Core 2 Duo - An OC adventure

[Mac2580]

Good luck guys. If it werent for these chips, I would probably be a console gamer. Parents bought me a PC for high school work as well as a PS3 for gaming. E8400 + 8600GT. It didnt take long for me to realise that on PC, there are strategy games, and that shooting is better with a mouse. (In my opinion) I didnt care about frames then, surely it outperformed the PS3 though. I ran LGA775 for 9 years straight. I wish I could join you guys but unfortunately not with this board. Kept it regardless. Personally I know it will BSOD at 341x9, but I look forward to checking what they are really capable of.

 

[Dinnercore]

Good luck guys. If it werent for these chips, I would probably be a console gamer. Parents bought me a PC for high school work as well as a PS3 for gaming. E8400 + 8600GT. It didnt take long for me to realise that on PC, there are strategy games, and that shooting is better with a mouse. (In my opinion) I didnt care about frames then, surely it outperformed the PS3 though. I ran LGA775 for 9 years straight. I wish I could join you guys but unfortunately not with this board. Kept it regardless. Personally I know it will BSOD at 341x9, but I look forward to checking what they are really capable of.

Hey don´t worry to much about being competitive! While this is a competition, I think we all agree to do this for fun. Why not join us and push your board to the limit?

Btw, how long do you all think will we keep the competition open? I will not be able to join on something other then nForce for atleast the next week and a half. After that I could try to push the Maximus II Formula.

I´m in the middle of preparing myself. You should see my desk... Or better you don´t.

 

[delshay]

I'm ready.
View attachment 119927

That's a monster, I mean a real monster. You may want to experiment with a single Noctua 200mm fan, that's what I would do if I had this heatsink.
It would be interesting to see what it looks like with a single 200mm fan, & i would expect noise level to be much lower.

EDIT Experiment Two: You could also experiment with four fans below, if they can fit underneath, ie 15mm or 10-12mm depth fans. Don't think there's enough space for 25mm depth fans.

EDIT 2: Found another interesting down-draft cooler. Don't think I will be buying this as it's only rated @135w http://www.jonsbo.eu/en/cpu-cooler/cr-301/124/cr-301-black-rgb?c=88#prettyPhoto

 

[Dinnercore]

111% from my e4300

Hwbot links in my sig.

I´m currently trying to find some values of what 65nm can tolerate for a short bench session, those 1.744V seem very high for air cooling. Was that any bench stable at 1.7V+? If so, how did you cool that with ambient air?

 

[Caring1]

EDIT 2: Found another interesting down-draft cooler. Don't think I will be buying this as it's only rated @135w http://www.jonsbo.eu/en/cpu-cooler/cr-301/124/cr-301-black-rgb?c=88#prettyPhoto

The included fans are a pathetic, 45CFM only.
Upgrade those and that cooler would be more efficient.

 

[Bones]

That's a monster, I mean a real monster. You may want to experiment with a single Noctua 200mm fan, that's what I would do if I had this heatsink.
It would be interesting to see what it looks like with a single 200mm fan, & i would expect noise level to be much lower.

During development Scythe did a setup of one with a massive fan but it didn't fare so well.
After testing they concluded the 4 fan layout was best and it probrably is, it's very quiet on normal operation yet if you turn up the speed dial those fans can sing too.

 

[delshay]

During development Scythe did a setup of one with a massive fan but it didn't fare so well.
After testing they concluded the 4 fan layout was best and it probrably is, it's very quiet on normal operation yet if you turn up the speed dial those fans can sing too.

I'v seen a review of that heatsink, it's 1c better than the discontinued Noctua NH-D14.

The included fans are a pathetic, 45CFM only.
Upgrade those and that cooler would be more efficient.

Ok. I did not check the fans performance, so I will look at it again & consider it on my buy list.

 

[Dinnercore]

During development Scythe did a setup of one with a massive fan but it didn't fare so well.
After testing they concluded the 4 fan layout was best and it probrably is, it's very quiet on normal operation yet if you turn up the speed dial those fans can sing too.

