Overclocking the e8400

[NoM8s]

I warned you all of the possibly silly questions....here is maybe one of them.
I have a second rig I built to tinker with, the spec is as follows :

E8400 @ 3.0GHz stock (running @ 3.45GHZ)
Asus P5N32-E SLI Mobo
4GB Corsair XMS2 DDR2 PC 6400 (4-4-4-12)
OCZ Stealth Xtreme 600W PSU
Zalman CNPS9700-NT
9800 GTX+
Vist 64bit

I achieved a 15% overclock by selecting that option in the Bios. I am having trouble locating the FSB option to enter 400 so as to get a 3.6GHz overclock, I have found and set the multiplier to 9 (9 x 400 =3.6). I must again tell you all that I am new to this and am looking for some guidance.

As I posted in another area my 3.6 oc on the Core i7 was a simple affair, unlike this attempt .

 

[nenforcer]

I have the exact same motherboard with an E6600 and will be looking to upgrade to a Wolfdale, possibly the E8700 when it eventually comes out. If you don't figure it out let me know and I will check out my BIOS to see if I can find the setting. I prefer not to overclock and void any warranty but have never actually tried.

 

[BonzaiDuck]

Originally posted by: nenforcer
I have the exact same motherboard with an E6600 and will be looking to upgrade to a Wolfdale, possibly the E8700 when it eventually comes out. If you don't figure it out let me know and I will check out my BIOS to see if I can find the setting. I prefer not to overclock and void any warranty but have never actually tried.

Tell me if I'm wrong, but doesn't the P5N32-E SLI deploy the nVidia 680i chipset? Wait a minute -- lemme see for myself . . . . Yeah . . . same as my Striker Extreme board.

You need to upgrade the BIOS to make it Wolfdale compatible. As far as Penryn is concerned, Wolfdale is as far as it goes. Try and find the BIOS revision that is several months more recent than the release of the Wolfdales back in January, '08.

You can get the E8400 to 3.6 Ghz -- I did it on the Striker Extreme. It won't quite make it to 3.8 Ghz without some fine-tuning -- and then again, maybe not, even if you fine-tune until your vision begins to blur.

3.6 is a 20% over-clock. That means the FSB has to go up to 400 Mhz, and with the 680i chipset, that's where it gets tricky. You'll need to adjust the CPU_VTT or CPU_FSB (same voltage setting) upwards. But with the 45nm Wolfdale, you should not push it above 1.4V. In fact, you are wise to limit VTT increases to a slightly lower limit.

I need to look into this again -- my mind is foggy about it -- but if an acceptable VTT adjustment doesn't quite get you to 400 Mhz, you may be able to adjust the GTLREF voltage settings -- although these are for use with higher VTT -- not lower VTT.

On my 780i motherboard, I can get way past 400 Mhz with a VTT setting of only 1.3V. The actual value shown in BIOS monitor may be more like 1.28V. And I recall on the 680i Striker that a setting of 1.4 resulted in a reading of between 1.35 and 1.38.

Pushing the VTT above 1.4V risks burning out a Penryn prematurely.

Please, though, above all -- read Graysky's sticky, and take the time to do this right with PRIME95 multi-core version. If you can get it to post and load Windows, and you get the VCORE set so that it fails in between 10 and 30 minutes, your additional VCORE adjustments should be incremental to achieve stability.

Knowing these nVidia chipsets, make sure to "unlink" the FSB and RAM speed settings; run the RAM at a CPU : RAM ratio of 1:1 -- stock latencies and at RAM voltage that is just comfortably within the warranty maximum. You will make adjustments with a target FSB of 1600 and target RAM speed of 800 [in DDR terms].

 

[NoM8s]

Thanks for all the info., that gives me a lot to fiddle with and points me in the right direction. I have already updated the Bios to version 1701, which seems to be the most rescent. I'll let you know how I get on (or not). Thanks again for the input.

