stand up!?

 

[Technonut]

Anybody with a C1 overclocked higher than 2.8GHz that can successfully run P95 please stand up!?

Standing Up

 

[THUGSROOK]

why would C1s be effected and not B0s?

 

[RalfHutter]

Originally posted by: ShoNuff


Anybody with a C1 overclocked higher than 2.8GHz that can successfully run P95 please stand up!?

Standing tall here.

2.66 @ 3.0Ghz, passed 24hr loops of Prime95 on several different occaisions.

 

[Technonut]

Originally posted by: nealh
Just an FYI update..I have run this P4 2.4@3.0 ghz at 1.65v (actual is 1.625) with crucial pc2100 ddr ram at 166fsb and have encoded about 4 different dvd movies..they average 1 1/2 hrs of video with encoding times of 2-3 hrs ..I used DVD2SCVD which uses tmpgenc as the video encoder

Also just so you know the VCD I created look awesome on a TV much better than I ever expected given dropping resolution almostinhalf from DVD quality to VCD..so the overclock did not produce a video with dropped frames and poor quality

I use DVD2SVCD / TMPGEnc Plus as well, and am quite impressed with the results. I will have to try it on my TV sometime. The quality is great on my 21" iiyama Vision Master Pro though.

 

[ShoNuff]

Originally posted by: RalfHunter[/i]

Standing tall here. 2.66 @ 3.0Ghz, passed 24hr loops of Prime95 on several different occaisions.[/i]

Awesome...Hmmmm. We may be on to something here?

 

[kponds]

I for one think there is an issue with P95 and C1's or maybe C1's running at high FSB. I came here specifically looking to see if anyone was having problems with Prime95. Here's my story with two C1 stepping 1.8As:

Both tested on Albatron PEV Pro with core voltage set to 1.75 in bios (reading 1.7). Memory set to 1.5x to eliminate memory issues. Using known good OCZ PC3200 which has ran at DDR450 Cas2. Both tested under an Alpha cooler and AS3.

CPU 1 ran 3dMark2001 Looped Benchmark for 1 plus hours without a hitch at 173 Mhz FSB (3114 Mhz). I ran prime95 starting at 173 Mhz FSB and kept getting errors, usually within about 10 seconds, dropping the FSB by 2 Mhz after each error. I finally got it to run for about 10 minutes at 160 Mhz FSB (2880 Mhz) before an error and I stopped and changed CPUs. Delta CPU speed = 234 Mhz

CPU 2 ran 3dMark2001 Looped Benchmark for 1 plus hours without a hitch at 182 Mhz FSB (3276 Mhz). I ran prime95 starting at 182 Mhz FSB and kept getting errors, usually within about 10 seconds, dropping the FSB by 2 Mhz after each error. I finally got it to run for about 10 minutes at 164 Mhz FSB (2952 Mhz) before an error. Delta CPU speed = 324 Mhz.

I am/was a big believer in prime95 being the only test for CPU/system stability but the differences I am seeing here versus other CPUs is too significant for me to not consider that something else is going on that I do not understand.

 

[kponds]

and immediately after I posted this I decided to do a search on google using "prime95 C1". The first hit was a link to an article on overclockers.com. Following is a quote from the article:

"6) Prime isn't fine A number of people are reporting that Prime95 is the bugaboo in their stability tests; they're running other benchmarks at higher speeds, but Prime collapses at much lower speeds. "