Core 2 Duo - An OC adventure

[rockit00]

Yes they do. Thanks.

 

[Komshija]

Is it socketed or BGA? If it's socketed maybe you could do the FSB mod. It's been a long long time but I remember you could jump a couple of pins (BSEL mod I think it was called) to trick it into running at 333mhz, but that would mean 3.8ghz which will prob be too far, and the board might not even boot at 333mhz fsb.

edit - there's a guide HERE, but most of the pics are gone unfortunately.

It's socketed. I have a general idea how to physically alter or connect CPU pins, but never done it in practice. The farthest I went modifying the CPU was converting LGA771 Xeon CPU to fit LGA775 motherboard.
If something goes wrong, my back-up laptop (I hardly ever use it, but still...) is gone. Since I fully restored it and invested quite a lot in this laptop, that would be a shame. I was even considering installing Core 2 Extreme QX9300 (same 478-pin socket P CPU), because I saw that someone managed to wire certain pins and installed custom BIOS, but I think he had a different laptop model. Then I found that I'm also missing fine tools for the such job , so I tried with software with no luck... I didn't plan to push it past 3,33 GHz anyway.

 

[dorsetknob]

But I´m not giving up on nForce just yet

Never was a fan (pun because nforce needs it) of nvidia Chipsets.
Known for melting .....>and corrupting data on drives.

 

[Dinnercore]

Never was a fan (pun because nforce needs it) of nvidia Chipsets.
Known for melting .....>and corrupting data on drives.

I can understand that, really that 780i I use in my SLI-rig gets toasty even with multiple fans all over it. Okay the poor thing is busy with an X5470 @4.44GHz + quad SLI but still.

Btw that data corruption was related to raid and velociraptors I think + the latest bios versions should include a fix. I think I´m safe.

The Striker board has it under control tho, as far as I´ve seen by now.

This is with 1.52V on the NB, a good starting point I´d say. Maybe temps will be worse when I reach 1.6 - 1.7V but I have to get there first.
Currently I´m busy probing the thing and playing with all the different voltages and offsets. I have practially little idea of what I´m doing, so I set memory clocks and timings way down to have them dead stable and not interfering first. Then tried to go up on FSB but with the CPU at max. multi at the same time. This was a bad idea because I had to jump the exact same FSB-hole as I had on my 780i and had to jump another one to get it to boot beyond 1800MHz. It managed that, I can get it to post between 1850 and 1950, but every single value is unstable to the point that windows does some freaky stuff. Like suddenly loading with the Windows XP loading bar into BSOD.
It got a little better by raising Vcore higher and higher, but I stopped at 1.45V in bios reading as something like 1.5V actual value. If the CPU could take the frequency it would do it by then. So I´m still on an unstable FSB I think, or my VTT is off?
It doesn´t help that the C0-Stepping wants completly different GTL-offsets compared to my E0-Xeon that I´m used to. Might just switch CPU out for one of my 3 E0 soon. But before that I will now fix the multi to 6x and just try FSB OC by itself.

EDIT: Oh the stuff you encounter when touching windows with unstable settings. I killed my windows partition table without noticing it, thinking my settings suddenly are all no longer booting. Was scratching my head why even stock settings don´t boot anymore until I plugged in a repair stick and found out that windows died on me

 

[storm-chaser]

I've had a little fun with these chips... all these benches were done with air cooling (Thermalright Ultra 120 extreme)
And I made all these runs before I learned of the bugged 3:5 mem divider...

E8600

E8400

E7500

Q9650 peak clocks

Q9650 daily driver settings

QX6850 peak clocks

 

[Deleted member 24505]

Just wondered if anyone else managed over 100% oc like my e6300?

 

[Dinnercore]

I´ve been busy too and tried the approach of FSB first, then back a bit and try to go up with the multi again...

Well this does not work at all for me on the Striker board. This board is one tricky beast. It wiggles itself out of stability on every chance it gets. Don´t get me wrong tho, I love it! Sorry that this might delay any useful charts on my first post for now, but I´m gonna be busy this weekend tackling the monster.



