Help me stabilize my E8400 @ 4 GHz

C

cold_steel

Gawd
2FA
Joined
Jun 27, 2004
Messages
581
Hey guys, I recently picked up a DFI LT X48-T2R mobo with an E8400 and some corsair PC2-8500 dominator ram. I previously had an Opty 170 that I ran at 2.8 GHz, so I'm newly back in the Intel camp and need to get up to speed again. I have been reading the various guides/threads around here and I seem to be requiring a lot more juice to get this thing to pass prime than most, like close to 1.45v and I see that that isnt recommended. It seems to get pretty hot, but not dangerously so- around 65c on realtemp. It seems *almost* stable bc it will run games perfectly 99% of the time (then maybe a random hang once in a blue moon) but now if I leave it running prime I come back to a shut down system, I'm guessing thats my board shutting down bc of temps? seems odd bc I've never seen anything past 67c. Do I need to mess with the GTL voltages, and where should I be putting the CPU VTT volts? Thanks for the help in advance- Oh I'm using a TRUE 120 on it.
 
It seems *almost* stable bc it will run games perfectly 99% of the time (then maybe a random hang once in a blue moon) but now if I leave it running prime I come back to a shut down system, I'm guessing thats my board shutting down bc of temps?
Click to expand...

Your board will not just shut down the whole system because of the temps. Something is failing. What is the highest speed you can pass ~12 hours of Prime 95 small FFTs at?
 
What PSU are you using? To get to 4Ghz + you typically need a quality unit.

Also have you bumped the voltage on memory and/or the northbridge?
 
ocellaris said:
Your board will not just shut down the whole system because of the temps. Something is failing. What is the highest speed you can pass ~12 hours of Prime 95 small FFTs at?
Click to expand...

Actually, it could very well be shutting down due to high temps. As you probably know, most boards have a setting that automatically shuts down the computer if the CPU hits a certain temperature, and since the OP was just leaving the system unattended, it's possible that the threshold temperature was reached.

OP, it's possible that you simply aren't able to hit 4GHz with your setup. 1.45v is getting into the dangerous area for a Penryn chip, so if it isn't stable I would advise that you back down to something that is stable, or upgrade your cooling system to something that would be able to better cool your CPU.
 
Zero82z said:
Actually, it could very well be shutting down due to high temps. As you probably know, most boards have a setting that automatically shuts down the computer if the CPU hits a certain temperature, and since the OP was just leaving the system unattended, it's possible that the threshold temperature was reached.

OP, it's possible that you simply aren't able to hit 4GHz with your setup. 1.45v is getting into the dangerous area for a Penryn chip, so if it isn't stable I would advise that you back down to something that is stable, or upgrade your cooling system to something that would be able to better cool your CPU.
Click to expand...

Well he can check the BIOS, however if he is running 65C I do not think it would be getting into thermal shutdown for quite a while...
 
The temp is fine, you are 30C off TjMax, I run mine 25C off TjMax (70C) and have had no issue at 10C from TjMax.
You shouldnt put 1.45V into it though, the established safe figure is around 1.38V although I run mine at 1.4V (4.13GHz) cos I have a deathwish.
The chips are known to irreparably degrade if pushed too far.

Your problem may be your Ram.
Make sure the timings are set to Auto / SPD in the CMOS
The default voltage for the memory is often set to 1.8V which is enough to boot your system but usually the memory will need a higher voltage under load, especially when overclocked.
If that fails try a slightly higher MCH voltage.

Check the specs of your ram to find the maximum voltage and dont exceed it.
You probably wont need to run at its max voltage, find the lowest voltage that works.
My Ocz Platinum DDR2-1000 will do 913MHz at 1.9V rock stable, its rated as 2.1V.
Your ram is running at 888MHz while you are clocked to 4GHz (on a x9 CPU multi).
fyi
 
My old pc2-8500 dominators required 2.2 volts @ 500fsb, double check volts on nb, sb and keep cpu vtt voltage under 1.2v if you dont wanna kill your cpu.
 
I'm running the Dominator 8500 ram at stock close to stock volts/settings, 2.10 volts at 1069 (534.5). Timings are 5-5-5-15, stock settings. The divider I'm using actually lets the FSB be at 445 for the 4.0 GHz OC while still allowing the ram to be just *slightly* OC'd at like 1069 mhz DDR. I believe it is like 5/6. Do I need to mess with the GLT volts? Where should I be trying to stick with the NB volts?
 
cold_steel said:
Hey guys, I recently picked up a DFI LT X48-T2R mobo with an E8400 and some corsair PC2-8500 dominator ram. I previously had an Opty 170 that I ran at 2.8 GHz, so I'm newly back in the Intel camp and need to get up to speed again. I have been reading the various guides/threads around here and I seem to be requiring a lot more juice to get this thing to pass prime than most, like close to 1.45v and I see that that isnt recommended. It seems to get pretty hot, but not dangerously so- around 65c on realtemp. It seems *almost* stable bc it will run games perfectly 99% of the time (then maybe a random hang once in a blue moon) but now if I leave it running prime I come back to a shut down system, I'm guessing thats my board shutting down bc of temps? seems odd bc I've never seen anything past 67c. Do I need to mess with the GTL voltages, and where should I be putting the CPU VTT volts? Thanks for the help in advance- Oh I'm using a TRUE 120 on it.
Click to expand...

