ASUS Maximus II Formula - new P45 king?

[Mr.Pink]

Try NB GTL's, made major difference from me but same as you above 475MHZ it works 1 reboot and will not next reboot.

Even had 500FSB benching but not for long as it freezes, I blame Bios and lack of Memory support as Corsair ain't even on the list.

Skews did nothing for me, but I cannot boot with a negative NB GTL but can at Auto or Positive.

Currently I am @ +40 NB for 475FSB and CPU GTL's are +20/+40/+20/+40, Tried to leave the 20's at Auto and also tried all Negatives but not sure what's best as of above info. because its simply random each reboot.

It makes no difference if I use 1 Module or all 4 Modules or even my older Corsair Modules, still hard to get over 475FSB and 1175MHZ seems limit though my Ram can do more.

My advise would be to sell of some of the non needed kit and buy a Q9650 and see how far it can go on an A3 after all that's what the Gigabytes are.

that quad still giving you grief???

I found in the beginning positive GTL values worked better for me I'm using a negative value now and am limited to 471FSB above this and my clocks fail...
525FSB is a walk in the park for my 8400 but either my 9550 or 9650 struggles after 470FSB...

[GFORCE100]

i know, but this thing haven't any connection with e.v.o.'s modded bios ...

Well yes and no. E.V.O's BIOS patching is a more elegant way of going about it rather than having to install additional software to trick the drivers.

[humeyboy]

that quad still giving you grief???

I found in the beginning positive GTL values worked better for me I'm using a negative value now and am limited to 471FSB above this and my clocks fail...
525FSB is a walk in the park for my 8400 but either my 9550 or 9650 struggles after 470FSB...

Stable for 4 days with multiple reboots now @ 475FSB.

I think its the LLC ain't 100%, I now think I know what to look for in voltages each reboot to know if its going to be stable or not (obv is stable when not Priming).

I need about 1.168v for 3.6GHZ and that means 1.2v is Bios even with LLC as it drops (not droops).

Now it seems to be 1.192v with 1.2v in Bios so LLC is holding better but its not ideal as my previous Mobo which had LLC (NF 780I) would be exactly what you set it to, no VDrop and no VDroop.

I will keep an eye on it every reboot and see if 1.2v ends up 1.168v or 1.192v on load.

I do not get why that would cause issues though as I lowered the Multi and raised Voltages throughout all settings to make up for this to test and it again was a hit and miss.

[Mr.Pink]

I'm currently testing the mosfet temps, I think its the reason this board is so unstable at high FSB..
I have 2 MIIF boards one with stock TIM and cooling the other with mosfet blocks, from just the first few crude tests the water cooled mosfet board clocks higher.

[Haes]

Hi Guys,

bought a Q9550 E0 yesterday. I want to get 4,0 Ghz 24/7 and my first impression is that these little 45nm suckers are really tricky to handle compared to my Q6600 G0.

My Q9550 is a 1,25V VID, pretty standard i would say, but testing Small FFT last night got me turning wild, cause the cpu shows no stable area instead while reducing Vcore stable GTLs swtch over the whole bandwidth.

Small FFT (all BIOS 1901):

8.5*471 (4Gs):

VC/PLL/NB/VTT/CPU GTL

1.35/1.5/1.29/1.2/10 -35 10 -35 stable , else BSOD about 5 to 10 minutes
1.34/1.5/1.29/1.2/60 10 60 10 stable, else BSOD about 5 to 10 minutes
1.33/... nothing stable for now, BSOD about 5 minutes
<1.3/... can boot win7


Got the sucker at 1.36/1.55/1.44/1.31/40 -5 40 -5 stable at 475*8.5(~4.1Gs), else BSOD while stressing

Large FFT BSOD for now

Temps are all normal and 24/7:

4 Gs load 54-58°C
4.1 Gs load 56-60°C

(all air-cooled with Noctua U12P)

...so that's not the problem.

Compared to my Q6600 (http://www.xtremesystems.org/forums/...postcount=3797 - could even lower VTT to 1.34V last week) I can't hit the max stable FSB with the Q9550. I need more Vnb to get the same results like with my Q6600 and 475 is for now Small FFT stable but with (for me) no reasonable Volts for 24/7 - even when temps are pretty ok.

Perhaps I've to reset my ocing and take a break.


Anyway, I aprecciate 24/7-settings Q9550 SLB8V with M2F ...

Best Regards

[humeyboy]

I'm currently testing the mosfet temps, I think its the reason this board is so unstable at high FSB..
I have 2 MIIF boards one with stock TIM and cooling the other with mosfet blocks, from just the first few crude tests the water cooled mosfet board clocks higher.

