Winchester + Prime95 = no go? - Page 2

Venomous21 · 2005-06-12, 03:17

I have a athlon 3700 (2.2ghz 90nm 1mb l2cache). I am running at stock settings, passing memtest86 with flying colors, and failing prime95 almost immediately.

Is this a known bug? I'm at a loss now....I do not know what is wrong with my system or how to fix it.

dsouza123 · 2005-06-12, 04:57

Download the Prime95 program again maybe the copy you have is corrupt.

Still can be RAM problem.
If there are multiple sticks of RAM retry with one at a time.
If all pass alone try in combo.

Other tests.
Underclock by a large amount ie down to 1 Ghz then retry.

If using Windows test in Safe Mode.

If the PC cover can be taken off or side panels easily removed,
try with case open.

While the PC is off check/clean out dust from CPU heatsink and fan.

The CPU voltage could be too low, if the BIOS allows it try increasing it
a small amount.

The power supply could be giving bad/noisy or low power to the CPU.

Cruelty · 2005-06-12, 14:42

I have had a similar problem with my A64 3500+ (E3-stepping) - it turned out to be a BIOS problem (DFI NF4) + RAM (to some extent).
Generally, version 24.11 and 24.12 stress tests (blend) seem to be harder on A64 than 23.8...

db597 · 2005-06-13, 14:12

This seems to be a common problem. I've seen long posts about this over at the [H]ardOCP forums. It was even suggested that there's a bug in Prime95, but that's not true. Prime95 goes further than any other program to stress test the CPU and memory subsystem. The code in Prime95 has been tried and tested for a long time, on many CPUs. It's worked on Opteron and A64 up to now as well.

As Cruelty mentioned, a BIOS update might fix the problem. I have a similar issue with one of my P4s. I've used that P4 CPU in a different board before with no problems. But when I swapped it over to an Asrock board, I couldn't get it stable unless I underclocked the CPU by about 5% and turned off "V-Link" in the BIOS - it's still running on that today.

Maybe the hardware quality control at Asrock wasn't up to the standards required by Prime95.

2005-06-12, 03:17	#12
Venomous21 Jun 2005 2₁₆ Posts	I have a athlon 3700 (2.2ghz 90nm 1mb l2cache). I am running at stock settings, passing memtest86 with flying colors, and failing prime95 almost immediately. Is this a known bug? I'm at a loss now....I do not know what is wrong with my system or how to fix it.

2005-06-12, 04:57	#13
dsouza123 Sep 2002 2·331 Posts	Download the Prime95 program again maybe the copy you have is corrupt. Still can be RAM problem. If there are multiple sticks of RAM retry with one at a time. If all pass alone try in combo. Other tests. Underclock by a large amount ie down to 1 Ghz then retry. If using Windows test in Safe Mode. If the PC cover can be taken off or side panels easily removed, try with case open. While the PC is off check/clean out dust from CPU heatsink and fan. The CPU voltage could be too low, if the BIOS allows it try increasing it a small amount. The power supply could be giving bad/noisy or low power to the CPU.

2005-06-12, 14:42	#14
Cruelty May 2005 658₁₆ Posts	I have had a similar problem with my A64 3500+ (E3-stepping) - it turned out to be a BIOS problem (DFI NF4) + RAM (to some extent). Generally, version 24.11 and 24.12 stress tests (blend) seem to be harder on A64 than 23.8...

2005-06-13, 14:12	#15
db597 Jan 2003 7×29 Posts	This seems to be a common problem. I've seen long posts about this over at the [H]ardOCP forums. It was even suggested that there's a bug in Prime95, but that's not true. Prime95 goes further than any other program to stress test the CPU and memory subsystem. The code in Prime95 has been tried and tested for a long time, on many CPUs. It's worked on Opteron and A64 up to now as well. As Cruelty mentioned, a BIOS update might fix the problem. I have a similar issue with one of my P4s. I've used that P4 CPU in a different board before with no problems. But when I swapped it over to an Asrock board, I couldn't get it stable unless I underclocked the CPU by about 5% and turned off "V-Link" in the BIOS - it's still running on that today. Maybe the hardware quality control at Asrock wasn't up to the standards required by Prime95.