CPU 700MHz Bus Speed Project

[wytiwx]

6 hours ago, ground said:

Have you tried any of the really low end gigabyte p45 boards too? I wonder if they can get anywhere close to 700 as well...

I think UD3L board may have a chance, as long as it has the key BIOS options we can tweak. Still, we need to find good NB more than good CPU. Will try at next stage.

[wytiwx]

4 hours ago, ludek said:

I had once gigaboom GA EP45 UD3LR. I have done 624 mhz fsb on my old old e8200 c0 stepping, all air. I was very proud, but have no validation, only screen. It was back in 2009 or 2010, when I was starting my fascination on OC. It was before I found hwbot (polish overclock site with only screenshot to validate)

 

Wytiwx, thanks for sharing the knowledge!

 

Edit:

I have found this! LINK
it was 1.7 Volt, stock NB AIR cooling

Good job, man, and nice to see you.

[wytiwx]

E8500 + EP45-UD3, CPU-Z 6.70GHz done!

https://hwbot.org/submission/4724125

can do SuperPi 1M @ 6.40GHz but low efficiency 

Edited April 11, 2021 by wytiwx

[ground]

711 fsb 1M kinda tells me there is much more to come, given that you already showed off 710 fsb booting... Excited for what is coming next...

[wytiwx]

the 710 air boot cpu is a E8600, not this E8500.

[Antinomy]

10 hours ago, wytiwx said:

can do SuperPi 1M @ 6.40GHz but low efficiency

Gigabyte P45 is not about efficiency

I think you need MCHbar editing for speed. But this is what makes them clock nice.

[George_o/c]

I remember this thread from back in the day at XS, might be helpful 

 

Congratulations @wytiwx awesome results! 

Edited April 12, 2021 by George_o/c

[wytiwx]

6 hours ago, Antinomy said:

Gigabyte P45 is not about efficiency

I think you need MCHbar editing for speed. But this is what makes them clock nice.

tried mchbar edition but saw littile progress on mem bandwith or super pi

[wytiwx]

10 minutes ago, George_o/c said:

I remember this thread from back in the day at XS, might be helpful 

 

Congratulations @wytiwx awesome results! 

THX, man. I will try this method.

[Noxinite]

1 hour ago, wytiwx said:

tried mchbar edition but saw littile progress on mem bandwith or super pi

I've been playing with my board the last two days, and I can say that the MCHBar is loose on the -UD3 boards as well as the -Extreme(s). So it is definitely doable to squeeze some efficiency out of them. Probably 4-5s at least in 32M. Might give is a spin on 32M 4GHz LCC for funsies.

Edit: not sure high fsb 1T is possible though, I set 1T at 640MHz and OS froze.

Edited April 12, 2021 by Noxinite

[wytiwx]

1 hour ago, Noxinite said:

I've been playing with my board the last two days, and I can say that the MCHBar is loose on the -UD3 boards as well as the -Extreme(s). So it is definitely doable to squeeze some efficiency out of them. Probably 4-5s at least in 32M. Might give is a spin on 32M 4GHz LCC for funsies.

Edit: not sure high fsb 1T is possible though, I set 1T at 640MHz and OS froze.

Perhaps I edited the HEX value at the wrong mchbar address, I need to study that topic about strap tweaks.

[ground]

IDE drive needed or is SATA not a problem?

[ground]

currently trying a pair of UD3Ps on Dice... so far

3/3 CPUs >715 FSB

2/2 Boards >720 FSB

Given what I’m managing on unbinned boards with unbinned CPUs on dice I am starting to believe in 800...

https://hwbot.org/submission/4737529_

 

Out of 6 random chips I've taken cold i've got now:

E8400 705

E8400 710

E8500 715

E8500 720

E8600 720

E8400 725 C0

E8400 725

E8400 726

 

EP45-UD3P rev 1.0 - 726 confirmed

EP45-UD3P rev 1.0 - 725 confirmed (did not try 726 chip in it)

EP45-UD3P rev 1.0 - 710 confirmed (only tried 1 random chip in it)

EP45-UD3 rev 1.0 - 725 confirmed (only tried 1 random chip in it)

Edited June 13, 2021 by ground

[ground]

On 4/9/2021 at 4:13 PM, wytiwx said:

Vcore: 1.40V to 1.60V (no need LLC)

Vtt: 1.40V to 1.60V

Vpll: 1.35V (air) to 1.59V (LN2)

Vnb: 1.46V to 1.70V (get a IHS on NB chip! make sure NB cooling is good!)

(see previous comment for my results, just adding a note for volts I ran)

Last couple days I ran a pair of UD3Ps and a couple Wolfdales and, at least for my chips and boards on Dry Ice, these voltages varied quite a bit from yours. Maybe I was still doing something else wrong, but with my chips on Dry Ice I had the following sweet spot voltages:
Vcore: 1.66V-1.74V (depending on chip)
Vtt: 1.6V-1.66V (depending on chip)
Vpll: 1.35V seemed just fine cold too, saw no gain going higher at all, rather lost some frequency. Maybe there are some chips that benefit from higher PLL, but mine certainly didn't.
Vnb: 1.64V-1.74V (depending on both board and chip; one board was higher voltage then the other. NB had an IHS and a waterblock base filled with dry ice (no contact fluid))

