GA-EP45-Extreme, E8500 4.2ghz, 533mz FSB

[cyclo]

Hi all. Just got myself one of these boards and i've been trying to get it stable with a 4.2ghz oc (had the same CPU running fine at 4ghz on my Asus P5N-D)

I really don't have a lot of knowledge when it comes to OC'ing but what i'd like to achive is to get my FSB to 533 so my 1066mhz RAM is at a 1:1 ratio.

It doesnt seem to matter what I try I always end up getting an error in orthas. My temps are fine, idles at 33c and never goes above 60c at full load.

I've tried various vcore settings from 1.3-1.4, but theres so many other settings on the board that I just dont understand. I've played around with the CPU termination settings and MCH core settings, set both of those at 1.2v and went all the way up to 1.3v. Set CPU PLL to 1.5v, but when I start messing around with either of those or any other voltage settings, windows fails to boot.

Right now i have it running at 425x9.5 and my ram at 510mhz with everything on auto except vcore, which is showing as 1.31v in windows. Testing it out in orthas as i type this, but ideally I want my FSB higher.



does anyone have any experiance with this board? and its it really that important to get a 1:1 ratio? right now its 5:6

[sharpbiohazard]

Re: GA-EP45-Extreme, E8500 4.2ghz, 533mz FSB

While I don't know about your board, I have an E8500 as well and I've heard that anything above 1.4v on vcore= instant death. Just FYI

[Psycho101]

Re: GA-EP45-Extreme, E8500 4.2ghz, 533mz FSB

Your current settings for PLL and CPU Termination are dangerously high! The first thing you need to do is take them off auto and set them manually. Suggested limits are 1.61V for PLL and 1.4V for CPU Termination. 1.6V CPU Term is extremely dodgy indeed.

There's zero performance advantage from having a 1:1 FSB to RAM ratio. The performance advantage is due to the RAM at 533 FSB 1:1 running 23MHz faster than it currently is, plus a small performance boost from the FSB itself being 23MHz faster too.

I would suggest setting PLL to 1.57V/1.61V and trying values from 1.3V to 1.38V for CPU Termination. Also at 533 FSB, you'll probably need to lower MCH Reference.

After setting PLL to 1.61, set CPU termination to 1.2V. Now set MCH Ref manually to 0.760. Now use the Page UP/Down key to set MCH ref to 0.740V (1 tick down). You can now set CPU termination to the voltage needed (as said, try 1.30V-1.38V with max of 1.4V if you really need it). Is you still get failures, set CPU Term to 1.20V again and lower MCH ref another tick to 0.720V. Again, raise CPU Termination back up. You will notice that as you increase CPU Termination, MCH Ref rises. This is normal. MCH ref will only be eg 0.740V at 1.2V CPU Term, and is an indication of the ratio between CPU Term and MCH ref. The ratio between these values is more important than actual voltage values.

If you still aren't stable, then again lowering MCH Ref to 0.700V @ 1.2CPU Term is fine to try.

To emphasise how important MCH ref values are, take my system at 3.9GHz. At the stock MCH Ref of 0.760V and my needed CPU Termination of 1.4V, the system fails Prime 95 instantly. At 0.740V MCH Ref it fails within 15 minutes. At 0.720V it is somewhat stable, to around about 7 hours on blend or 3 hours on large fft. At this voltage, perfectly OK .rar files fail to decompress without errors such as "archive corrupt". At 0.720V @ 1.2V CPU Term (actual MCH Ref value is higher after it scales upward with CPU Term set to 1.4V) I am stable as stated in my sig, plus I can un-rar files to my hearts content.

Do be careful with VCore, but as long as cooling is ok and temps are roughly in check, 1.4V is fine. I'm running at 1.4625V in the BIOS and 1.424V in Windows (CPU-Z) without issue. 5 months of heavy usage so far and not a single blue screen, freeze or crash related to CPU or RAM. I've had GPU related crashes but only when finding my OC limits for my pair of 4850's.

To summarise, I'd set 1.4V or there abouts in the BIOS, with LLC on. Then set PLL to 1.61V, set CPU term to 1.20V temporarily while adjusting MCH Ref to 0.740V. Now, starting at 1.3V CPU Term, test with Prime/Orthos, upping CPU termination to a max of 1.4V. If all values fail, set CPU Term back to 1.2V, MCH Ref to 0.720V then try again with CPU Term values 1.30V-1.40V. Again, if it fails readjust MCH Ref to 0.700V and go again.

After achieving stability for a 12 hour run of prime, or what ever you deem stable enough, I would work on getting Vcore down as low as possible. Setting 1.4V is only to give you enough (possibly) to have that value not be a factor. That said, it is possible, though not likely that you might need more, so if all these steps fail, go for 1.45625 Vcore, check temps and start the whole thing again.

Edit! Search for posts by Boombastic sloth, D-an-W and Grishkafool. They all have threads here concerning E8400's, and the exact same principles apply to an E8500/E8600. All these threads are many many pages long (D-an-W's thread is 86 pages).

[cyclo]

Re: GA-EP45-Extreme, E8500 4.2ghz, 533mz FSB

Thanks for the detailed reply, really helpfull :) Just to clear it up in my head cos its a little bit complicated. I set cpu term to 1.20, mch ref to 0.740 then raise cpu term from 1.30 to 1.38 whilst keeping mch ref at 0.740? Then if neither of those voltages work, I lower ref down to 0.720 and try 1.30-1.38 cpu term again?

i dont need to keep lowering mch ref every time i raise cpu term up a notch? just after i've tried the 1.30-1.38 range right?

