X48 Rampage Formula Preview.

[leo27]

I'd say that the build you recieved A-Grey was meant for the engineers to test the new features implemented. If all went well with that they'd then make it into a ezyfalsh II compatible bios and send it to you.

If this is indeed the case a proper beta bios will be released in a week or two like they did with bios 802 and then once we do their testing they'll release a final bios revision like 803.

[Antineutrino]

Then lets hope the guys at Asus will prepare a new bios soon....

Intel C2Q 9650 @ 4.2GHz (9x467) 1.256v / 4.05GHz (9x450) 1.192v (24/7)
ASUS Rampage Formula (BIOS 0902)
cooled by Prolimatech Megahalems
G.Skill 2x2GB F2-9600CL5D-4GBPI
Gainward Bliss 8800GTS/512 @ 792/1890/1102
2x WD 640GB SATA@Raid0
Seasonic X-750
Antec Nine Hundred Two Case
Windows 7 Ultimate 64

Hi!
How did you manage to reach these high frequencies with so low voltage?

Is it LinX - Prime95 rock solid stable ?

[Trust13]

Hi!
How did you manage to reach these high frequencies with so low voltage?

Is it LinX - Prime95 rock solid stable ?

My Q9650 has a VID of 1.200V, so it needs only 1.84v for 4GHz, 1.192-1.200v for 4.05GHz and 1.256v for 4.2GHz.

I use nearly the same settings like Alien Grey posted a while ago (Thanks for your finding Alien Grey )

Here are some older screenshots:

4.0Ghz


4.05GHz

i changed this setting actual to memory at tREF 12800, vNB 1.28v (1.29v BIOS) and VTT 1.12v

4.2GHz



my older Q9650 needed much more voltage for 4GHz (VID 1.275) to be stable:




but take a look at this guy:
4 GHz with only 1.152GHz LinX stable VID (1.162v)


i don´t use LinX for stress-testing, because it stresses only the CPU and produces more load than you will ever have with every other application

[Alien Grey]

but take a look at this guy:
4 GHz with only 1.152V LinX stable VID (1.162v)


i don´t use LinX for stress-testing, because it stresses only the CPU and produces more load than you will ever have with every other application

Fixed it for you.


That's also something I would like to see fixed in the BIOS and that I've asked to the ASUS Tech Support.

My Q9650 has a VID of 1.1875V and I also can't use less than 1.20V to have it 100% stable with Prime95 or I end up with a Machine_Check_Exception BSOD. I Never had a failing core in Prime95 or a wrong result in LinX because of to low voltage so it should be possible to use less than 1.20V and I need it to be fixed or I won't get it stable at 4.5GHz with 1.36V.

Edit:

I think that this has to do with the Ai Clock Twister. With Moderate I need 1.22V and with Lighter 1.20V to have it stable at 4.05GHz.

[Trust13]

Fixed it for you.

Oops...

to get my setup stable with LinX i need exactly the same Voltage at load then with Prime 8K, but then i have to raise the vCPU in Bios for 2 notches.
Maybe try to change the CPU-GTL, my actual setting is at 0.67. There is a good chance, that you can lower your vCPU.
Much harder was to get Prime Custom to be stable. You can run Prime Small and Large for hours, but then it fails at Custom (FFt´s in place). Most times when it changes the testing from large to small.
I had to play a little with VTT, GTL, and vNB to find the right setting.
I couldn´t get a tREF of 16383T stable. Prime95 Custom failed at 512k. With 12800T it runs fine. I dunno why, but maybe because my troughput is a little higher than yours (Ai Clock Twister Lighter):

[Alien Grey]

I think that you've got a slightly better bandwidth because you use a different OS. I never had any problems using tREF 16383T with BIOS 0902. With BIOS 0803 it was sometimes stable and the next reboot it was unstable. If I use a different CPU GTL setting it doesn't allow me to use less CPU Voltage nor does changing FSB Termination Voltage. For 4.05GHz it's 1.20V+ or BSOD at 6+ hours Prime95 Blend.


Changing from Moderate to Lighter didn't only let me use less voltage for the same performance there is also a nice temperature improvement. The core temperature dropped with 3° Celsius under full load with LinX.

These are two tests that I did to see if the dye that I'm using in my loop isn't causing problems and gunk up my rad or blocks. Both tests where done with the same room temperature of 18° Celsius. The first one is with Ai Clock Twister Moderate and the second one with Ai clock Twister Lighter.

[Antineutrino]

My Q9650 has a VID of 1.1625 Volts but I cannot be LinX stable with a Vcpu lower than 1.42500 Volts in BIOS.... Am I missing something?



Perhaps is there more than one range where raising the voltage makes the CPU stable?

