Quirks while testing for instability: Prime95

Antoneo

Diamond Member
May 25, 2001
3,911
0
0
Here's the computer that I am trying to overclock:

Sempron 64 Palermo 3100+ currently testing at 2610MHz w/1.50 vcore
XP-120 with 120mm Panaflo M1A
DFI NF3 250GB 2005/05/04 BIOS
1GB Corsair Value Select PC3200 RAM running at a 7/10 divider
ATI 9800 Pro
Antec TruPower 330 Watt
3x Seagate 'Cuda V 200GB

I haven't been able to get Prime95 to fail with a hardware/rounding error. I have only been getting freezing inside of windows while running Prime95 where I am forced to press the reset button. Thus, in my testing so far, I have never gotten a toture test failed with the cause being a hardware/rounding error. I have not seen a warning either, it has simply been a complete freeze of the system.

My overclocking experience with my sempron 2800+ computer has been hardware/rounding errors while running Prime95 when there wasn't enough vcore, or when I pressed the FSB too high. Those errors were an indication to up the vcore or reduce the FSB, but on the Sempron 3100+ machine I'm only getting freezes.

I've run the Sempron 3100+ machine for about 8 hours at 1.475 vcore before it froze up. I rebooted and am testing at 1.500 vcore for the past 6 hours today. I've tested before at 1.500 vcore for about 10 hours without any crashes before trying lower voltages.

Configuration for the other computer:

Sempron 64 Palermo 2800+ @ 2400MHz w/1.55 vcore
XP-90 with 92mm Panaflo LBX
DFI NF3 250GB 2005/05/04 BIOS
1GB Corsair Value Select PC3200 RAM at a 2/3 divider
ATI 9800 Pro
Fortron 300 Watt
1x Seagate 'Cuda V 200 GB

I've taken screenshots of CPU-Z, speedfan, and Prime95 on my Sempron 3100+ computer:
pic1.JPG
pic2.JPG

One of the odd things I've noticed is that in the memory tab for CPU-Z, the information for Frequency or FSB:RAM is missing. At stock settings, these were both reported but now they are not. Has anyone experienced this? On my Sempron 2800+ machine, the 2/3 divider is correctly reported.

The RAM is running at 203MHz according to my calculations and its at 2.7 vdimm. I don't remember doing memtest at the current settings because I was trying to find a stable CPU configuration.

So my main question is whether or not such freezing behavior (without actual Prime95 reported errors) is normal. Do you guys think it is pointing towards an inadequate PSU?
 

1Dark1Sharigan1

Golden Member
Oct 5, 2005
1,466
0
0
Originally posted by: Antoneo
Here's the computer that I am trying to overclock:

Sempron 64 Palermo 3100+ currently testing at 2610MHz w/1.50 vcore
XP-120 with 120mm Panaflo M1A
DFI NF3 250GB 2005/05/04 BIOS
1GB Corsair Value Select PC3200 RAM running at a 7/10 divider
ATI 9800 Pro
Antec TruPower 330 Watt
3x Seagate 'Cuda V 200GB

I haven't been able to get Prime95 to fail with a hardware/rounding error. I have only been getting freezing inside of windows while running Prime95 where I am forced to press the reset button. Thus, in my testing so far, I have never gotten a toture test failed with the cause being a hardware/rounding error. I have not seen a warning either, it has simply been a complete freeze of the system.

My overclocking experience with my sempron 2800+ computer has been hardware/rounding errors while running Prime95 when there wasn't enough voltage, or when I pressed the FSB too high. Those errors were an indication to up the voltage or reduce the FSB, but on the Sempron 3100+ machine I'm only getting freezes.

I've run the Sempron 3100+ machine for about 8 hours at 1.475 vcore before it froze up. I rebooted and am testing at 1.500 vcore for the past 6 hours today. I've tested before at 1.500 vcore for about 10 hours without any crashes before trying lower voltages.

Configuration for the other computer:

Sempron 64 Palermo 2800+ @ 2400MHz w/1.55 vcore
XP-90 with 92mm Panaflo LBX
DFI NF3 250GB 2005/05/04 BIOS
1GB Corsair Value Select PC3200 RAM at a 2/3 divider
ATI 9800 Pro
Fortron 300 Watt
1x Seagate 'Cuda V 200 GB

I've taken screenshots of CPU-Z, speedfan, and Prime95 on my Sempron 3100+ computer:
pic1.JPG
pic2.JPG

One of the odd things I've noticed is that in the memory tab for CPU-Z, the information for Frequency or FSB:RAM is missing. At stock settings, these were both reported but now they are not. Has anyone experienced this? On my Sempron 2800+ machine, the 2/3 divider is correctly reported.

The RAM is running at 203MHz according to my calculations and its at 2.7 vdimm. I don't remember doing memtest at the current settings because I was trying to find a stable CPU configuration.

So my main question is whether or not such freezing behavior (without actual Prime95 reported errors) is normal. Do you guys think it is pointing towards an inadequate PSU?

Are you using Blend mode?

Your temps look okay so shouldn't be a heat issue . . . If you are using blend mode, freezing tends to suggest memory instability . . . but @ 203MHz value ram should be just fine . . . maybe loosen up timings and try again?

Your CPU doesn't look like it's unstable though if you can run Prime 8 hours with no round error . . .
 

dunkster

Golden Member
Nov 13, 1999
1,473
0
0
Instability can be due to low Vcore, poor voltage regulation (voltage droop under load) or due to incorrect RAM settings.

Don't skip the Memtest86-testing for RAM stability.

For example, I can run Prime95 for an hour at 1T command rate, but my system will crash (in the real world) as soon as a significant page-write is encountered. Memtest86 (Test#8) revealed that 2T command rate is required for complete stability.

Hope this helps!
 

Thor86

Diamond Member
May 3, 2001
7,888
7
81
Any type of freezing or hardlock is usually a memory issue. Make sure your memory settings/timings in your bios are set correctly, and yes, do test your memory for errors using Memtest as the previous poster posted.
 

Antoneo

Diamond Member
May 25, 2001
3,911
0
0
I ran Prime95 for 7.5 hours using Large FFTs (Max heat/power and some RAM tested) without any freezes. I stopped that, placed my system configuration back to stock (200 FSB with 9x multiplier, 1:1 DRAM:FSB, 1.4 vcore, 2.6 vdimm) and ran memtest86 for 6 hours without any errors. The RAM timings for both the Prime95 and memtest runs were the same as SPD settings: 2.5-3-3-8.
 

calguy

Junior Member
Nov 26, 2005
7
0
0
I had the same problems with my Sempron 3000+ and was able to fix it by setting Tras to 10 (I think it's related to the nForce 3 chipset).