I have a later Mobo (still a Rev A2) but no heat issue with Heatpipes so for now as lazy I left the TIM.

I did ask weeks ago if any7one got more stability redoing the TIM and who also did not have heat issues but got no feedback.

I redid the TIM on the 680I and 7680I as they are hot Mofo's.


@ Haes, IMO no, the 45nm CPU's are easier to clock than 65nm CPU's.

The Q6600 is an old dog and ain't great to get a high FSB on at best of time.

Its just this Mobo to blame TBH, a Dual is a breeze to get high FSB on but a Quad is harder.

The Gigabyte does far better.

[Haes]

I have a later Mobo (still a Rev A2) but no heat issue with Heatpipes so for now as lazy I left the TIM.

I did ask weeks ago if any7one got more stability redoing the TIM and who also did not have heat issues but got no feedback.

I redid the TIM on the 680I and 7680I as they are hot Mofo's.


@ Haes, IMO no, the 45nm CPU's are easier to clock than 65nm CPU's.

The Q6600 is an old dog and ain't great to get a high FSB on at best of time.

Its just this Mobo to blame TBH, a Dual is a breeze to get high FSB on but a Quad is harder.

The Gigabyte does far better.

Well, it can be that Giga does better in the bandwidth with Quads, but with my Q6600 and M2F I get 490FSB stable with reasonable 24/7-setting, temps are really good ~52°C NB at load. So for me Asus did a good job, not brilliant, but good. I agree that a 45nm is better to oc as a 65nm and I hoped to take my settings with few GTL/Skew adjustements to the Q9550. But like we all know each CPU is different, so I've start over.

I forgot to mention, that I tried to stay in a 40-45mV gap like simps guide shows, but I found on forum delux some use 10mV gaps or else with the SLB8V für 24/7 by reasonable volts.

[humeyboy]

No way m8, your going to need put in as much time with the new CPU as you did with the old CPU.

The GTL's are different for a 45nm over a 65nm.

As for the "GAP" in GTL, only the Negatives have 40/45/50 so 0.5's, the Positives have 40/50/60 so 10's

You adjust the GTL going by your VTT (FSB Termination Voltage) into the proper Calculation (in a Sticky in this very section), it could end up nowhere near 40-45, I have seen peeps on +/-60 here

[Haes]

No way m8, your going to need put in as much time with the new CPU as you did with the old CPU.

The GTL's are different for a 45nm over a 65nm.

As for the "GAP" in GTL, only the Negatives have 40/45/50 so 0.5's, the Positives have 40/50/60 so 10's

You adjust the GTL going by your VTT (FSB Termination Voltage) into the proper Calculation (in a Sticky in this very section), it could end up nowhere near 40-45, I have seen peeps on +/-60 here

Yeah, that's what I thought too...

And I resetted all, put all to my starting position... calculated GTL accordingly to the VTT. which would 1.35*0.67-GTL=1.35*0.63 => GTL= 54mV.

So I played up from 40mV with a gap between 40-60mV and get Win7 boot + some benches/validation by 40 -5 40 -5 NB 30 (which is pretty standard for the NB GTL with my M2F) and the CPU GTLs are 1 of 3 functional setups with 471*8.5 by Small FFT that I did last night.



Vcore is way too much, just don't turned the VC down...

[humeyboy]

I get 1.35v to work out at 0.0432 so 42mV so choice of -40 or -45 or +40 or +50 GTL's

[Haes]

Ok, could bench a little first LargeFFT without errors by 490 FSB like with my Q6600. That's a start .

[seban]

...I have seen peeps on +/-60 here

Well I am +50/+90 nice and stable at 4ghz q9550 - also the higher you go with vtt and vc from nominal values the bigger allowed gap becomes.

And Simps guide is ok for mediocre clocking - if you want to squize max from MB and CPU you got to tune all parameters individually all the time not only at beginning as he suggests - this is why he did not notice that gap is getting bigger as you adjust vcc and vtt while clocking cpu.

[Mr.Pink]

I did not find simps guide very usefull, it did not clock my board higher than my own method....
If I was looking for a 10-15% overclock ..... its built in to the bios

[Haes]

So, I got my Q9550 E0 stable in SmallFFT with 470*8.5 by 1.34V VC , 20/70/20/70 GTL and good temps 55-59°C (Noctua U12P) on the 1901.

It's ok, but my own expectations are a bit higher like 470*8.5 by 1.3-1.32V VC or better 482*8.5 (4.1Ghz) by 1.34V VC.

In fact, I got 475*8.5 (4.05 Ghz) by 20/-25/20/-25 stable in SmallFFT with max 62°C Temps and 1.36V VC.