For example a 1.1625-1.27500 first range... CPU is stable with low heat.
Than it becomes instable due to increased heat...
Raising the voltage more to find a second range and so on till the maximum heat dissipable by the heatsink...

Is that true?

[Trust13]

I think that you've got a slightly better bandwidth because you use a different OS. I never had any problems using tREF 16383T with BIOS 0902. With BIOS 0803 it was sometimes stable and the next reboot it was unstable. If I use a different CPU GTL setting it doesn't allow me to use less CPU Voltage nor does changing FSB Termination Voltage. For 4.05GHz it's 1.20V+ or BSOD at 6+ hours Prime95 Blend.


Changing from Moderate to Lighter didn't only let me use less voltage for the same performance there is also a nice temperature improvement. The core temperature dropped with 3° Celsius under full load with LinX.

These are two tests that I did to see if the dye that I'm using in my loop isn't causing problems and gunk up my rad or blocks. Both tests where done with the same room temperature of 18° Celsius. The first one is with Ai Clock Twister Moderate and the second one with Ai clock Twister Lighter.

You really found your perfect settings.. i never saw a Q9650 with only 1.10v VTT and 1.25v vNB at 450MHz FSB before..
I for myself never test my setup for more then 2+1/2 hours with Prime Custom, had never problems. So i mangaged to run my setup with 1.10v VTT too, but it crashed while i was gaming. I had to raise VTT to 1.12 and vNB a little.

My Q9650 has a VID of 1.1625 Volts but I cannot be LinX stable with a Vcpu lower than 1.42500 Volts in BIOS.... Am I missing something?



Perhaps is there more than one range where raising the voltage makes the CPU stable?

For example a 1.1625-1.27500 first range... CPU is stable with low heat.
Than it becomes instable due to increased heat...
Raising the voltage more to find a second range and so on till the maximum heat dissipable by the heatsink...

Is that true?

There is no "first and second range". The only things you have to look at, is the voltage for the CPU, which should be about 1.365v max. at load (24/7) and the temperature of the cores, which should be about 70°-75 C max. for a short peroid. But with high temperatures like this the system could get unstable.
I think you should start at the beginning. With such a low VID you should be able to run your system with 4.2Ghz at about 1.25v (full load). There is definitly something wrong in your settings.
Maybe you post your template here? Let´s take a look in it.

[Alien Grey]

I've just received a PM from Ket who knows how to mod a BIOS and he thinks that cross flashing the BIOS between the Rampage Formula and Rampage Extreme is probably possible and replacing the memory tables shouldn't be a problem at all.

My only concern is if the CPU GTL settings from the ASUS Extreme BIOS are going to work on the Rampage Formula.

[Trust13]

I've just received a PM from Ket who knows how to mod a BIOS and he thinks that cross flashing the BIOS between the Rampage Formula and Rampage Extreme is probably possible and replacing the memory tables shouldn't be a problem at all.

My only concern is if the CPU GTL settings from the ASUS Extreme BIOS are going to work on the Rampage Formula.

..and what happens when you access the additional stuff, which isn´t existend in the Formula Bios, so the EPU II Phase Control and the BIOS-Flashback part

[Alien Grey]

..and what happens when you access the additional stuff, which isn´t existend in the Formula Bios, so the EPU II Phase Control and the BIOS-Flashback part

Who knows. We will only know it if we have a working BIOS.

[Fritz_the_germ]

Yoh guys,
I can´t believe how much you are obsessed to get a new BIOS. Even to risk a bad flash and destroying your board by that way.
Why don´t you realize that ASUS won´t create a new BIOS and buy yourself a better board to get all kinds quality you guys want to?

[SoLoR]

because spending 200€ on new board seems like a waste of money to squeeze 10 more FSB and you dont destroy board with bad bios flash anyway, you can tecnicaly recover kinda easy if you have some old computer with LPT port and you waste like 1€ for materials to make your self SPI cable however everyone would prefere to stay away from doing this kind of stuff, but its fun in certain way

[Alien Grey]

And if we screw up there's still the ASUS Repair Kit.




http://www.overclockers.com.au/image...ics/3oct/1.jpg

[SoLoR]

Ye and then you can buy this, if you have 2k € to spend for cpu, mobo and rams I said ill never buy another asus mobo again, but never say never Rampage III Extreme seems impressive.

[OLDcomer]

[Alien Grey]

For the price that we paid for the board this should have been included.

[Alien Grey]

I did the pencil mod today and it didn't fix my problem with Loadline Calibration but I didn't really expect it to be fixed because it isn't really a voltage problem.

The vdrop with the pencil mod is only 0.008V between idle and full load with LinX so I don't need the Loadline Calibration anyway.

[Trust13]

I did the pencil mod today and it didn't fix my problem with Loadline Calibration but I didn't really expect it to be fixed because it isn't really a voltage problem.

