X48 Rampage Formula Preview.

[papatsonis]

@ A-Grey , how much VMCH u use for 450fsb/1200ram on formula? and what vtt for cpu?

[Alien Grey]

Eyecandies




A-Grey i wish i had your lowVol. chip in my hands

By the way our g.skills cost about 105€-130€ in Germany right now:

http://www.hardwareschotte.de/hardwa...28PC2-10000%29

but the thing is can 4 sticks work without having stability problems???

Nice very nice but to high vTT, vMCH and probably vCORE for these speeds.


After I finished my tests at 4.23GHz and DDR 1253MHz you're for sure going to wish you had a low vid Q9650.

I think that FSB 500MHz and Quads are possible with BIOS 0902 with lower voltage.


I'll give you a clue: I've seen the LIGHT.


I have no idea if it's possible to use 4 X 2 Giga DDR 1200MHz and a Quad.

@ A-Grey , how much VMCH u use for 450fsb/1200ram on formula? and what vtt for cpu?

I need 1.33V in the BIOS or 1.31V real for vMCH and 1.24V in the BIOS or 1.14V real for vTT for FSB 450Mhz.

[papatsonis]

omg!!!... i envy you :p for such low vmch :S

[greg.m]

I think that FSB 500MHz and Quads are possible with BIOS 0902 with lower voltage.


I'll give you a clue: I've seen the LIGHT.







I need 1.33V in the BIOS or 1.31V real for vMCH and 1.24V in the BIOS or 1.14V real for vTT for FSB 450Mhz.

Give me your quad or... maybe ill take the risk and rma
my quad for her faulty sensor

[Alien Grey]

Give me your quad or... maybe ill take the risk and rma
my quad for her faulty sensor

I just discovered that mine has three stuck sensors.

core#1 at 35° Celsius, core#2 at 30° Celsius and Core#3 at 40° Celsius. For core#0 I have to wait for lower room temperature. The lowest I've seen on idle was 24° Celsius.


My G.SKILL F2-9600CL5D-4GBPI is stable at 1253MHz with 1.92V.

[Alien Grey]

omg!!!... i envy you :p for such low vmch :S

I think that everyone can have it stable with such a low vMCH. You have to use CPU Clock Skew Delay 200ps and NB Clock Skew Delay 100ps.

I'm just lucky that my NB can do it with the low voltage and my other BIOS settings.


My clue for greg.m isn't for him only. It's for everyone with a Quad.

[papatsonis]

a gave a try the above skews (not because i had high hopes.. but in case...) with almost instant freez.. at my previous stable settings with 0.025 less vmch. form my point of view, (correct me if im wrong) u only need to "correct" the NB clock skew "delay". at least in my "gigabutt":P its description, says more or less, "defines how much time the nb clock ADVANCES cpu clock".. naively thinking that if i have instability from X38, could be due to chipset's failure to "follow" cpu's clock, i played with nb clock "delay" and found small stability gains to 50mv. Also assuming these two settings counter react to each other, whats the point of setting both of them !!!!!!? eitherway the bottom line is that i had no voltage gains with any other skew setting than NB clock skew "DELAY"-> advance in fact , of 50ps.

[leo27]

To be able to lower vNB you also have to have vTT and the GTL reference voltages set correctly in addition to the CPU and NB clock skews.
Otherwise the clock skews are compensating for bad GTL votlage settings rather then allowing for an increse in stability and thus allowing for lowering vNB.

[Alien Grey]

There's also another important setting in the BIOS that needs to be adjusted. Something that we forgot it was usefull because it didn't work with the older BIOS revisions.


I'm going to explain it later and probably post the most important picture of this thread.

[leo27]

You've got me curious A-Grey whatever it is must be good since i see your q9650 clock is significantly better than whats in your sig

I might be able to get some more out of my q6600 after all

[Alien Grey]

My new 24/7 overclock.



Q9650@4.23GHz - DDR 1253MHz - tRD 6

Ai Overclock Tuner [Manual]
CPU Ratio Setting [9.0]
FSB Strap to North Bridge [400MHz]
FSB Frequency [470MHz]
PCIE Frequency [100MHz]
DRAM Frequency [1253MHz]
DRAM Command Rate [2N]
DRAM CLK Skew on Channel A/B [Auto]
DRAM TimingControl [Auto]
DRAM Static Read Control [Enabled]
Ai Clock Twister [Light] -> Here's the clue
Ai Transaction Booster [Manual]
Common Performance Level [06]
Pull-In of CH A/B all disabled
CPU Voltage [1.38750V]
CPU PLL Voltage [1.50V]
North Bridge Voltage [1.41V]
DRAM Voltage [1.92V]
FSB Termination Voltage [1.32V]
South Bridge Voltage [1.05V]
SB 1.5V Voltage [1.50V]
Loadline Calibration [Disabled]
CPU GTL Voltage Reference [0.63X]
NB GTL Voltage Reference [0.67X]
DRAM Controller Voltage REF [Auto]
DRAM Channel A/B Voltage REF [Auto]
CPU Spread Spectrum [Disabled]
PCIE Spread Spectrum [Disabled]
CPU Clock Skew [Delay 200ps]
NB Clock Skew [Delay 100ps]

CPU Ratio Setting [9.0]
C1E Support [Disabled]
CPU TM Function [Disabled]
Vanderpool Technology [Disabled]
Execute Disable Bit [Disabled]
Max CPUID Value Limit [Disabled]
Intel(R) SpeedStep (TM) Tech. [Disabled]

I have to test it for a longer period with Prime95 Blend. I did a 1.5 hour run yesterday without problems but I want it to see Prime95 Blend stable for 6 hours or more.