German press reviews of this thing mentioned that the massive fan had QC issues and was much less reliable with high fault rates. Thats the only reason they choose not to use it. I think you could try a large Noctua or maybe a Silverstone AP-181. But 4x 120s work good so I don´t think there will be much of an improvement other than noise.

I´m ready too (except for the mb):

 

[Japie073]

Seems my G41 hates anything above 350MHz on the FSB. Got my E7500 running at 3.8GHz (345*11)

 

[Bones]

German press reviews of this thing mentioned that the massive fan had QC issues and was much less reliable with high fault rates. Thats the only reason they choose not to use it. I think you could try a large Noctua or maybe a Silverstone AP-181. But 4x 120s work good so I don´t think there will be much of an improvement other than noise.

I believe that was the reason.
If it ever ran into a problem with the fan somehow getting out of balance or just warbly like some do it was a problem - A big one.

Noise doesn't bother me esp since I'm already 25% or so deaf, medically proven years ago and it's probrably worse by now.

Not 100% sure if I'll actually be "In" it but want to. Lots going on here and never enough time for it all.

Will have to dig out the cooler, give the fans a good cleaning, mount it up and get it all ready.
That's an older pic of the rig but what it will be once it's setup again.

 

[M0rafic]

I know this isn't a proper test, but since I'm sitting watching TV with my old dell m1330 sat on my lap and too close too bits I'd rather not cook, it will have to do. Besides its not as if I can cram a 150mm noctua fan in it to disperse the heat But a whisker shy of 3.4GHz isn't too shabby for a 12 year old laptop

 

[storm-chaser]

Impressive indeed for an older tech laptop... I saw your post in the laptop overclocking adventures thread. Great work!

Welcome to TPU and hope you can see 4.0Ghz with that X9000. She will get a little hot, best you stick your laptop in the freezer before attempting anything over 3.6Ghz.

 

[M0rafic]

I did have a quick go at getting over 3.4GHz and it did go over 3.5, but would only hold it for 10s or thereabouts before dropping down. So I nudged up the voltage to 1.5v and after about 15secs BSOD :-( Which screwed up thottlestop as this particular laptop won't overclock properly if the Disable Turbo box is checked and I can't un-check it nor can I remember how I did so the last time. The problem with the m1330 is that the cpu (hot) sits close to the nVidia GS8400M (very hot) and right between the two is the intel chipset (which gets cooked from both sides.

It was the original T7500 with the same tdp as the x9000 it now has that killed the first two motherboards. Poor design for something that cost me over 2 grand even after the sizeable discount I got whilst I was doing platinum server support working Dell at the time. Such were the numbers of fried laptops that Dell gave everyone an extra 12 months warranty. Anyhoo, I missed the notification and as I was busy; I just shoved it in a cupboard. Fast forwar to 2012 so 4 years after my extended warranty ran out I was looking for another laptop and wondered what Dell would charge to fix my m1330 compared to buying something with an i7 in it. So I gave them a call, explained the situation and lo and behold they sent out a guy with a new motherboard and it didn't cost me a penny!

So I've become sort of attached to it and its still pretty quick, when compared to your average laptop Though with 8Gb ram, the x9000 and an ssd it has had a few upgrades. But over the years it's run vmware, its been an oracle server, it had Adobe CS4 on it for a while, it still dual boots ubuntu runs an apache server and Postgresql. So you could say that apart from the years it spent in the cupboard, It's earned its keep

Mike

 

[vega22]

I´m currently trying to find some values of what 65nm can tolerate for a short bench session, those 1.744V seem very high for air cooling. Was that any bench stable at 1.7V+? If so, how did you cool that with ambient air?

-11c air temps really helped.

I was just after high gigglehurts, suicide shots tbh.

 

[Dinnercore]

-11c air temps really helped.

I was just after high gigglehurts, suicide shots tbh.