I got it up to 500 FSB and NB temps are still under control with air cooling at 1.66V. I had to raise VTT into slightly uncomfortable levels to get any sort of memory stability on this. On top of all that the voltage readings from the boards sensors are all over the place, varying up to 50mV from the actual value. The 1.45Vcore on the screenshot are actually 1.40 set in bios and 1.39ish real voltage...

To keep the mcp on the same temp as before with moar volts I used an old AMD fan:


Found this mad lad catching dust in a box and it was a perfect fit for the gap between GPU and CPU-Cooler. Put some of the noctua rubber thingys on it and just let it sit there.

Oh and after raising FSB the boards LEDs came on, this little 'speedometer' thingy started to burn my eyes so I grabbed a VGA-cable cover to diffuse it a bit.



And then I started to fiddle with my multi... and it all went downhill fast. I went back up in the 0.5x steps and every single step was an instant no post again. Back down on 475 FSB it was the same, but I had to adjust everything to 475 too, since FSB of 475 is unstable at the voltages of 500. So I spend 2 hours going up and down and over and out on all the settings I could find, tried memory timings manual from 8-8-8-24 to 11-11-11-30 1T/2T, AUTO, all of these things with 1300-1800 on the memory and every single time I got the same result:

I get past POST but can´t boot stable. Windows loading screen appeared multiple times, even went into login screen @ 9.5 multi with 500FSB but no luck after that. BSOD and other fun stuff:
EDIT: Oh and yes I used 2x Sticks instead of 4 too.


Now I´m at the point that I think my windows install nr. 2 broke again...
So I have no clue if some settings may have been stable if windows was ok, or if the settings are all wonky or what.

So much to learn... BUT I will get there. Just gotta take my time and maybe fry a CPU.

Just wondered if anyone else managed over 100% oc like my e6300?

I want to try that but not on the higher clocking skews. That would require LN2 which I can´t get in my country.

 

[Deleted member 24505]

I want to try that but not on the higher clocking skews. That would require LN2 which I can´t get in my country.

That chip really was something special, 560fsb easily, had more but I did not have the cooling for it It did require 1.57v for that though eek

 

[Dinnercore]

Update, going back to 2x2 RAM was a stupid move. Threw off all the settings I noted as stable on my way up with the FSB...

Instead back on 4x2. I decided to stick to the rule a wise man with physically unplausible hair once shared: Keep calm and raise Vcore.
At ~1.54 ish V on the CPU and 1.42ish V VTT I managed to get it running.



Core temps settle at ~78°C. Memory speed on auto and still garbage, got to try and get that in line. IF that is possible on an nForce chipset.

 

[freeagent]

Just wondered if anyone else managed over 100% oc like my e6300?

I had an original e6300 Conroe that ran at 3700 and change on a P5B Deeluxe wifi. I wish I still had it because it was the best overclock ever got out of a cpu. It might have went higher, but I only had 667 ram and it was maxed. I traded it and cash for an e6600. I was hoping for a 4ghz e6600 but didnt get it. The cache was a nice bump though.

 

[king of swag187]

have wanted a t7600g (unlocked multiplier) for my Dell XPS1710 laptop

could get a C2D X9100 if you wanted, might be BIOS limited or something else though

 

[Deleted member 24505]

I had an original e6300 Conroe that ran at 3700 and change on a P5B Deeluxe wifi. I wish I still had it because it was the best overclock ever got out of a cpu. It might have went higher, but I only had 667 ram and it was maxed. I traded it and cash for an e6600. I was hoping for a 4ghz e6600 but didnt get it. The cache was a nice bump though.

