P95 - FATAL ERROR: Rounding was 0.5, expected less than 0.4

DarkStar02

2[H]4U
Joined
Mar 1, 2006
Messages
2,144
What does this mean? I read on another forum that it meant I needed more volts, but P95 still fails at 260x10 (5/4 dram freq) on 1.6v

I can still cruise around Windows, but apparently I'm not stable. Why is this?
 
it means your oc is unstable, try turning up your voltage or lowering your oc.
 
7/22/2006 11:33 PM Beginning a continuous self-test to check your computer.
Press Stop to end this test.
Test 1, 4000 Lucas-Lehmer iterations of M19922945 using 1024K FFT length.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture Test ran 0 minutes 26 seconds - 1 errors, 0 warnings.
Execution halted.


How much higher can I go? I'm already at 1.6v! :eek:

EDIT: Without increasing the voltage, I re-ran P95 and it's been going for over 4 minutes now.

EDIT2: Over 17 minutes now without failing. Was the first fail some sort of qwerk? The system not finishing booting caused instability in P95?
 
24 minutes passed and my PC rebooted. Why did it restart instead of stopping the torture test?
 
Just curious, have you removed the IHS? The heat may not be dissipated quickly enough if it's still on there.

IME that rounding error is what I get with too low of VCore, but I have seen insufficient VDimm/tight timings cause it as well. Instant reboot sounds like a possible PSU issue, sometimes it's also low VCore/something screwy with the memory controller.

At 1.6v, if the IHS is off, I don't think there is much if anything to do. Those BW steppings just don't have the oomph of the B1/BB chips *shrug*
 
PliotronX said:
Just curious, have you removed the IHS? The heat may not be dissipated quickly enough if it's still on there.

i dont think that taking off the IHS is nesessary. you should either turn down the overclock, turn up the voltages, or loosen up the ram timmings/ more voltage to ram.


 
try checking if your RAM is the problem, that's usually where it was if i got a rounding error
 
omz said:
try checking if your RAM is the problem, that's usually where it was if i got a rounding error
if it's randomly rebooting, i second this motion
 
try getting your cpu stable without overclocking the memory seperately. some memory controllers simply cannot handle much over ddr400, opteron or not. Timings also come into play, but find the cpu limit first, ram limit second, and settle for a happy medium. there's no way you should need 1.6v for 2.6 on a 170, that may even be too much voltage for the cpu, in fact limiting your oc.
 
Make sure the HTT Multiplier is dropped down to 3x and isn't 1.6v a little high?
 
Your RAM is definatley bonked. I had the exact same problem, ran memtest and one of my sticks was bad. Called Crucial and they sent me two new sticks, on of which was bad, but I just used one from the old and one from the new. Working very smoothly now.

Run MEMtest and check to see if your RAM is pooped or not.
 
sdk said:
Your RAM is definatley bonked. I had the exact same problem, ran memtest and one of my sticks was bad. Called Crucial and they sent me two new sticks, on of which was bad, but I just used one from the old and one from the new. Working very smoothly now.

Run MEMtest and check to see if your RAM is pooped or not.

I have had memtest report no errors, while P95 reports errors. Low and behold it was a memory problem. So dont necessarily believe memtest if it tells you everything is ok.
 
I bet the new RAM will help. Even running the divider you are using, the RAM is running higher than 200 which it may or may not be good for.

Tons of people on the boards have the G.Skill HZ's (myself included), and they are consistently good overclockers. I would just about gurantee they'll be good for 260 at stock timings. Mine do about 275-ish, and that's around average. Try running a bigger divider, or see if the new ones help. Heh, kinda like everybody else said i guess........
 
Agreed. From Most to least likely.

Your at your Ram's Limit (apply a higher divider)
Your running too Hot (get better cooling, try with the side panel off)
Your at your board's limit (lower the FSB and increse the multiplier)
Your at your CPU's limit (lower the multiplier and increase the FSB)
Your PSU is screwy (get a better PSU)
Your board is screwy (hey it happened to me).
 
thunderstruck! said:
I am surprised no one mentioned temps. If P95 errors, you are simply unstable.

QFT. I would try lowering the overclock or loosening the timings rather than check for hardware problems. When you're overclocking, Prime95 is supposed to be used as a tool to tell you when you've gone farther than your hardware can handle.
 
Not necessarily the memory, although that'll do it. I always set my memory to 1:2 divider and HT multiplier to 1x when testing OCing though.

Usually when I get a rounding error it's not enough VCore or overheating. Download a program called Core Temp and check your temperatures when running P95. Anything over 70C is hot and your CPU is very likely to crash beyond 75C. 1.6V is too much for my CPU to handle but YMMV (depending on your cooling, I use a Thermaltake Silent Boost K8 RX). As some have said, raising VCore may not increase stability. In fact, my CPU won't boot to Windows if I push it past 1.55 VCore.

Athlon64 X2 3800+ @ 1.55v @ 2.5 GHz
 
I have to thank all of you for helping to solve my ‘FATAL ERROR: Rounding was….’ problem. Indeed, when I increased Vcore to bios 1.450 (droops to 1.328 loaded by Orthos), my system now runs Orthos error free. The intensified Vcore necessitated the addition of a side-panel fan to keep CPU Tcase within spec, but hey, who cares. When I am finished torture testing I might remove the new fan and set Vcore down a bit since my normal video conversion applications, with their disk-intensive operations, do not produce such extreme, prolonged CPU performance. Again, thanks to all. --jerry
 
Back
Top