Live MB OC Report :: ASUS Rampage Extreme

[Ghostleader]

If you are reduced weight you should consider these from Enzotech for your SB.





I´m using CNB-S1L for my SB and the temp is ~40C

[rob2k]

Hi Aln688, looks like a great CPU & NB combo

Have to agree with Ghostleader that the Thermalright HR-05 SLI/IFX is a bit OTT if it's not too late and you have it already. I have the CNB-SL1 which maxes out at 53 degrees, the SLF-1 should arrive tomorrow which should reduce that to ~40.

[aln688]

So you both think the Enzotech CNB-S1L is a better choice? Actually I still have the Enzotech CNB-S1L I just wasn't planning on using it. I was just put off putting the heatsink on at an angle, but I suspect it covers the entire SB chip.

I was going through the numbers again of using the Noctua U9B -vs- the Noctua U12P, after adding up the weight of the fans as well, I'm only saving just over 100g using the U9B. If I take on your suggestion of the CNB-S1L I'm sure I'll save a bit more as well.

The calculations look like this:

Using Noctua U12P:
--> SB heatsink, Thermalright HR-05 SLI/IFX (125g)
--> NB heatsink, Thermalright HR-05 SLI/IFX (125g)
--> NB heatsink fan, Sanyo Denki 80x15mm (68g)
--> CPU heatsink + fan (760g)
--> Total : 1078g

Using Noctua U9B:
--> SB heatsink, Thermalright HR-05 SLI/IFX (125g)
--> NB heatsink, Thermalright HR-05 SLI/IFX (125g)
--> NB heatsink fan, Sanyo Denki 80x25mm (110g)
--> CPU heatsink (460g)
--> CPU fan, Sanyo Denki 92x25mm (150g)
--> Total: 970g

If I use the Enzotech CNB-S1L, it is only 50g, saving 75g in both cases. I'm weighing up still using the U12P, if the difference was a few hundred g's then the U9B would be a definite choice but it's only just over 100g difference.

Don't know if you noticed, but Enzotech is bringing out a new heatsink, the MOS-C10, three of these would fit the chokes near the NB perfectly:

http://www.enzotechnology.com/mos_c10.htm

[aln688]

I´m using CNB-S1L for my SB and the temp is ~40C

Thanks for the information, to be honest I wasn't sure how much heat the SB pumps out. I suspected it was a low heat component but read a few forums where people experienced heatsinks on their SB that they couldn't touch with a finger due to it being too hot.

[aln688]

Is the Sanyo Denki 80x15mm fan for the Northbridge sufficient? It's rated at 20 CFM @ 2000 RPM, only 21 dbA. I suspect that *any* fan blowing on the Northbridge is sufficient.

[aln688]

This was in the early days of my RE ownership. I was struggling to get more than 420FSB and getting very frustrated at my OCZ memory not running at its rated speed so I went for the TRUE120 for a no compromise CPU cooler. From memory the Zalman ran 10 degrees hotter under load. The TRUE120 is HUGE and I had to remove a side fan from my case at the time which was a C/M Stacker 832 to fit it inside. Of course there were other issues, early bios, the huge number of bios options and N/B temps that also hampered me. When I moved to water I found the board to have warped 3-4mm when sat in its free state on the table although when fitted in my new case it 'bolted up' flat and there are no long term ill effects. Of course I never measured the mobo flatness when it was new so it may not have been perfect then and may not have warped at all - how is your new board ?.

Here is a useful comparison review of the Noctua U12 Vs the Noctua U9 :-
http://www.silentpcreview.com/article295-page1.html

and a more general H/S summary table :-
http://www.silentpcreview.com/article30-page1.html

HTH

My board minus Asus's heatpipe assembly is very slightly bowed, with both sides pointing up and the middle down, but it's very slight. So far my board hasn't had a CPU heatsink installed on it, so it's more or less bare as from the box. Thanks for the links, very useful to review.

One thing I did read about on the reviews of Newegg.com regarding the Zalman 9500 LED, is that it *used* to fit the Rampage Extreme, then Zalman changed the mounting mechanism to a two piece arrangement, which now means the bracket hits chokes/capacitors - I experienced a similar problem with my old Noctua C12P. I'm guessing regarding the problem, I tried going to Zalman's web-site for motherboard compatibility, useless, it was last updated August 2005!

