Strange Problem 500FSB+ E8400

[zfooz]

Orthos stable 16+ hours at 500FSB * 8 and *7.5
but when I load up 3d games , and I tried different games, they crash fairly quickly.
Has anyone else experienced this?? I tried many voltage settings..
IT must be something with corruption on the PCI-E bus but what would be causing that?

thanks!

 

[LOUISSSSS]

did u test with OCCT 2 hours of CPU + 2 hours of RAM? pass?

if the cpu/ram is stable then its probably the video card/video driver... try a reinstall

 

[zfooz]

I did 1 hour of OCCT RAM test.. I did 16 hours of Orthos Blend test with no errors.
I really don't think it's the Video card or drivers, but I'm setting it back to 450 * 8.5 which I know it works at (48+ hrs of testing and playing games for days..)
and i'll see what happens.

I did notice one thing.. It used to be stable at 1.3875v at 4.05ghz and now it takes 1.4v , so something changed.. I guess with the burn in it needed a little more volts, but I'm back down to 1.3v @ 3.8ghz . stupid to have .1V just to get 250mhz more so it's staying at 3.8 but I really would love the 500FSB to work with games.. There must be something I'm missing here... How does the FSB effect the PCI-E bus and the video card.. It shouldn't really right? PCI-E is locked at 100mhz

 

[Gannon]

Originally posted by: zfooz
I did 1 hour of OCCT RAM test.. I did 16 hours of Orthos Blend test with no errors.
I really don't think it's the Video card or drivers, but I'm setting it back to 450 * 8.5 which I know it works at (48+ hrs of testing and playing games for days..)
and i'll see what happens.

I did notice one thing.. It used to be stable at 1.3875v at 4.05ghz and now it takes 1.4v , so something changed.. I guess with the burn in it needed a little more volts, but I'm back down to 1.3v @ 3.8ghz . stupid to have .1V just to get 250mhz more so it's staying at 3.8 but I really would love the 500FSB to work with games.. There must be something I'm missing here... How does the FSB effect the PCI-E bus and the video card.. It shouldn't really right? PCI-E is locked at 100mhz

IT most likely has to do with the PCI express frequency, do you have features in the bios to change it?

 

[M1A]

Why are you lowering the mult to get to 3.8Ghz? Most will do 3.8 at stock 9 mult. I agree with gannon you have to be sure the PCI is locked at 100. What direct x are you running and what vid card? Are you running 32 or 64 bit? Vista or XP?
If you are running with lower mult to end up with a higher FSB I think you will not see any difference well mot much depending on what application you are running. Like I said most of these 8400's do 4GHZ on air. But you are correct I found if its stable for hours running orthos you should be able to run games so check the PCI freq..........

 

[Gillbot]

Probably overclocking the PCIe frequency.

 

[devilchrist]

try Primeing with large fft. see if it passes for 6 hours at least.

when I playing Crysis. it would crash unless it was large fft stable.

it's the only thing I use for stability. If it passes Large FFt it passes everything else for my computer

 

[zfooz]

Lowering the multiplier because I need over 1.4volts to do 4ghz stable and I don't want to push it.. 3.8ghz is 1.3v
Running XP SP3 32 bit, with Radeon HD3870
PCIe is locked at 100mhz so the BIOS says.. I don't have any other way of testing it, if it was the PCIe then 450mhz bus wouldn't work either.
I did prime blend for 16 hours so it hit them all.
I think it must be something with the southbridge not liking to talk to the northbridge at that bus speed, that's all I can think so maybe I have to mess with the memory timings because I believe it uses those also for DMA through PCI bus but i'm not positive. I don't think I'm the only one having this problem, I posted in another forum and someone said it was happening to a lot of others who are prime stable at high FSB but can't run 3d apps.

 

[Shimmishim]

Prime blend test is not the same as largeFFT test.

Try largefft test for at least 6 hours... 12 hours if you can.

man, these 8400/8500's are beastly!

also what mem settings are you using, mem modules and what motherboard?

these all are going to factor into your 500+ mhz fsb stability

 

[zfooz]

Lol.. well, after doing a lot of tests at 4ghz it now takes 0.0250 more volts than it did previously to run at 3.8ghz which is weird.. usually processors take less after they have been run a while..

I'll try large FFT to test the ram. Ram 5-5-5-15, Corsair PC8500 1066mhz two 1gb sticks
Motherboard is ABIT IX38

Prime/orthos blend does everything from 8K to 8192K fft or even higher I think so 16 hrs of that should be sufficient..