 

[M1A]

As a whole the E8xxx's do 400FSB easily. Above that you will soon need faster ram. Better cpu cooling, and watch your temps. I thought the 680i and 780i were better overclocking boards than the p35's but looking at Bonzai's post I guess not.
I have had very good OC luck with my setup thats below in my specs.
Good Luck!

 

[Tempered81]

My P35 did 500mhz+ FSB easy, but my Nvidia 750i won't stay stable above 478mhz

 

[Gillbot]

Originally posted by: jaredpace
My P35 did 500mhz+ FSB easy, but my Nvidia 750i won't stay stable above 478mhz

I've had my 780i up to 500fsb with a X6800, but with a 45nm chip it's been impossible to get over 375MHz or so.

 

[BonzaiDuck]

"The E8x00 CPUs are easy to overclock above 400 Mhz . . " -- TRUE.

"I thought the 680i and 780i are better over-clocking boards than the P35 [chipset] boards . . . "

CAVEATS:

The 680i boards -- and I can only speak to my Striker Extreme firsthand -- don't quite make it above 400. I had my Striker 680i above 405 with the E8400 chip, and then -- one day -- it simply wouldn't post. I may not have attended to chipset cooling -- didn't use the little fan that ASUS bundles with those boards. [I was kicking myself, because I should've known to use the fan from the 'git-go.'] Keep in mind I'm a very conservative over-clocker, and many forum members here have taken 680i boards much higher -- if not at the stock CPU multiplier, than at a lower multiplier and higher FSB. I simply don't recall the limits -- and those limits were reported in comparisons between 680i boards and other boards.

The 780i board that I have -- not an ASUS but the eVGA 132-CK-NF78 -- seems to effortlessly over-clock to -- and past -- the advisable limit for the CPU. I am certain -- dead certain -- I could push further with the stock multiplier, but my idle VCORE will creep outside the 1.3625V spec-limit. The NB/SPP temperature only rises two or three degrees going up from 400 to 425. Certainly, just as eVGA did when they designed the board and its stock cooling solutiion, I would advise special attention to chipset cooling if you attempt to push it higher. I never bothered with dropping the CPU multiplier and pushing the FSB as high as 500, but I've seen several people do that on the eVGA forums -- maybe even some posting here. For me, getting memory spec'd at DDR=900 to run at DDR=850, with the lowest voltage in the RAM spec and timings 4,3,3,8 -- is just freakin' wonderful happy-fine with me.

I wouldn't know about the P35's performance in comparison. But my 680i associates here jumped into the P35 and other Intel chipsets with both feet -- touting a sigh of relief. On the other hand, I've posted here in threads from people who had trouble OC'ing their P35 systems, and their troubles may not have been inherent to the chipset. They seemed to need "guidance."

EDIT FOR GILBOT:
Did you tweak the CPU_VTT/FSB voltage to at least 1.3V (and NEVER -- NEVER higher than 1.4V)?? As far as I know, you posi-lutely abso-tively have to bump it up when getting close to 400 and above.

Here are my settings for E8600: 1.3625V (set)-- 1.32 to 1.34V (idle from sensor) -- 1.288V to 1.304V (load from sensor) -- VCORE
1.30V (set) for CPU_VTT/FSB
2.0V (set) RAM for G.SKILL spec'd for 2.0 to 2.1V -- sensor = 2.02V
1.30 to 1.35V (set) for NB/SPP
1.5 (? I need ot check again) for SB
1.2V for 1.2V_HT

That puts me at 4.25 Ghz CPU, and 425 Mhz (CPU) FSB -- 20 iterations LinPack IntelBurnTest; 6 hours or more PRIME95.

Also -- check for BIOS revision -- it may make a difference.

 

[Tempered81]

wow, bonzai! 1.288v @ 4250mhz?

i'm currently fluctuating between 1.376-1.392v @ 4300mhz on an e8400

that must be a nice e8600 you have.

 

[BonzaiDuck]

Originally posted by: jaredpace
wow, bonzai! 1.288v @ 4250mhz?

i'm currently fluctuating between 1.376-1.392v @ 4300mhz on an e8400

that must be a nice e8600 you have.