[cstkl1]

all those with quaddies stable on rampage extreme..
can u guys post ure IBT/LINpack for 2x1gb around 15 runs..
will just take 15 minutes..

and ure core gtl..
curious.

[aln688]

Hi Aln688, looks like a great CPU & NB combo

Have to agree with Ghostleader that the Thermalright HR-05 SLI/IFX is a bit OTT if it's not too late and you have it already. I have the CNB-SL1 which maxes out at 53 degrees, the SLF-1 should arrive tomorrow which should reduce that to ~40.

Well the combo might be good, but after studying my new HR-05 SLI/IFX, I've came to the conclusion the quality seems good, but the great idea got lost somehow in putting the package together. The fan clip the way its designed means the fan sits on the back of the heatsink, I don't know if this is a good idea as it means the heatsink further sticks out and coming near to hitting the video card, it would be better to have been positioned inside the heatsink, hovering above the chip it's used on.

The other very confusing item is the mounting method. Possible mounting methods including the hook system (not applicable here), white plastic pushpins or screws coming up from behind the board, just which system am I supposed to use? The paper instructions mention using the spring loaded plate with the pushpins, yet if I look at Thermalright's web-site for installation, they don't even mention the white pushpins and list using the screws coming up from behind the board.

Looking online here at some reviews, it seems the pushpins aren't used at all, only the screws method where the screws come up from under the board, the clip goes over the heatsink and the thumb screws screw down the heatsink, one item I wonder about is, how much are you supposed to tighten the screws? I don't want to over tighten it, do I tighten it until the heatsink won't move?

One other item, the Enzotech CNB-S1L, while a nice heatsink, does not properly fit the Rampage Extreme's SB chip. Sure, I could angle the black push pins inwards then they'll push through, but I refuse to angle a pushpins just to get it to fit, there's the potential for the pushpins to come undone. Not much choice for SB heatsink in my case. This link contains a picture that proves the point, 3rd picture down:

http://forums.techpowerup.com/showthread.php?t=52217

[cstkl1]

another interesting discovery
the gtl for NB adding it only makes things worst
e8600 and team xtreem pc3-15000 2x1gb and 4gb

on 4ghz 500x8 1600 CL7 1T tRD 8 dram read/write disable
cpu vcore 1.1875v
vtt/sb/nb/pll/dram - 1.12/1.5/ 1.249/1.5/1.8
cpu gtl 0 +40
nb gtl +30
dram A/B vtt +25.

ibt test pass 30 runs so around 2 hours
if add nb dram vtt +12.5 hangs on 4th

up multi to 8.5 and vcore to 1.2875
hang on start or ibt.
cpu clock skew to delay 200... pass 30...

so higher vcore/speeds require cpu clocks skew adjustment...
cant test 4.5ghz on ibt.. has it exceeds my cooling ability cause 4.25ghz on load is already 66C.
but testest with orthos priority 10 with blend then custom and max fft at 8 to stress ram/nb/cpu.
shows similar results...

but so far IBT has proven that nb gtl doesnt need any increment and for dual cores only gtl 0 needs a increment of +40. to make the gtl to 0.67.
the gtl results of this board is exactly same as maximus 2 formula
so hence y i am asking all those posting results to show ibt...results...
the cause of the hanging issue was the dram vtt channel A/B.. can be proven with extended runs in memtest86
wrong pll voltages will hinder ure results in memtest86 and showing long runs of stable system but in IBT failure is imminent.


also another interesting result
memtest86 test 5 shows that certain strap/trd prefers a certain clock twister.
sometimes higher and sometimes lower.

the current difficulty is getting 2x1gb 2000mhz tRD 8 gtl's correct.. so far IBt can run 50 runs without failure
but all the residuals are varying...
cheapseat.. i think this was ure issue when u tried to go more than 2 hours on prime/orthos blend..

micron d9jnl's should be able to do 2000 8-8-8-24 110 2x1gb clock twister strong strap 400 vnb 1.6 dram vtt a/b +25 at 1.9v only...
so cellshock 1866 / team xtreem 1866 owners .. heheh nice stuff..

hope somebody can help with this and show posting of ibt results...

for 4x1gb
so far for booting at 2000mhz
setting skew adjustment as cheapseat did but on all channels shows booting but will fail memtest 3/5

[rob2k]

One other item, the Enzotech CNB-S1L, while a nice heatsink, does not properly fit the Rampage Extreme's SB chip. Sure, I could angle the black push pins inwards then they'll push through, but I refuse to angle a pushpins just to get it to fit, there's the potential for the pushpins to come undone.

