Unofficial Wolfdale E8400/E8500 OC Reports

[Zenoth]

Good stuff I'd encourage you to still make a Blend run to be sure. Memtest is not the end all to memory stability. Windows can still find ways to make Memtest stable memory unstable.

Thanks

I'm currently doing a Blend test, nice coincidence. It's been on for 6 hours already, no errors so far. But I've also tested my memory with HCI (the Windows variant of Memtest), filling up the whole 3.3GB of memory detected under the x86 environment (I'll try with Vista x64 later next week), for a solid seven hours, no errors.

Add to that a two hours Team Fortress 2 session (while Blend testing in Orthos) and, after that, a two hours session in UT3 (which is a multi-threaded game), and so far no crashes, no errors, pure bliss. Right now I believe that I've attained absolute stability.

The only thing that I believe could have been better is the vCore, but I'm still content with it at 1.328v (I don't know why CPU-z reported it at 1.312v in the picture, I just noticed it... when I closed and re-opened CPU-z it came back at 1.328v).

 

[harlam357]

Good to hear... and I'm one to talk really. I'm still having issues with my rig after passing 36 hours of Prime Blend. I don't think it's my cpu though, I think it has more to deal with the fact that I'm not familiar with my board yet.

Me and mobos kinda take a while to get acquainted... makes me miss my DFI nF4. Except the fact that this rig is like waaaaayyyyy faster.

 

[Zenoth]

Good to hear... and I'm one to talk really. I'm still having issues with my rig after passing 36 hours of Prime Blend. I don't think it's my cpu though, I think it has more to deal with the fact that I'm not familiar with my board yet.

Me and mobos kinda take a while to get acquainted... makes me miss my DFI nF4. Except the fact that this rig is like waaaaayyyyy faster.

Still having problems after a stable 36 hours Blend test? That's... rare.

Maybe your Power Supply is going bunkers on you? Or else what could it be... hmmm, memory timings maybe and/or voltage... or heat... or one of the modules is dying... or your motherboard isn't compatible with it (or the other way around)... or erm... gah, so many possibilities man!

 

[Anubis_386]

Mine failed at 3h so the voltage went up to 1.38125v .. 8h Orthos Stress CPU stable now even tho it gets errors on all the other tests ..

I've been reading things about these chips dying under (what was normal with Conroe) high CPU VTT (FSB Termination) voltage .. Anyone know anything about this cause i've got mine set to 1.46v in BIOS ..

 

[Reefa_Madness]

Mine failed at 3h so the voltage went up to 1.38125v .. 8h Orthos Stress CPU stable now even tho it gets errors on all the other tests ..

I've been reading things about these chips dying under (what was normal with Conroe) high CPU VTT (FSB Termination) voltage .. Anyone know anything about this cause i've got mine set to 1.46v in BIOS ..

I believe that guys are looking at absolute vcore and thinking that the voltage looks good.

If you look at % increases over default voltages, then the picture gets a little clearer, or at least more informative.

If, for example, your E8400/E3110's default voltage is 1.1v and you are running 1.46v, then that .36v represents a 32.73% increase. The same increase to a S478 processor with default voltage of 1.55v would yield an ending vcore of 2.057v and we would all call you insame for feeding it that much. Another comparison would be an E6420 Conroe with say vcore of 1.325v (the high range for these)...the equivalent voltage (same 32.73% increase) would be 1.759v. Again, way more than most of us would feed them unless under Xtreme cooling.

I caution everyone to keep these % in mind when pumping the voltage into these new procs. The problem is that they respond/scale so well with voltage so we say "heck, its only 1.5v and my temps are OK, so all is well", but in fact, all may not be well if you are wanting to keep this proc alive for longer than a few months. These 45nm process chips appear to be way more delicate than the 90nm and 130nm procs of yesteryear.

Build yourself an Excel spreadsheet or keep a pencil and paper nearby when you are upping your vcore and do the math first and save the proc in the process.

 

[Blkout]

Mine failed at 3h so the voltage went up to 1.38125v .. 8h Orthos Stress CPU stable now even tho it gets errors on all the other tests ..