Like I said -- I'm a "conservative" over-clocker. I wouldn't push my "set" VCORE above the 1.3625V notch, but apparently, it is the actual load and idle values that matter in the 0.85 to 1.3625V Intel "safe" range. @ 74F-room-ambient, my tCase temperature peaks at 71C under IntelBurnTest, while the processor thermal spec is around 72. I'd be interested to hear your tCase maximum under LinPack IBT. I think the thermal spec is related to throttling, and I have that feature turned off in BIOS -- common practice for OC'ing.

I think I said before, the Anandtech article on the QX9650 from fall 2007 (a 45nm CPU) explains vDroop and vOffset, or how the voltage spikes momentarily when returning from load to idle -- the spike being about 0.02V above the idle equilibrium. That equilibrium, for me and my E8600, is 1.34V as measured by the sensor. Well -- the sensor bounces between 1.32 and 1.34. It's all a matter of how much you can trust those readings.

Whether the momentary spike going outside the 1.36+ range-limit has any adverse effect, I couldn't say. The spec quotes a wider range limit of 1.45, saying that the processor will still "operate," but will degrade above 1.36+.

I'm cautious -- that's for sure.

 

[Tempered81]

mine stays around 73-74C max small FFT load. IBT, linx etc is about 8-9C higher (but very jumpy)

 

[NoM8s]

Well its really nice to know that my request for help generated such interesting discussion and invaluable data information, I'll see what I can do with all the info, and let you all know how a newbie gets on.

 

[BonzaiDuck]

Originally posted by: NoM8s
Well its really nice to know that my request for help generated such interesting discussion and invaluable data information, I'll see what I can do with all the info, and let you all know how a newbie gets on.

I have plenty of time on my hands. People here say I'll "write a book to say what can be said in a page." See if you can't find some performance reviews on the P5N32-E SLI -- they should still be available at benchtest web-sites, and should give you an idea of the limits for the board. The ASUS forums may have archival threads and posts, as well.

Those limits would be no different than I found with my Striker Extreme. Pay attention to chipset cooling -- even for installing the little fan on the NB HS fins. Part of my problem with the STriker and the E8400 stemmed from not volting the processor sufficiently, and instead of simply "terminating" a thread upon error in PRIME95, it would crash to BSOD. A little more voltage would've made the difference. I'm much bolder now in volting the E8600, and most would admit that I'm still "cautious."

 

[NoM8s]

Thanks for the input Bonzai...I would like to hear your thoughts on my i7 920 overclocks. If you look at my other post I would be very interested in what you think. Cheers

 

[BonzaiDuck]

Originally posted by: NoM8s
Thanks for the input Bonzai...I would like to hear your thoughts on my i7 920 overclocks. If you look at my other post I would be very interested in what you think. Cheers

Firsthand experience would make my opinions more worthwhile -- and I haven't any. I have a lot of catching up to do. It's a new and different architecture, and what used to be "FSB" is now "QPI" measured in giga-transfers per second. All I can see at this point, is that you achieved something like a 35% over-clock.

I'd say that 35% is good, with air-cooling for a processor that has a stock TDP of 130W.

This Akasa Nero AK-967 cooler seems to be a "direct-touch" heatpipe cooler with one of the better thermal-resistances, gauged by performance of load temperature above ambient. But I don't trust the FrostyTech reviews that much. I'd only assure myself that it is a good cooler performing well under a 125W thermal-level test -- just short of the stock TDP for that processor.

Lord knows precisely what the thermal wattage of your 35% OC would be. How close to the "auto" VCORE value could you set the voltage for stability? Or how far beneath the Intel voltage spec-range maximum?

 

[NoM8s]

OK I'll bare all that in mind, thanks again for sharing your thoughts.