Ah that explains it I´m impressed at how tough these 65nm chips are. I had one running at 1.63V with ambient water, it got pretty warm and didn´t help with my clocks as the problem was elsewhere, but it didn´t degrade during a couple test runs either. Still rocking the same voltage / frequency curve after the torture.

I got another mainboard to try out:


Asus P5Q-E, a classic. Replaced the by now bone dry stock TIMs.

Got it on my bench and made some early tests to get a feel for it:


As far as I can tell it handles high FSB well, BUT again gives me headaches with the memory. 2nd and 3rd RAM timing table does not seem to suit my OCZ memory. It can only run on this board on the very specific clock-band between DDR2 1060-1130. Anything above OR below fails to post. Doesn´t matter which settings I choose outside of the timings themself. Relaxing the primaries to something like 7-7-7-21 doesn´t do it either. Only when I touch the 2nd and 3rd information myself I can get it to post at speeds from 800-1200 (my kit is rated DDR2 1200 5-5-5 @2.1V, I´m using 2.2V atm) but it never gets stable. BSODs as soon as I get to the login screen.
Likely because I have absolutly no idea what I´m doing on those timings, I´m a total RAM-OC noob. It is a bit disappointing that it does that. On the nForce I could just set primary timings 5-5-5-16 and was good for 1100MHz @2.1V. No chance on the P5Q-E.

Oh and some more test candidates arrived:


Very kind of the seller to include 3x the tray, I´m not sure why he filled the bottom of one tho. Still, thanks to this I now have some neat trays to organize my CPUs.

 

[DR4G00N]

@Dinnercore I have a P5Q-E also it's a decent board just a bit finicky. Highest FSB I did with mine was 640MHz with chilled NB & CPU. https://valid.x86.fr/x065iy

And yeah, Asus P45 DDR2 board's don't seem to like OCZ memory, I have a couple kit's here and neither work well past DDR2-1100 or so on both my P5Q-E & Maximus II Formula. My good Crucial & Buffalo kit's do DDR2 1300+ C5 @ 2.2V just fine though.

I've attached KET's Modded bios for the P5Q-E with Xeon code's added to it. Makes a nice improvement to the board.
http://www.xtremesystems.org/forums/showthread.php?200109-Modded-ASUS-P5Q-Pro-Deluxe-Premium-BIOS

 

[vega22]

Not all chipsets are made equal

You want to get a T power or ud3f for the best ddr2 mobo. Maxi extreme will be the easiest way as the ddr3 removes that limitation but none of them are cheap now, if you can even find one for sale.

 

[Dinnercore]

@Dinnercore I have a P5Q-E also it's a decent board just a bit finicky. Highest FSB I did with mine was 640MHz with chilled NB & CPU. https://valid.x86.fr/x065iy

And yeah, Asus P45 DDR2 board's don't seem to like OCZ memory, I have a couple kit's here and neither work well past DDR2-1100 or so on both my P5Q-E & Maximus II Formula. My good Crucial & Buffalo kit's do DDR2 1300+ C5 @ 2.2V just fine though.

I've attached KET's Modded bios for the P5Q-E with Xeon code's added to it. Makes a nice improvement to the board.
http://www.xtremesystems.org/forums/showthread.php?200109-Modded-ASUS-P5Q-Pro-Deluxe-Premium-BIOS

Thank you! That is very useful stuff. I already flashed it and will try if that improves the situation a bit. If not, I still have the board ready for Xeons. Might give my G.Skill sticks a try.


EDIT: Well the auto timings still don´t work at all above 800, but it did not get worse and with manual primary timings it does seem to get a bit further now. Will have to confirm with extended testing

 

[theFOoL]

THIS Guy

 

[DR4G00N]

Some -20c coolant testing with that 4.9GHz E8400.
Seems garbage for SuperPi 32M though, failed even at 4.5GHz 1.5V.

 

[storm-chaser]

Looking good dragoon. 5.2Ghz isn't too shabby! I will pull my e8600 out of storage later today. I want to first get results from my Q9650 and then move on to the dual core chips.