As you can see Ai Clock Twister at Light makes it possible to have a stable overclock above FSB 450MHz with reasonable NB Voltage and FSB Termination Voltage.
This isn't possible with any of the previous BIOSes. Changing Ai Clock Twister to Light didn't make any difference in stability because the problem was situated in the memory configuration in the BIOSes.

I was lucky that I could do 10-15 MHz above FSB 450MHz with Ai Clock Twister at Moderate. I guess that my NB clocks a little better than average.

[papatsonis]

quick quesiton, which bios u used to have these results? (0902 i see in your signature?) , and also.. can tha tbe crossflashed to P5E? , or needs special version of it?

[Alien Grey]

quick quesiton, which bios u used to have these results? (0902 i see in your signature?) , and also.. can tha tbe crossflashed to P5E? , or needs special version of it?

It's BIOS 0902 and I think you can crossflash it without problems.

[leo27]

I notice your using a tREF of 3120T, at those memory clocks wouldn't the highest tREF 16383T serve you better

[Alien Grey]

I notice your using a tREF of 3120T, at those memory clocks wouldn't the highest tREF 16383T serve you better

Yes it would but I wanted to know if it's possible with the BIOS default 3120T.

I'm going to test it with 16383T to see if it stays stable without changing anything in the BIOS.

[greg.m]

Yes it would but I wanted to know if it's possible with the BIOS default 3120T.

I'm going to test it with 16383T to see if it stays stable without changing anything in the BIOS.

Im gonna use the clue you gave us and try to achieve higher speeds
Lets see whatll happen

[leo27]

just out of curiosity how much bandwidth do you think would be lost going from clock twister stronger to clock twister light????
Cause atm i'm at ddr2 850 with clock twister stronger.

[Alien Grey]

just out of curiosity how much bandwidth do you think would be lost going from clock twister stronger to clock twister light????
Cause atm i'm at ddr2 850 with clock twister stronger.

You don't have to use light go with moderate first. I think that you are going to loose around 100 or 150Mb/s. But you gain a lot more with the higher FSB if you can get higher memory speeds stable.

[Alien Grey]

Im gonna use the clue you gave us and try to achieve higher speeds
Lets see whatll happen

Let's see some results.

[leo27]

To run my Q6600 at 3.4Ghz i'm at 8x425 so to keep this frequency i'd need 7x486 which might be pushing the FSB a little.
I could try 7x457 for 3.2Ghz but i think when i get a cool day i'll see what max FSB i can get stable using a CPU multiplier of 6 to elimate the CPU as a factor assuming my memory can clock above 900Mhz. Although since i'm at 850Mhz with very tight timings at 1.80v in the bios (which gives me 1.87 in windows thanks to load line callibration) i think it's likely i could get upto around 1000 ish with CAS 5.
I can't do any stress testing at the moment due to an ambient tmep of about 30C thanks to the wonderful Australian summer though.

[greg.m]

I was wondering when asus will give us a new bios (if its going to happen)

Someone from Asus here???

I know our lovely Rampage is kind of "old" but i think we deserve something new to play.

[Alien Grey]

A new BIOS would be cool. A new BIOS with tREF.

[neo_rtr]

@ GREG.M
Nice Sig.. Nice results..

-> Project Blue Orbit - Phase 2 Initiated . (More to come..)

[Fritz_the_germ]

Yoh Alien-Grey,
it looks like ASUS won´t release a new BIOS. The 0902 is about a year old
But I am happy to see that you managed to oc your memory over 1,200 MHz.
The clue was ai clock twister "light".
Are you waiting for the "Fermi" from Nvidia?

@all,
did anybody of you guys hear when NVidia will bring out the GT300 ?

[Alien Grey]

ASUS is trying to fool us.

Ai Clock Twister Light is exactly the same as Ai Clock Twister Moderate and DRAM Static Read Control Disabled. It's exactly the same bandwidth in EVEREST.
I also tried Ai Clock Twister Lighter. It's also exactly the same bandwidth.

After many more tests I came to this conclusion. The memory is stable at DDR 1253MHz and tRD 6.

The CPU is stable at 4.23GHz.

The board fails to keep it stable.

BIOS 0902 is good but not good enough. The problems with the memory are fixed but there's still a problem with the BIOS that makes the NB fail after some time.


FSB 500MHz and my Q9650 ... not with this board and/or BIOS if I want it really stable.