The S1L has 2 fittings 54.6mm and 59mm, not sure exactly what the RE hole spacings are but all you need to do is to file the overlapping holes into a single slot and position it centrally over the S/B chip.

[Ghostleader]

another interesting discovery
the gtl for NB adding it only makes things worst
e8600 and team xtreem pc3-15000 2x1gb

on 4ghz 500x8 1600 CL7 1T tRD 8 dram read/write disable
cpu vcore 1.1875v
vtt/sb/nb/pll/dram - 1.12/1.5/ 1.249/1.5/1.8
cpu gtl 0 +40
nb gtl +30
dram A/B vtt +25.

ibt test pass 30 runs so around 2 hours
if add nb dram vtt +12.5 hangs on 4th

up multi to 8.5 and vcore to 1.2875
hang on start or ibt.
cpu clock skew to delay 200... pass 30...

so higher vcore/speeds require cpu clocks skew adjustment...
cant test 4.5ghz on ibt.. has it exceeds my cooling ability cause 4.25ghz on load is already 66C.
but testest with orthos priority 10 with blend then custom and max fft at 8 to stress ram/nb/cpu.
shows similar results...

but so far IBT has proven that nb gtl doesnt need any increment and for dual cores only gtl 0 needs a increment of +40. to make the gtl to 0.67.
the gtl results of this board is exactly same as maximus 2 formula
so hence y i am asking all those posting results to show ibt...results...
the cause of the hanging issue was the dram vtt channel A/B.. can be proven with extended runs in memtest86
wrong pll voltages will hinder ure results in memtest86 and showing long runs of stable system but in IBT failure is imminent.


also another interesting result
memtest86 test 5 shows that certain strap/trd prefers a certain clock twister.
sometimes higher and sometimes lower.

the current difficulty is getting 2x1gb 2000mhz tRD 8 gtl's correct.. so far IBt can run 50 runs without failure
but all the residuals are varying...
cheapseat.. i think this was ure issue when u tried to go more than 2 hours on prime/orthos blend..

micron d9jnl's should be able to do 2000 8-8-8-24 110 2x1gb clock twister strong strap 400 vnb 1.6 dram vtt a/b +25 at 1.9v only...
so cellshock 1866 / team xtreem 1866 owners .. heheh nice stuff..

hope somebody can help with this and show posting of ibt results...

for 4x1gb
so far for booting at 2000mhz
setting skew adjustment as cheapseat did but on all channels shows booting but will fail memtest 3/5

Sorry for asking but what is IBT?

And a other thing, I thought that 45nm should use - on GTL and not +?

How about the NB GTL, NB GTL Vref, DD3 Ch A B Vref + or - ?

Some using + and some -, quite confusing.

[cstkl1]

Sorry for asking but what is IBT?

And a other thing, I thought that 45nm should use - on GTL and not +?

How about the NB GTL, NB GTL Vref, DD3 Ch A B Vref + or - ?

Some using + and some -, quite confusing.

well for ure e8600

try core 0 +40
nb gtl = auto..
if cannot enter or hang try upping pll voltage.

dram vtt +25

not minus..
u can see from IBT - Intel Burn Test.. .that ure gtl is stabalizing.

since ure using a e8600 which is the proc that i have bought/given/tested the most amount in my entire 27 years of living in this planet.