 

[phill]

Looking good dragoon. 5.2Ghz isn't too shabby! I will pull my e8600 out of storage later today. I want to first get results from my Q9650 and then move on to the dual core chips.

I' d personally go the other way, just makes those quad cores seem even faster than going fast and then much slower lol

I'll definitely need to have a look at what CPUs I have or what you guys would like to use for this competition....

 

[infrared]

Ah that explains it I´m impressed at how tough these 65nm chips are. I had one running at 1.63V with ambient water, it got pretty warm and didn´t help with my clocks as the problem was elsewhere, but it didn´t degrade during a couple test runs either. Still rocking the same voltage / frequency curve after the torture.

I got another mainboard to try out:


Asus P5Q-E, a classic. Replaced the by now bone dry stock TIMs.

Got it on my bench and made some early tests to get a feel for it:


As far as I can tell it handles high FSB well, BUT again gives me headaches with the memory. 2nd and 3rd RAM timing table does not seem to suit my OCZ memory. It can only run on this board on the very specific clock-band between DDR2 1060-1130. Anything above OR below fails to post. Doesn´t matter which settings I choose outside of the timings themself. Relaxing the primaries to something like 7-7-7-21 doesn´t do it either. Only when I touch the 2nd and 3rd information myself I can get it to post at speeds from 800-1200 (my kit is rated DDR2 1200 5-5-5 @2.1V, I´m using 2.2V atm) but it never gets stable. BSODs as soon as I get to the login screen.
Likely because I have absolutly no idea what I´m doing on those timings, I´m a total RAM-OC noob. It is a bit disappointing that it does that. On the nForce I could just set primary timings 5-5-5-16 and was good for 1100MHz @2.1V. No chance on the P5Q-E.

Oh and some more test candidates arrived:


Very kind of the seller to include 3x the tray, I´m not sure why he filled the bottom of one tho. Still, thanks to this I now have some neat trays to organize my CPUs.

Nice, that's basically the same board I had, mine was the deluxe but there's very little difference. From what I remember two important settings (that are a pita to set via trial/error) are clock skews and GTL reference settings, the board does an okay job of setting these on auto, but you do gain a lot by getting them dialed in correctly. You'll have to spend some time googling if you want ballpark settings though, I can't remember what settings worked best, there's probably tons of pages on here discussing them.

Also, try setting 2T timings and using a different memory divider, 1:1 gets great bandwidth and latency if you can push the fsb high enough. Maybe go back to auto sub-timings to test these changes, then tweak afterwards.

Good luck, it's a bit of a finicky board but should reward you for doing the fine tuning.

 

[Dinnercore]

Again a big thank you for @DR4G00N the modded bios did help with memory stability.

I can now get 1060 running at the advertised timings of the kit. Which was impossible before.


But to get any further on the FSB I need to change the division to something like 1:1 as @infrared suggested. This however is still impossible with this kit and board :/
I went back two steps in FSB for maximum stability and switched over to DDR2 860 (1:1) but it now fails to boot every single time. CPU is stable at the clock / voltage, FSB did not change just the different ratio kills it. Timings seem to stay the same on all 3 informations. It´s the same for both 333 strap and 400 BUT the 333 strap on 1:1 fails to post while the 400 one does atleast attempt a boot.
Very strange. I have to go for faster clocks or else it dies. I guess this board/memory combo has a real need for speed. It does better on 1:1 ratio when FSB is getting up to 500+ again because then my DDR2 speeds are back over 1000...
Maybe clock skews are the magic juice that I need.

Oh and on that screenshot, I got one if those rare Intel locked-temperature CPUs it seems. It stays at 37°C no matter what, so I guess 2.1V benching it is. Will take this thing with me during winter, the 37°C will keep my hands warm.
(note this is a joke, I don´t mean it serious and I know these sensors can get stuck).

 

[DR4G00N]

What you need is a kit of memory that aren't OCZ. Use a 2x2GB kit if that's all you have.

Different straps can affect stability, so 333 1:1 can be rock solid but 400 1:1 doesn't post. The FSB you boot with can also affect the maximum fsb oc.