Some ppl in the forumluxx (german community) have better results with the 1307 Bios,although it wasn't designed for Q9XXX. I'll give it a try and I've E.V.O.'s 1901.2XX Bios too.

A good thing about mine is a very plain templvl. There are times in the stress where all cores have the exact same temp for a long period (1hour or more) and jump together. Never seen such a tempbehavior and the max tempdifference is 4°C.

But I've another matter:
My RAMbehavior is really strange. Corsair sent me a new pair (2nd pair already) DDR2 8500er Dominators last week. While with my Q6600 a never had a problem with memtest86+, I've now errors no matter which setting I used all on defaults or loose timings, more VDimm, less frequency or more VNB and no matter which slot or single/dual channel.

Blend is stable by even 460FSB and even ~ 1100MHZ RAM, PL10 or a bit tighter timings and I know Blend isn't such a stress test like Memtest. So what can it be?

Best Regards

[Mr.Pink]

test one stick at a time on stock speeds (1066) , if they pass I would start to play with dram clock skew or up vdimm there good till 2.2volts iirc

[eSp!s0]

Here is an article: "SLI on X38, X48 and P45 possible?": http://www.computerbase.de/news/hard...i_x38_x48_p45/
The article is in german, so you have to translate it!

[humeyboy]

Corsair are not even on the QVL.

Only the older Rev 1.1 Dominator can do 1200mhz but still cannot do it on this Mobo but can on other Mobos.

[Haes]

test one stick at a time on stock speeds (1066) , if they pass I would start to play with dram clock skew or up vdimm there good till 2.2volts iirc

I did it already and all failed on every slot in default by about 4% to 10% and the curious thing about that is, that with my Q6600 a month ago or so they passed memtest every run.

I do exclude the new cpu, that can't be an error source.
Perhaps something happened with my mobo while ocing, although I never went over the max allowed specifications.

I had to switch my cpu, just to be sure.


P.S. Switched back to 2104, set all on default and bumped VDIMM to 2.15V. Memtest seems to run without problems. Very strange indeed, never had any mem issues on 1307 or 1901 at all. Dualchan test is running now.

[humeyboy]

That's the temperamental bug back again.

I changed Bios setting to try higher than 475FSB and after failing loaded my saved profile again with stable 475FSB settings.

Now I'm unstable, this sure is 1 buggy POS.

The fix is to just enter and exit the Bios (don't change anything) till its a stable boot up again.

[Mr.Pink]

Corsair are not even on the QVL.

Only the older Rev 1.1 Dominator can do 1200mhz but still cannot do it on this Mobo but can on other Mobos.

i would say yes they can

[GFORCE100]

i would say yes they can

I think what he meant is you cannot do 1200MHz RAM stable on the M2F. I.e. not just claim it's stable with an everest benchmark but do the full agenda which at the minimum would signify Prime95 Large FFT and LinX for a few hours.

Also post your BIOS settings if you're not too lazy No seriously, then others can confirm your findings too and put this myth/fact to rest.

[Mr.Pink]

many kits will do 1200MHz on this board, no my 1240 run was not stable
1200 is doable with alot of tweaking, it wont come easy..
My bios profile is here, you just need to look for it

I have a 1200MHz stable shot someplace I'll hunt it down and post it up, iirc it was a prime run for 9 hours so thats good enough for me to call stable
I dont run that Corsair kit 24/7 its a benching kit, I have a PC8500 kit of Mushkins that will do 1200 on this board, sadly my 2 kits of GSkills will not go above 1140-1150MHz

[humeyboy]

Nope I meant what I said.

I and many others inc Gold***** (cant remember his nick) cannot get 1200mhz on ram proved in other Mobo's

I can only get 1175mhz then it will not even post Bios, that's with 1 stick or all 4, no matter what voltage/settings.

Someone here showed Gold**** the 1200mhz Corsair and I think he went back and tried it and finally got 1200mhz but I aint sure if stable

All I know is its not worth the bother, Asus need to get better support in new Bioses.

[Mr.Pink]

I agree the min gains from 1200MHz ram is not worth the hassle...
If Asus had put effort into this board and released a bios that unleashed its full potential I think you would see similar performance that the UD3P has with quads.. sadly this a dual friendly board and seems to hates quads

[humeyboy]

This Memory should post the Bios esp only 1 stick with major voltages, it simply ain't supported.

I'm tempted to get 8GB of 1200mhz G.Skill sent here from USA by a buddy.

I wonder if it will work as its also not supported and if this will help me get 500FSB as it will be o the harder 1:1 ratio.

I stand by the fact we need better Bioses.