Mine was the best, luckily my patriot ram did 1120, 1:1 with the 560 FSB, also was the best clocking CPU I have ever come across. My only problem was the cooler, with a better one, i'm sure I could have cracked 4ghz

 

[dorsetknob]

could get a C2D X9100 if you wanted, might be BIOS limited or something else though

on ebay their £60 this was a £5 nuff said

Just wondered if anyone else managed over 100% oc like my e6300?

managed to get a P11 300 to 650Mhz years ago

 

[Bones]

Did this a few years ago.
Not a 100% OC but close and do consider the chip and platform it was done with.
https://hwbot.org/submission/2658219_bones_cpu_frequency_opteron_165_3514.75_mhz

 

[storm-chaser]

I´ve been busy too and tried the approach of FSB first, then back a bit and try to go up with the multi again...

To keep the mcp on the same temp as before with moar volts I used an old AMD fan:


Found this mad lad catching dust in a box and it was a perfect fit for the gap between GPU and CPU-Cooler. Put some of the noctua rubber thingys on it and just let it sit there.

I did the same thing with my Asus P5Q3... just enough room between the GPU and CPU cooler for a small fan:

 

[Deleted member 24505]

I can't remember if the P5B-dlx had a fan on the NB to cool it for 560fsb, it might have had a stock one on, but I don' remember putting one on it.

 

[Dinnercore]

I did the same thing with my Asus P5Q3... just enough room between the GPU and CPU cooler for a small fan:
View attachment 119356

Comparing the space situation to my recent PCs I wonder where all the space suddenly comes from. Like these old boards are packed so tight with heatsinks, fans and the high-profile RAM. It seems like a miracle that the coolers fit. While with todays boards there are thin slabs of metal as heatsinks and no fans. Much more space all around, everything from socket to PCIe slots seems to be set apart just a tad bit further so that you have all the room you need.


I meanwhile got 500FSB on 9,5x to boot, but not stable enough for benchmarks.

https://valid.x86.fr/6d0rvh

It would need a few steps more in voltage to get there, and that is critical with my air cooling at ambient temperature (20°C). On second note I got the memory above 1600 too, but timings are still eh.
All in all, I´m making progress. Btw at this voltage level I saw an idle power draw of 160 W for the whole system, its 100 W on stock voltage.

I hope you all don´t mind if I take my time with the results and charts. I first need to learn what is working and how, then I can start to plan out the frequencies and voltages I want to use for testing. As soon as I get there I can start with the games too.
Still have not decided if I will just switch over to a P45 board and do it all on this. I think it´s very important to decide on and stick with one single mainboard. Same goes for CPU revision.
I learned that C0 is not as efficient as E0 and differs in a way that it needs more voltage for the same frequency but puts out less heat with higher voltages. Overall it seems that the average E0 is a slightly better overclocker compared to C0, but a golden sample C0 can still beat most E0s, so the difference is not very obvious.

New material arrived, this is the waterblock I will use for watercooling, and I found some cheap CPUs that I will add to the pile. These Xeons are not a priority, got them out of curiosity for the higher multiplier.

 

[storm-chaser]

You have a good deal of voltage being pushed to that CPU. If I recall correctly you shouldn't need above 1.53-1.57 vcore for that clock speed. And maybe I missed it but what are you running for cooling on the E8500?

As for the P45 board option they do best with the core 2 quads. For what reason are you interested in moving to the P45 platform?

 

[Dinnercore]

You have a good deal of voltage being pushed to that CPU. If I recall correctly you shouldn't need above 1.53-1.57 vcore for that clock speed. And maybe I missed it but what are you running for cooling on the E8500?

As for the P45 board option they do best with the core 2 quads. For what reason are you interested in moving to the P45 platform?

I´m running the Asus V60 for air-cooling atm. Like I said before, the voltage might appear a bit higher vs. the actual value, the Strikers sensors are not very accurate. I set 1.58 in Bios and it should equal to about 1.59V real voltage. The reading is always a bit higher, this became clear when temperatures on this board @1.4V were lower than with the same CPU and cooler on another board @1.36V.
And on top of that I think the C0-Stepping CPUs need a little more voltage to maintain the same frequency as an E0-Stepping CPU.