The vdrop with the pencil mod is only 0.008V between idle and full load with LinX so I don't need the Loadline Calibration anyway.

What was your problem with LLC enabled in first place? Everywhere i read not to use it with 45nm CPU´s, because of the overshooting voltage pikes which may damage your CPU. Does a pencil mod not does exact the same thing?
True or not, i tried it yesterday and on full load everythings seems to be fine. I could lower my CPU Voltage in Bios from 1.28xv to 1.20v. The vdrop was only 0.008v and the vdroop was +0.008v! (I think you ment vdroop, not vdrop in your post). But in idle state suddenly the computer crashed, no video, and DET RAM in the LCD-Poster...
Is it stable now for you with the pencil mod?

[Alien Grey]

What was your problem with LLC enabled in first place? Everywhere i read not to use it with 45nm CPU´s, because of the overshooting voltage pikes which may damage your CPU. Does a pencil mod not does exact the same thing?
True or not, i tried it yesterday and on full load everythings seems to be fine. I could lower my CPU Voltage in Bios from 1.28xv to 1.20v. The vdrop was only 0.008v and the vdroop was +0.008v! (I think you ment vdroop, not vdrop in your post). But in idle state suddenly the computer crashed, no video, and DET RAM in the LCD-Poster...
Is it stable now for you with the pencil mod?

The problem with LLC enabled is running in to a BSOD. It's so bad that most of the times I can't get in to my OS.


With the pencil mod it's perfectly stable with the CPU Voltage at 1.25V in the BIOS. That gives me 1.224V idle and 1.216V full load with LinX. I can't run it stable anymore at 1.20V after the pencil mod but that's no problem because my load voltage with BOINC was always around 1.22-1.23V before the pencil mod.

If I try to use lower voltage it's Machine_Check_Exception sooner or later.

[Leeghoofd]

Flashed you guy's Asus bios yesteday on the Rampage Extreme : got twonew features now :



Using these I got 640FSB sort of stable (superpi till 16m, 32m still freezes), beforehand my rig would constantly freeze at 635FSB with just Superpi 1Mb... PI32mb is stable now up to 635FSB ( before 620FSB ) Will test 3D stability tonite...



No idea what special 133 mode is...

[OLDcomer]

Leeghoofd:


5МHz FSB increase would be lovely. Man, these ASUS guys are slow. How long should we wait for a new Rampage Formula bios.

[Alien Grey]

I think I know what Special 133 Mode is. It's probably a strap change at FSB 466MHz. This what we need to run our Quad at FSB > 466MHz.

The setting is by default disabled to have the same performance at FSB > 466MHz but if you enable it you should be able to push it higher on FSB. Maybe a Quad at FSB 500MHz stable is now possible.


This is just great. After all my hard work trying to figure out what could be the problem and how to fix that in the BIOS, they added it to the Rampage Extreme BIOS and we've got nothing and they probably say you Alien Grey, if you want to overclock your Quad, buy an ASUS Rampage Extreme.

[SoLoR]

Well somehow more and more i think, more and more i suspect it was a mistake on Asus part and they where developing Extreme bios instead of Formula one :s because they did added features you requested.... All Extreme users should praise you! But its funny, how Fomrula topics/forums are stil kinda busy and Extreme is not and they make bios for Extreme and not Formula... zeh!

I seriously have no idea what the hell is so big deal of converting Extreme bios in to Formula one (outside change from DDR3 to DDR2 and probably timings that go along). I mean, i bet that for some asus engineer in bios department its like a day of work... They have bios code written in assembler, its just matter to put memory stuff from Formula to Extreme bios and disabling things that Formula mobo does not have (dual bios for ex).... SERIOUSLY SLACK! For someone that understands what needs to be done it should be pice of cake... Sadly outside mobo manufactures bios departments, people like that are almost not existant. Even people that have lots of coding skills and went to disassemble bios, they said its almost impossible to add new functions, unless you have code of that bios. You see in coreboot support (coreboot is opensource bios), that they basicaly dont support any mobos except few "test" models and even those in 90% lack functionality. Unless you dont have help from mobo engineers and access to specs or whatever, you cant make your own bios.


edit: maybe even in this opensource department something will move in a year or two, just noticed this "coreboot now has initial support for the AMD RS780 / SB700 chipset. The code was contributed by AMD, thanks! ". Sadly i dont see Intel or Asus contributing anything in last 2+years would make things faster if manufactures would start supporting opensource, instead of paying licenses to ami,award,phoenix... silly people :s

[Alien Grey]

And now they've got a BIOS with a new feature that probably nobody is going to use because just enabling it wont make the board stable at higher FSB but it's going to need a lot more tweaking of other settings I think that they also don't use on that board.

And we. Well we wait and wait because that's what we've been doing all the time and that's something that we're very good in.