most e8600 that i have tested has one thing in common
the stock vcore at 3.33ghz at whatever vid is the 4ghz vcore
example mine is set 1.1875 is the stock vcore..which runs 4ghz
and then try 1.12v vtt which is good up to 500 fsb
nb 1.25
strap 333
1600 CL6 at 1.8v

and try ure gtl's using linpack
for 2x1gb i advice 50 runs
for 4gb kits advice 15 runs
both will be around 1 hour.
make sure u have sufficient cooling

e8600's also has one more thing in common
say 4ghz is stock at example mine at 1.1875
then the 4.25ghz is at 1.2875

and 4.5 ghz is around 1.4v for asus mobo cause they is a bigger droop after 1.35v.
but for ddr2 up to 4.25ghz not much setting to change other than vcore
but ddr3.. ure rams will start acting up..
so need to change cpu skews..

and ghost a lot of ppl are quoting stable etc
its given..
i can run ibt for 2000mhz CL8 for 50 runs that shows vcore stable and nb /cpu gtl and even pass memtest etc.
but all the residuals are different. and theres a few common pairs popping up..

this is y i think some ppl are failing to run the full orthos etc...
and it will fail my test 3 that i posted using orthos a few pages back..

also those having a maximus 2 formula board and some ddr2 rams .. u can confirm this findings on cpu gtl and vcore and vtt
its exactly the same.

also the use of pll for dual core..
advice not needed until u have weird results .. example start can run ibt for full run
restart it cannot and hang etc...
it always messes up memtest86 results for vdimm for test 5

tested without dram vtt.. and pll.. memtest86 hang in 45 minutes
with pll it passes to 1 hour 30 minues but fails suddenly in test 5 which is mainly nb/dram voltage and tRFC settings.
reduce pll and increase dram vtt to +25 for both A/B.. passed 4 hours.. no hitch and i stopped it.

lastly to answer that question
for p45
it favours gtl 64 for nb and cpu up to a certain vcore
but i see on high vcore and high speeds gtl 67 is still the best for core and nb.

[cstkl1]

oh yeah and another thing i want to clear about e8600's
i had a vid many vids and many different stock vcores
one thing i noted
whatever the voltages were
the stock vcore = 4ghz temp for all of them were the same
and the +0.1v for 4.25ghz temp for all of them were the same
when i ran IBT with the same cooler
the difference was the tweaking of the gtl/skews and sometimes rams... which depended on that vcore and speed.
i guess thats y a lot of those 600+fsb clocks seems to do well on the high vid rather than the low vid.

oh yeah and disable legacy support..

oh yeah remember the nb+30 thing is still inconclusive.

but i think i am almost got it for this board on what voltages are insufficient and what test to confirm it..
need to run more test..and make sure its not ram chipset dependent.

[vega22]

all those with quaddies stable on rampage extreme..
can u guys post ure IBT/LINpack for 2x1gb around 15 runs..
will just take 15 minutes..

and ure core gtl..
curious.

It's a nightmare. My Q9550/2x2GB G.Skill 1600 is able to make 200+ IBT @ 425 FSB/PL6/AI C.T. Stronger, with GTLref all 4 in Auto and NBGtlref=+20 (VERY IMPORTANT). But if the system hangs and reboot in the #201 IBT (i.e. because of some instability occurs in pcie -yes!-), then the internal management of the GTLrefs become crazy and then the system is unable to make #1 only IBT with the same settings.
Then, trying to restore the internal stability of the GTLref again is the problem more heavier, boring and depressing that you may find with this motherboard.
Time to go for i7, i think....

[cstkl1]

It's a nightmare. My Q9550/2x2GB G.Skill 1600 is able to make 200+ IBT @ 425 FSB/PL6/AI C.T. Stronger, with GTLref all 4 in Auto and NBGtlref=+20 (VERY IMPORTANT). But if the system hangs and reboot in the #201 IBT (i.e. because of some instability occurs in pcie -yes!-), then the internal management of the GTLrefs become crazy and then the system is unable to make #1 only IBT with the same settings.
Then, trying to restore the internal stability of the GTLref again is the problem more heavier, boring and depressing that you may find with this motherboard.
Time to go for i7, i think....

true..
but does i7 support ibt.. and if does must be freaking fast
hmm result on 4.25ghz
the nb gtl .. also is a issue
it can stabilize fsb
as pll voltage can do the same...

but which one is better.
so far seems to be pll..
so guess nb voltage for now can leave on auto..