Well the 790i Ultra SLI that I´m on right now is very inaccurate for measurements as a start. Voltages are off and I would have to measure real voltage with a DMM every time. On top of that it is tricky to work with, for example it gets stuck during post after changing anything in bios, I need to reset twice before it finally takes the settings and starts. It can´t restart from windows, it gets stuck during shutoff. No power state setting in Bios could fix this so far.
It has huge FSB holes, one spanning from 380 - 450 FSB. Any value inbetween will not post. No matter the voltage, CPU or memory setting. It does not handle the memory all to well as off now. Once it goes above 400 FSB it automatically reduces timings to try and maintain stability (its a build in 'feature' called P1 and P2) and generally seems to perform poorly with memory.
For very high OC it has a big FSB wall above 520-540 and is tricky to get any more.

I hope that some of those things might be avoided by using a P45 board like the TPower I45 or a Maximus II Formula. Maybe not tho, I´m not experienced with these chipsets. I could even get back to DDR2 RAM, which would help me because I have many different DDR2 RAM sticks around to switch out if I run into trouble with one specific combination of manufacturer.

EDIT: And it killed the partition table again. Windows install Nr.3 coming right up.

 

[Caring1]

I did the same thing with my Asus P5Q3... just enough room between the GPU and CPU cooler for a small fan:

It's almost like they are designed for a top down cooler, not some hideous tower.

 

[storm-chaser]

It's almost like they are designed for a top down cooler, not some hideous tower.

I dont really get your point? but yes the fan I used came from an older top down cooler.

 

[Caring1]

I dont really get your point? but yes the fan I used came from an older top down cooler.

The open fins on the VRM cooler and Northbridge are begging for airflow which top down coolers give (and also negates the need for extra, small fans on those areas), and tower coolers don't.

 

[Dinnercore]

The open fins on the VRM cooler and Northbridge are begging for airflow which top down coolers give (and also negates the need for extra, small fans on those areas), and tower coolers don't.

That is why my Asus V60 has the lower 50% of its heatsink fins angled slightly and with a little bend in the middle, to direct airflow directly into those VRM heatsinks. But this of course only covers one side.
My problem with top down stuff is that it usually only supports small fans and a very limited fin area. I prefer one thing doing one job. A fan on the VRM heatsink, one fan on the NB, one fan on the SB and the CPU-Cooler cooling just the CPU. And not having the compromise of less mass and smaller fan on the CPU for some minor airflow around the socket.

I do understand your point tho, and on an open testbench it would be ideal to have something blowing top down, I´m considering to build a mount for my 200mm fans on top of it.

EDIT: Another update, I have to report the first casualty. Turned out that Windows got really shredded this time, had to recover the other drive attached to it too. Upon reinstalling Windows with stock settings after a full CMOS reset it crashed multiple times and now I´m sure the poor CPU had enough. It degraded beyond the point of usability, it could run on higher voltage at stock speeds but I´ll let it rest in peace now. Big F for the champ, doing 4.7GHz one last time @1.6V.

Things I´m noting now, don´t go above 1.55V with ambient cooling. Don´t push over 1.4V VTT, and maybe stick to E0.
One thing that I just realized after inspecting my other E8500s is, that those parts are now over a decade old. They already have their expected lifetime behind them and were used for more than 10 years under unknown conditions. Many must have been overclocked before during daily usage. I don´t think I´ll get competitive results out of most of them. Looking at their bend pcbs and deep scars on the IHSs. Really makes me wonder what they saw during their life.
Rest in peace C0. I will remember you :,)

 

[hat]

Damn, 4.3GHz on a Q6600? That... can't actually be stable for daily use?

 

[delshay]

I did the same thing with my Asus P5Q3... just enough room between the GPU and CPU cooler for a small fan:
View attachment 119356

Question, why aren't you using a downdraft cooler? This will cool VRM & Chipset. Downdraft cooler is what I use here on my classic PC, but these type of cooler works best if the side panel has air vents. When I close the side panel here the difference is only around 1-2C because the cooler is pulling in it's own cool air from the side panel, & that's with a air filter fitted to the side panel.