That's ridiculously high. With Intel, there really isn't much to complicate it besides GTL - but this really is only useful if you are trying to lower the required vcore. I.e. you are in the last stage of OCing, and just trying to shave off vcore. Particularly with dual cores, all you need is vcore. That's it.

Raising VTT/NB/etc. is not necessary. I'm OCing at VTT set at stock, 1.10v. Likewise with the NB. Fact is - you may simply have a bad chip. I knew my first chip was bad when I took over 1.36Vcore and I still was stuck around 3.93Ghz. My 2nd chip did 4Ghz, OCCT stable, at the voltage you see in my sig, 1.296vcore.

What batch/pack date do you have?
 
cold_steel said:
I'm running the Dominator 8500 ram at stock close to stock volts/settings, 2.10 volts at 1069 (534.5). Timings are 5-5-5-15, stock settings. The divider I'm using actually lets the FSB be at 445 for the 4.0 GHz OC while still allowing the ram to be just *slightly* OC'd at like 1069 mhz DDR. I believe it is like 5/6. Do I need to mess with the GLT volts? Where should I be trying to stick with the NB volts?
Click to expand...

Run memory 1:1 with the FSB.
You will find no benefit clocking the ram higher, concentrate on CPU overclock and memory timings.
 
1.45v for 4ghz is crazy. I can run 4ghz with 1.24v with my e8400. You may just have a bad chip. You shouldn't have to do much tweaking in order to hit 4ghz with the e8400. Your problem could be the northbridge. I would recommend using a mulitplier of 9 and bus of 450 (4.05ghz). If this runs ok at 4ghz it may tell you that you will have to increase the northbridge voltage in order to run higher than 500mhz bus. I've got a DFI DK P35 and have my northbridge at 1.37v running at little over 500mhz.

I would say that it is a bad chip or either a very bad stepping if it can't do 4ghz with 1.38v.
 
Mav451 said:
Raising VTT/NB/etc. is not necessary.
Click to expand...
Not true in my experience. For example it is common knowledge for Abit IP35-Pro owners that you have to raise northbridge (MCH) voltage 10%-15% to stabilize the board when running higher FSB speeds ( >450 mhz). My old Asus P5B-Deluxe also required bumping up the northbridge voltage to stabilize it. I recently built a system for a friend based on an e8400 and the only way to reach stable 4.1(456x9) was to raise NB.

OP, keep in mind there are scattered reports in the hardware forums of permanent processor degradation for users running high voltages on these new Wolfdale processors. Intel recommends maximum voltage as 1.365v so it's best if you keep your vcore at or below this amount.

HB
 
HenryBravo said:
Not true in my experience. For example it is common knowledge for Abit IP35-Pro owners that you have to raise northbridge (MCH) voltage 10%-15% to stabilize the board when running higher FSB speeds ( >450 mhz). My old Asus P5B-Deluxe also required bumping up the northbridge voltage to stabilize it. I recently built a system for a friend based on an e8400 and the only way to reach stable 4.1(456x9) was to raise NB.

OP, keep in mind there are scattered reports in the hardware forums of permanent processor degradation for users running high voltages on these new Wolfdale processors. Intel recommends maximum voltage as 1.365v so it's best if you keep your vcore at or below this amount.

HB
Click to expand...

I hope that in your rush to "correct me", you did notice that we are discussing DFI mobos?

My experience is that if you got a bad chip, throwing more voltage at it is fruitless. Especially NB. You can pretty much instantly tell if you got a good or bad chip, and if you gotta raise your NB just to stabilize it, well too bad then.

While I do appreciate the sentiments (I actually had an IP35-E for a good 30 seconds, before selling that to my friend), my experience is primarily DFI boards and has been for about 3 years. I think for mainstream OCing, throwing voltage at it is hardly the wise thing to do. If you are going for an XS record (5Ghz+) kind of thing, then it is necessary. For 24/7 OCs, it is not necessary from my experience.

*I also understand that "sensible or 24/7 OCs" may be an oxymoron for a DFI owner...well I'm in the minority then lol.
 
You must log in or register to reply here.
Back
Top