Also, can I change the cpu term and ref within windows using easytool? I always thought stuff like that had to be changed in the bios but i have no idea how this tool works

[Psycho101]

Re: GA-EP45-Extreme, E8500 4.2ghz, 533mz FSB

That's right. Set CPU Term to 1.2V, set MCH ref, then set CPu Term back to where you want it. You then don't touch MCH Ref again unless CPU Term of 1.3-1.4V all fail. Basically, any time you wish to alter MCH Ref, do so with CPU Term at 1.2V, then procede to increase CPU Term to what is needed.

I'd recommend strongly against using ET6 to change any settings. It's useful to use for screenshot demonstrating what voltages you have used, but always set things in the BIOS.

[cyclo]

Re: GA-EP45-Extreme, E8500 4.2ghz, 533mz FSB

Ok, think I got it now. Up to 0.740 MCH, 1.38 Term so far, longest I've managed in orthas is 20 minutes. Looks like I've got a lot of rebooting to do :)

[cyclo]

Re: GA-EP45-Extreme, E8500 4.2ghz, 533mz FSB

Well it looks I've found the sweet spot. CPU term = 1.30. MCH Ref = 0.720. Passed 11 hours in orthas so far!

vcore is at 1.376v in CPU-Z, guess I should starting playing around with that next

edit: I take that back, as soon I started using my pc again after leaving it overnight, the system crashed

[Psycho101]

Re: GA-EP45-Extreme, E8500 4.2ghz, 533mz FSB

Keep tweaking a little. Note that if you mean you got a boot loop after disconnecting the AC over night, that's normal. Turning the PSU off at the switch on the back for ~10-20 seconds should make it boot fine.

To play it safe, I'd tweak Static tRead in advanced RAM options to one number higher than it's currently set. The other advanced settings on auto should be fine.

If you have only tried 1.30V CPU termination with 0.720 MCH Ref, I'd try the range of 1.32-1.4V CPU Term. You may have MCH Ref in the right place now, and just need to give it a little more CPU Term to finish things off.

Can you describe the crash in more detail? Was it a boot loop, a series of error beeps, a freeze on loading Windows, a blue screen etc?

[cyclo]

Re: GA-EP45-Extreme, E8500 4.2ghz, 533mz FSB

What happened was Windows completly froze when I was editing this thread :) (still had orthas running in the background)

This is what i've done so far

(XX min/hour) = Duration in Orthas

VCORE 1.3760 (1.43750 bios)
CPU PLL 0.1600
MCH REF 0.7400

CPU TERM 1.30 = FAIL (15 min)
CPU TERM 1.32 = FAIL (20 min)
CPU TERM 1.34 = FAIL (6 min)
CPU TERM 1.36 = FAIL (25 min)
CPU TERM 1.38 = FAIL (53 min)
CPU TERM 1.40 = FAIL (2 min)

VCORE 1.3760 (1.43750 bios)
CPU PLL 0.1600
MCH REF 0.7200

CPU TERM 1.30 = FAIL (11 hours, seemed stable the windows froze)
CPU TERM 1.32 = FAIL (1 hour)
CPU TERM 1.34 = FAIL (19 min)
CPU TERM 1.36 = FAIL (15 min)
CPU TERM 1.38 = FAIL (unknown, blue screen
CPU TERM 1.40 = FAIL (16 min, windows froze)

VCORE 1.3760 (1.43750 bios)
CPU PLL 0.1600
MCH REF 0.7000

CPU TERM 1.30 = FAIL (8 hours, blue screen related to memory_management)


//ERRORS FOUND IN MEMTEST
//ADJUSTED TIMINGS (6-6-6-18) - MEMTEST PASSED
//RESTARTING AT MCH REF 0.7400


VCORE 1.3760 (1.43750 bios)
CPU PLL 0.1600
MCH REF 0.7400

CPU TERM 1.30 = (8 min)
CPU TERM 1.32 = (4 min)
CPU TERM 1.34 = (7 min)
CPU TERM 1.36 = (5 min)
CPU TERM 1.38 = (22 min)
CPU TERM 1.40 = (7 min)


VCORE 1.3760 (1.43750 bios)
CPU PLL 0.1600
MCH REF 0.7200

CPU TERM 1.30 = FAIL (19 min)
CPU TERM 1.32 = FAIL (19 min)
CPU TERM 1.34 = FAIL (16 min)
CPU TERM 1.36 = FAIL (16 min)
CPU TERM 1.38 = FAIL (18 min)
CPU TERM 1.40 = FAIL (20 min)

VCORE 1.3760 (1.43750 bios)
CPU PLL 0.1600
MCH REF 0.7000

CPU TERM 1.30 = FAIL (21 min)
CPU TERM 1.32 = FAIL (1 min)
CPU TERM 1.34 = FAIL (7 min)
CPU TERM 1.36 = FAIL (19 min)
CPU TERM 1.38 = FAIL (5 min)
CPU TERM 1.40 = FAIL (7 min)


//ADJUSTED MCH CORE TO 1.400
//RESTARTING AT MCH REF 0.7400


VCORE 1.3760 (1.43750 bios)
CPU PLL 0.1600
MCH REF 0.7400
MCH CORE 1.400

CPU TERM 1.30 = FAIL (11 min)
CPU TERM 1.32 = FAIL (19 min)
CPU TERM 1.34 = FAIL (8 min)
CPU TERM 1.36 = FAIL (5 min)
CPU TERM 1.38 = FAIL (6 min)
CPU TERM 1.40 = FAIL (7 min)

VCORE 1.3760 (1.43750 bios)
CPU PLL 0.1600
MCH REF 0.7200
MCH CORE 1.400

CPU TERM 1.30 = FAIL (11 min)

//CHANGED VCORE TO 1.392 (1.45625 BIOS)

CPU TERM 1.30 =

Another thing is I have 4x2gb ram, could that make any difference?