 

[NoM8s]

Originally posted by: NoM8s
I warned you all of the possibly silly questions....here is maybe one of them.
I have a second rig I built to tinker with, the spec is as follows :

E8400 @ 3.0GHz stock (running @ 3.45GHZ)
Asus P5N32-E SLI Mobo
4GB Corsair XMS2 DDR2 PC 6400 (4-4-4-12)
OCZ Stealth Xtreme 600W PSU
Zalman CNPS9700-NT
9800 GTX+
Vist 64bit

I achieved a 15% overclock by selecting that option in the Bios. I am having trouble locating the FSB option to enter 400 so as to get a 3.6GHz overclock, I have found and set the multiplier to 9 (9 x 400 =3.6). I must again tell you all that I am new to this and am looking for some guidance.

As I posted in another area my 3.6 oc on the Core i7 was a simple affair, unlike this attempt .

Finally I have managed a stable 3.6GHz overclock on my E8400 and my settings were as follows
CPU FSB 400
PCI Freq.100
V Core 1.26250
FSB 1600
Memory 800

Core Temps running Orthos for two hours
Idle #0 37 deg. C
#1 36
Load #0 46
#1 46
These are max temps.
now to go for the 4Ghz O/C

Thanks again to all of you who gave me some invaluable advice and set me on the road to Overclocking Madness....I will bore you all with my meagre progress, as it unfolds!!
Cheers

 

[BonzaiDuck]

Originally posted by: NoM8s

Originally posted by: NoM8s
I warned you all of the possibly silly questions....here is maybe one of them.
I have a second rig I built to tinker with, the spec is as follows :

E8400 @ 3.0GHz stock (running @ 3.45GHZ)
Asus P5N32-E SLI Mobo
4GB Corsair XMS2 DDR2 PC 6400 (4-4-4-12)
OCZ Stealth Xtreme 600W PSU
Zalman CNPS9700-NT
9800 GTX+
Vist 64bit

I achieved a 15% overclock by selecting that option in the Bios. I am having trouble locating the FSB option to enter 400 so as to get a 3.6GHz overclock, I have found and set the multiplier to 9 (9 x 400 =3.6). I must again tell you all that I am new to this and am looking for some guidance.

As I posted in another area my 3.6 oc on the Core i7 was a simple affair, unlike this attempt .

Finally I have managed a stable 3.6GHz overclock on my E8400 and my settings were as follows
CPU FSB 400
PCI Freq.100
V Core 1.26250
FSB 1600
Memory 800

Core Temps running Orthos for two hours
Idle #0 37 deg. C
#1 36
Load #0 46
#1 46
These are max temps.
now to go for the 4Ghz O/C

Thanks again to all of you who gave me some invaluable advice and set me on the road to Overclocking Madness....I will bore you all with my meagre progress, as it unfolds!!
Cheers

I think what you're saying is that you've made promising progress getting to the 3.6 Ghz? And you're more familiar with the BIOS features as they affect your over-clock?

This is my opinion -- which no one has much challenged around here -- modestly asserted here and in other recent threads. If the 1.2625V value is the VCORE "setting" -- and the voltage sensors show VCORE (in BIOS OR in Win/CPU_Z) with idle <= "set" and load < "set," you can notch it up higher. [I'm getting old, and don't remember (without checking again) what I posted here days ago, but . . ] the Intel "safe" spec range is 0.85 to 1.3625V. You have "headroom" to go higher -- I think.

I just wish I had better understood that when I was OC'ing the E8400 on the 680i board.

Also, your temperatures seem just "too good." I'm guessing that you've only stress-tested with PRIME95? Because -- if I could read my true tJunction "core" sensors/temperatures under PRIME95 -- they would fit with yours. As it stands, my sensors are stuck at 51C, and the PRIME95 isn't sufficient to lift them beyond the stuck value. I can see my tCase temperature at both idle and load, and it corresponds roughly to what you report for your cores. Since my "set" VCORE at 1.3625V is higher than yours, it suggests again -- you have "headroom."

Some people pick up the "over-clocking art" faster than others, and you seem to be getting the hang of it on a short learning curve.