Further notes from the session
- I didn't run into a single <700 fsb chip with basically unbinned chips (quick test on asus x48 for 600 fsb air); this doesn't mean they don't exist, but it supports the claim that if you can't do 700 fsb the board or settings is the issue not the CPU
- even if a CPU can't boot 700 fsb doesn't mean its bad; one of my 725 valid chips failed booting anything over 660 fsb into OS.
- cold NB helps a lot; I only tested one of the boards with ambient NB and ran into a wall with CPUz open at 695 fsb. With cold NB, 700 worked really easily.
- I really need better DDR2; these sticks barely validated 750MHz on cold, I am pretty sure given the likely higher load through the 1:1 divider they are what limited me at 725 fsb, given it was 2 unique boards and 2 unique CPUs that walled there. If possible, I'll soon retry under Ln2 with some new sticks.
- also for next session: compare PCI card to no PCI card. Due to convenience I completely forgot trying a PCIe card, thus there is a decent chance it helped quite a bit with these results.
- I had one chip where MCH clock skew at 50 gave another 2-3 Mhz if I remember right, not a big difference. Wrong CPU clock skew (tried 0-150 with most chips) can cost quite a lot of frequency. Keeping CPU clock skew above MCH clock skew seems mandatory.
- don't bother trying asus p45 boards cold. I tried P5Q3 Deluxe and P5Q Pro, and both struggled even booting over 630 fsb with a 700+ gigabyte chip.
-710 fsb boot + 725 fsb in windows valid video

 

I'll certainly try again pretty soon, though first I gotta get some better memory and will likely airbin the CPUs on these boards now that I have a rough idea how to work them. Have a couple more boards I'm considering picking up...

 

[unityofsaints]

On 4/10/2021 at 12:13 AM, wytiwx said:

Sorry to keep you waiting so long, need some time to process my VCRs.

We all know, EP45 UD3R/UD3P are good boards to Core 2 Quard CPU for 500 FSB stability, 

but few OCer would like to use EP45/EP45T UD3 series for 700 FSB goal, they felt not so good, so as I did.

Since Mr. GRIFF made 708 FSB by EP45 UD3P, I begin to pay more attention to these UD3 boards. 

After severals weeks, I found that UD3 boards were really good.

Now, The most important point I want to say is, 

When CPU and NB are both with LN2 cooling, at least 50% of E0 stepping C2D E8X00 CPU can do 700 FSB CPU-Z validation, by EP45/EP45T UD3 series.

If you can't make this, the first thing you suspect should be NB, or BIOS tweak, not CPU.

The important BIOS options on UD3 boards for raw high FSB (in fact, they're almost the same as on the other P45 boards):

Vcore: 1.40V to 1.60V (no need LLC)

Vtt: 1.40V to 1.60V

Vpll: 1.35V (air) to 1.59V (LN2)

Vnb: 1.46V to 1.70V (get a IHS on NB chip! make sure NB cooling is good!)

Vmem: your mem choice

CPU Vref: 0.61x to 0.62x

NB Vref: 0.578x to 0.591x

CPU Clock Drive: 800mV to 1000mV

CPU Clock Skew: 50ps to 100ps

Performance Enhance: "Standard"

Mem strap: 1:1

CPU features: all to "disable"

Other BIOS options remain "auto"

Mem for DDR2 UD board: Micron D9GMH

Mem for DDR3 UD board: Elpida Hyper

VGA: PCI card for 720+ FSB

Thanks for this! Did you find any difference single vs. dual channel and between the different slots?

[wytiwx]

9 hours ago, unityofsaints said:

Thanks for this! Did you find any difference single vs. dual channel and between the different slots?

not yet

[ground]

7 hours ago, wytiwx said:

not yet

Coming back for more soon? Or do I and others have to try harder first? 

[wytiwx]

On 5/15/2021 at 9:12 AM, ground said:

Coming back for more soon? Or do I and others have to try harder first? 

You 750+ FSB first 

[TaPaKaH]

@ground: Still waiting for 300 BCLK X58

[ground]

17 minutes ago, TaPaKaH said:

@ground: Still waiting for 300 BCLK X58

I need LN2 again

20 minutes ago, wytiwx said:

You 750+ FSB first 

Got 3 Ep45-UD3Ps rev 1.0 ready now, just need a handful more chips and better D9GMH...

Edited May 16, 2021 by ground

[wytiwx]

updated 266 bus speed CPUs which did 700+ 

C2D E6300, 705 done, batch L632B.

https://hwbot.org/submission/4740813

[wytiwx]

C2D E7200, 730 done, batch Q748A

https://hwbot.org/submission/4754186

[ground]

I'd hope one of these will do it, now need good mem and a fitting CPU... And well, maybe I'll find more of these while looking for mem and CPUs.

[ground]

http://valid.x86.fr/ek3wlb

again confirming my suspicion that either me or my mem is the limit, I’m willing to blame the mem, for now at least. Tested 4 boards (3xud3p, 1x ud3) now, 3 confirmed 725, last I only tested with a random chip which walled at 710, though I still had >0.1V range on vmch so doubt the board was at its limit

[wytiwx]

23 hours ago, ground said:

http://valid.x86.fr/ek3wlb

again confirming my suspicion that either me or my mem is the limit, I’m willing to blame the mem, for now at least. Tested 4 boards (3xud3p, 1x ud3) now, 3 confirmed 725, last I only tested with a random chip which walled at 710, though I still had >0.1V range on vmch so doubt the board was at its limit

If you don't have good D9GMH, maybe you can try those crazy nanya d-die  just kidding