I've been reading things about these chips dying under (what was normal with Conroe) high CPU VTT (FSB Termination) voltage .. Anyone know anything about this cause i've got mine set to 1.46v in BIOS ..

I've only needed 1.1v CPU VTT to accomplish all of my overclocks. In my opinion that adjustment doesn't need adjusting, it makes little to no difference in stability. You should be adjusting the northbridge voltage or the GTL percentage.

 

[Reefa_Madness]

I've only needed 1.1v CPU VTT to accomplish all of my overclocks. In my opinion that adjustment doesn't need adjusting, it makes little to no difference in stability. You should be adjusting the northbridge voltage or the GTL percentage.

For what kind of overclocks have you not had to adjust your cpu's voltage. There are reasons why at some point increased vcore is required for stability as your processor clocks increase. It is grounded in science, and not simply a matter of opinion. Take a peek at this article. It used to have some visual representations, but it appears that the links to those is dead now. The text is still good, however.

http://www.hardwareanalysis.com/content/article/1482/

The questions that each must ask is how much are you willing to increase your voltage in order to obtain higher clocks...what is a given is that there will be a point where the increase in clock speed requires voltage increases in order to have reliable stability.

 

[harlam357]

Still having problems after a stable 36 hours Blend test? That's... rare.

Maybe your Power Supply is going bunkers on you? Or else what could it be... hmmm, memory timings maybe and/or voltage... or heat... or one of the modules is dying... or your motherboard isn't compatible with it (or the other way around)... or erm... gah, so many possibilities man!

I agree... it is rare. I've never had a rig handle that much Prime and then still be flaky afterwards.

Could be heat... when I Primed for that long it was cold in the house, I personally like clocking in the summer so I can be sure I'm not getting better stability because of lower ambient.

PSU... I doubt it. This Corsair 650TX seems right on spot when I check the voltage readings in bios. Really nice PSU IMO... right up there close with the PCP&C 510-SLI I swapped out from my Opteron rig.

Memory... it's GSkill HZ, they are known compatible with the IP35-Pro. However, I won't rule out RAM timings as I've got most of the subtimings set to Auto right now. But 4-4-4-12 @ 900MHz should not be out of the range of these dimms.

Last adjustments I've made were to bring up the CPU VTT from 1.10v to 1.17v, raise the ICH from 1.05v to 1.09v, and raise the ICHIO from 1.50v to 1.55v. I really think it's a board/chipset issue that didn't manifest itself under Prime. But when I run Folding@Home it comes out. We'll see, if I have any further reboots I'm thinking of swapping the dimms out for a set of Ballistix 1066 I've got in another rig, drop the timings to 5-5-5-15 for a while and see if that does the trick.

Just a note on CPU VTT to Blkout... for me to boot my setup @ 460FSB on the lowest possible vcore, raising the CPU VTT from 1.10v to 1.17v did the trick... so in my case it gave me a bit more stability in that configuration. I guess it also depends on how high the MCH is running since I've read that those two voltages are interdependent.

http://www.hardwareanalysis.com/content/article/1482/

Edit: That's a good read Reefa.

 

[Big Lar]

harlam357
I think you may be pumping too many volts into the HZ's,maybe... Mine will run at 910,(455x2) at 2.025v. Mine were made in 2007, you can check your year by looking at the first #/#'s of the serial #, which is the bottom line.
Larry

 

[Zenoth]

harlam357, yup, loosen the timings to 5-5-5-'x', perhaps x=18 to 24, try that (OCZ themselves recommend using a tRAS of 18 or more for over-clocking purposes). And by going from 4-4-4 to 5-5-5 you wouldn't notice any tangible differences in gaming, only in synthetic benchmarks, and by around 2% or 3%. If that mere performance "loss" is the price for absolute stability, I'd go for it any day (that's what I did myself, my memory is rated at and does work at 4-4-4-15, but I won't stay at such timings for OC'ing or else I'd better stick with the stock settings). And another advantage with loosing up the timings is the reduced necessity to increase the vDIMM.