 

[NoM8s]

Originally posted by: BonzaiDuck

Originally posted by: NoM8s

Originally posted by: NoM8s
I warned you all of the possibly silly questions....here is maybe one of them.
I have a second rig I built to tinker with, the spec is as follows :

E8400 @ 3.0GHz stock (running @ 3.45GHZ)
Asus P5N32-E SLI Mobo
4GB Corsair XMS2 DDR2 PC 6400 (4-4-4-12)
OCZ Stealth Xtreme 600W PSU
Zalman CNPS9700-NT
9800 GTX+
Vist 64bit

I achieved a 15% overclock by selecting that option in the Bios. I am having trouble locating the FSB option to enter 400 so as to get a 3.6GHz overclock, I have found and set the multiplier to 9 (9 x 400 =3.6). I must again tell you all that I am new to this and am looking for some guidance.

As I posted in another area my 3.6 oc on the Core i7 was a simple affair, unlike this attempt .

Finally I have managed a stable 3.6GHz overclock on my E8400 and my settings were as follows
CPU FSB 400
PCI Freq.100
V Core 1.26250
FSB 1600
Memory 800

Core Temps running Orthos for two hours
Idle #0 37 deg. C
#1 36
Load #0 46
#1 46
These are max temps.
now to go for the 4Ghz O/C

Thanks again to all of you who gave me some invaluable advice and set me on the road to Overclocking Madness....I will bore you all with my meagre progress, as it unfolds!!
Cheers

I think what you're saying is that you've made promising progress getting to the 3.6 Ghz? And you're more familiar with the BIOS features as they affect your over-clock?

This is my opinion -- which no one has much challenged around here -- modestly asserted here and in other recent threads. If the 1.2625V value is the VCORE "setting" -- and the voltage sensors show VCORE (in BIOS OR in Win/CPU_Z) with idle <= "set" and load < "set," you can notch it up higher. [I'm getting old, and don't remember (without checking again) what I posted here days ago, but . . ] the Intel "safe" spec range is 0.85 to 1.3625V. You have "headroom" to go higher -- I think.

I just wish I had better understood that when I was OC'ing the E8400 on the 680i board.

Also, your temperatures seem just "too good." I'm guessing that you've only stress-tested with PRIME95? Because -- if I could read my true tJunction "core" sensors/temperatures under PRIME95 -- they would fit with yours. As it stands, my sensors are stuck at 51C, and the PRIME95 isn't sufficient to lift them beyond the stuck value. I can see my tCase temperature at both idle and load, and it corresponds roughly to what you report for your cores. Since my "set" VCORE at 1.3625V is higher than yours, it suggests again -- you have "headroom."

Some people pick up the "over-clocking art" faster than others, and you seem to be getting the hang of it on a short learning curve.

What you have said about the Temps. is interesting, I'll test with Orthos as well and see what the effect on the Temps is....I'll let you know. What other testting software would you suggest?

 

[BonzaiDuck]

Originally posted by: NoM8s

What you have said about the Temps. is interesting, I'll test with Orthos as well and see what the effect on the Temps is....I'll let you know. What other testting software would you suggest?

It's hard to keep up with stress-tester revisions, but necessary.

There is a new version of the (Russian) OCCT program, which incorporates an option to run LinPack.

The most recent, multi-core version of Prime95 is useful for finding stability, but requires as much as a half-day's operation on each of the important tests (small-FFT's and Blend-test) to more or less "guarantee" perfect stability. Also, it doesn't stress the processor nearly as much as LinPack in its various forms.

Maximum temperatures will be achieved with LinPack, found as "IntelBurnTest," a prettier Windows version that has popped up here or at another forum, and the OCCT implementation. Since I have IntelBurnTest, I feel more comfortable using it for now.

@ VCORE load of 1.29V, 4.25 Ghz and the E8600 (VCORE setting of 1.3625V), IntelBurnTest shows a peak temperature @ 73F room-ambient of about 71C. Some of the hot-dawgs here who push their voltage and clock-settings higher are showing temps approaching 80C.