 

[aigomorla]

Originally posted by: zfooz
Lol.. well, after doing a lot of tests at 4ghz it now takes 0.0250 more volts than it did previously to run at 3.8ghz which is weird.. usually processors take less after they have been run a while..

I'll try large FFT to test the ram. Ram 5-5-5-15, Corsair PC8500 1066mhz two 1gb sticks
Motherboard is ABIT IX38

Prime/orthos blend does everything from 8K to 8192K fft or even higher I think so 16 hrs of that should be sufficient..

mmmm....

under different situations id be worried about that. Because its a sign of board degradation.

However since your board is brand new, i have no idea in why its doing that.

 

[error8]

Originally posted by: zfooz
Lowering the multiplier because I need over 1.4volts to do 4ghz stable and I don't want to push it.. 3.8ghz is 1.3v
Running XP SP3 32 bit, with Radeon HD3870
PCIe is locked at 100mhz so the BIOS says.. I don't have any other way of testing it, if it was the PCIe then 450mhz bus wouldn't work either.
I did prime blend for 16 hours so it hit them all.
I think it must be something with the southbridge not liking to talk to the northbridge at that bus speed, that's all I can think so maybe I have to mess with the memory timings because I believe it uses those also for DMA through PCI bus but i'm not positive. I don't think I'm the only one having this problem, I posted in another forum and someone said it was happening to a lot of others who are prime stable at high FSB but can't run 3d apps.

So when you lower your multiplier and increase FSB, the cpu clocks higher but with lower Vcore?

 

[dwcal]

I had a similar problem with my E6300 at 445FSB. CPU is stable with Prime and OCCT, and I can run loops of 3dMark06, but SecondLife would hard-lock the system after a minute. It was no problem at stock speed. I don't game on this one so I couldn't test a whole lot of 3d games on it, and it just has a 7100GS anyway. I never did figure it out, but one of the updates to the client eventually fixed it.

 

[M1A]

I do not believe that also error8. Zfooz can you explain exactly your mult settings?

 

[zfooz]

No no..
It was running at 450x9 and 500*8 but it needs 1.4125v, so i changed the multiplier to 500x7.5 and I need 1.3v or 400*8.5 I need 1.3125v but previously I only needed 1.2875v for 450x8.5 but now it needs the 1.3125 to be stable.. Like a month ago it was stable 36 hrs of orthos @ 450*8.5 with 1.2875v but it won't run for 5 mins at 1.2875 now.. works great at 1.3125
I put it back to 450x8.5, 1.3125v, and i'm not having any 3d issues for a few hours at least now..

So I guess there are two issues.. Why do I need more voltage after burning in for a few months, and why 500FSB crash my 3d games even though the CPU is fine?

Do any of the memory timings have anything to do with link between the NB/SB ? Searching the web I noticed someone changed some RAM settings and that helped the 3d games work at higher FSB although prime/orthos was stable.

My board and CPU are less than 3 months old.. I had the CPU running at 452*9 with 1.4125v for a while but temps were WAY too high.. Like 30C left to TJmax even with Tuniq 120 cooler.. Don't know why they are so high.. grr.. Even now stilll 40C to Tjmax , or about 60-65C under load, or 55-60C depending on which program I use.. that's why I just report dist. to tjmax temps because those are always the same no matter what program.

 

[Martimus]

Originally posted by: zfooz
Why do I need more voltage after burning in for a few months

While I am not sure this is the issue, I used to stress test electrical components, and the breakdown region of the transistors will creep up over time; especially if you increase the current flowing them. This means that it will require higher voltage to properly "switch", this higher voltage in turn will increase the current flow which will hasten the process of this breakdown region creep. It was a regular and irreversible phenomenon in the zener diodes I tested; but I don't hear much about it with microprocessors.

 

[M1A]

I bet from reading this again you are having heat issues due to your room ambient being higher, making your case temp higher, making your load temps higher, causing you to have to raise your voltages to run stable, you get it...........
Have you been checking your temps close?
Just a thought

 

[zfooz]

the room temp isn't that much different from day to day.. maybe maximum of 1.5C difference at any given time.. I don't think that is enough to raise the voltage that much.. I'll try lapping the cooler and putting it back on and see if that makes any difference in the stability. I think my cpu cooler and the processor are not flat

 

[jgigz]

Originally posted by: zfooz

Running XP SP3 32 bit, with Radeon HD3870

I've heard about some people having issues with SP3 and ATI drivers, might want to look into that, if everything else seems stable