 

[Blkout]

For what kind of overclocks have you not had to adjust your cpu's voltage. There are reasons why at some point increased vcore is required for stability as your processor clocks increase. It is grounded in science, and not simply a matter of opinion. Take a peek at this article. It used to have some visual representations, but it appears that the links to those is dead now. The text is still good, however.

http://www.hardwareanalysis.com/content/article/1482/

The questions that each must ask is how much are you willing to increase your voltage in order to obtain higher clocks...what is a given is that there will be a point where the increase in clock speed requires voltage increases in order to have reliable stability.

CPU voltage(Vcore) and CPU VTT are two different things. I said I haven't had to increase my CPU VTT beyond the default 1.1v. I've had my Vcore as high as 1.5v for my E8400 while testing, but currently running 1.39v.

 

[Blkout]

Just a note on CPU VTT to Blkout... for me to boot my setup @ 460FSB on the lowest possible vcore, raising the CPU VTT from 1.10v to 1.17v did the trick... so in my case it gave me a bit more stability in that configuration. I guess it also depends on how high the MCH is running since I've read that those two voltages are interdependent.

I've found the CPU VTT to be worthless, although the MCH voltage is absolutely crucial to high FSB overclocks.

 

[Anubis_386]

I've lowered my VTT to 1.32v in BIOS so with droop its at 1.23v ... I've changed some of my other voltages slightly as well

vCore - 1.38125v
CPU PLL - 1.64v
NB - 1.63v

Its stable through Orthos's Stress CPU with Gromacs Core but not on the Small, Large or Blend tests .. My RAM is the problem, its fails MemTest86+ v2.01 with errors in tests 6 and 7 .. Its got 2.32v and timings of 5-5-5-15 but still its not stable .. I'm going to get a better kit, DDR2-1200 with D9s

 

[Monty_Python]

My E8400's retail box stated max Vcore of 1.225v...so I'm happy at 4.005Ghz and 1.38vCore...thats only 13% over recommend max voltage.

 

[Reefa_Madness]

CPU voltage(Vcore) and CPU VTT are two different things. I said I haven't had to increase my CPU VTT beyond the default 1.1v. I've had my Vcore as high as 1.5v for my E8400 while testing, but currently running 1.39v.

My bad...I had my brain stuck on VID for some reason (my previous post perhaps) and that's all I saw.

One of those cases where you "see" something that isn't what's really there, even if you look at it more than once.

I guess I should have had that second cup of coffee this morning...

 

[Reefa_Madness]

Edit: That's a good read Reefa.

Glad you found it beneficial, at least something worthwhile came out of that otherwise stupid post.

I refer to it often, as it puts it in terms that I can understand/relate to. Too bad the visuals no longer are displaying, as they were helpful.

 

[NobodyNowhere]

Hi,

I have been reading this thread from the start and a very few have a E8500 .. Well i have ordered mine and i have spent almost 450 USD on a great Cooling kit and my goal is to get the CPU to 4.5Ghz to use 24/7 ... Do you guys think its possible ( Well yes it is possible but will the chip run and not die for 6-8 months?! )

Thanks

 

[Blkout]

Good stuff I'd encourage you to still make a Blend run to be sure. Memtest is not the end all to memory stability. Windows can still find ways to make Memtest stable memory unstable.

I concur. I prefer blend to small FFT's just for this reason. It's a better measure of overall stability.

 

[Blkout]

Hi,

I have been reading this thread from the start and a very few have a E8500 .. Well i have ordered mine and i have spent almost 450 USD on a great Cooling kit and my goal is to get the CPU to 4.5Ghz to use 24/7 ... Do you guys think its possible ( Well yes it is possible but will the chip run and not die for 6-8 months?! )

Thanks

4.5GHz is not going to be possible for 24/7 use...sorry. The best you could possibly hope for is 4.3GHz and even then, I wouldn't feel comfortable with those voltages. 4.2GHz is certainly attainable and that's where I run mine 24/7.

 

[Dawgdoc]

4.5GHz is not going to be possible for 24/7 use...sorry. The best you could possibly hope for is 4.3GHz and even then, I wouldn't feel comfortable with those voltages. 4.2GHz is certainly attainable and that's where I run mine 24/7.

Did you see he said he has an E8500 and not an E8400. With the right cooling i think thats a realistic possibility.

8500 has starts off higher than the 8400 and also has the 0.5 multi option.