[vega22]

true..
but does i7 support ibt.. and if does must be freaking fast
hmm result on 4.25ghz
the nb gtl .. also is a issue
it can stabilize fsb
as pll voltage can do the same...

but which one is better.
so far seems to be pll..
so guess nb voltage for now can leave on auto..

Mmmm...Ok...i´ll try with the PLL. So far, my tricks for FSB stability are NB GTL and PCI-e @ 105.

EDIT: IBT is the best for vCore, PLL and vNB. For vDram and VTT OCCT in RAM mode and Prime95 64bit run in custom mode 2048-4096 reveals failures that the IBT takes a lot more to discover. At least in my experience with my system

[Ghostleader]

well for ure e8600

try core 0 +40
nb gtl = auto..
if cannot enter or hang try upping pll voltage.

dram vtt +25

not minus..
u can see from IBT - Intel Burn Test.. .that ure gtl is stabalizing.

since ure using a e8600 which is the proc that i have bought/given/tested the most amount in my entire 27 years of living in this planet.

most e8600 that i have tested has one thing in common
the stock vcore at 3.33ghz at whatever vid is the 4ghz vcore
example mine is set 1.1875 is the stock vcore..which runs 4ghz
and then try 1.12v vtt which is good up to 500 fsb
nb 1.25
strap 333
1600 CL6 at 1.8v

and try ure gtl's using linpack
for 2x1gb i advice 50 runs
for 4gb kits advice 15 runs
both will be around 1 hour.
make sure u have sufficient cooling

e8600's also has one more thing in common
say 4ghz is stock at example mine at 1.1875
then the 4.25ghz is at 1.2875

and 4.5 ghz is around 1.4v for asus mobo cause they is a bigger droop after 1.35v.
but for ddr2 up to 4.25ghz not much setting to change other than vcore
but ddr3.. ure rams will start acting up..
so need to change cpu skews..

and ghost a lot of ppl are quoting stable etc
its given..
i can run ibt for 2000mhz CL8 for 50 runs that shows vcore stable and nb /cpu gtl and even pass memtest etc.
but all the residuals are different. and theres a few common pairs popping up..

this is y i think some ppl are failing to run the full orthos etc...
and it will fail my test 3 that i posted using orthos a few pages back..

also those having a maximus 2 formula board and some ddr2 rams .. u can confirm this findings on cpu gtl and vcore and vtt
its exactly the same.

also the use of pll for dual core..
advice not needed until u have weird results .. example start can run ibt for full run
restart it cannot and hang etc...
it always messes up memtest86 results for vdimm for test 5

tested without dram vtt.. and pll.. memtest86 hang in 45 minutes
with pll it passes to 1 hour 30 minues but fails suddenly in test 5 which is mainly nb/dram voltage and tRFC settings.
reduce pll and increase dram vtt to +25 for both A/B.. passed 4 hours.. no hitch and i stopped it.

lastly to answer that question
for p45
it favours gtl 64 for nb and cpu up to a certain vcore
but i see on high vcore and high speeds gtl 67 is still the best for core and nb.

Thanks

I gonna do a lot of test to see if what you´re suggestion is right, no doubt that it isn´t.

This board has a lot more tweaks then I have ever had before so I think it´s gonna take some time to figure it out.

What about run LinX, isn´t that almost the same as Linpack or IBT?

For the moment I focus at 450 9,5 4275 Strap 333 1800 777 21 1N tRD6 Strong tRFC88 to get it stable with so low volt set as possible and trying to get 475 9 4275 strap 333 1900 777 21 1N tRD6 Strong tRFC88 stable.

[Ozzfest05]

So im going to keep it here for a while decent Pi time for the speed im working on another project was thinking liquid cooling and keep it at 4.8 ish maybe ill do some air 4.5 with 2000mhz ram since this cpu needs literally nothing for vcore 4.4 at 1.23 is sick In my opinion but shooting for more phase change may be around the corner and ill have it running 5ghz or something for 24/7 and ram will settle in at 1800 or so since I know it wont do over 2000.

[cstkl1]

found one silly test for dram nb vtt

kekek ure guys going to laugh
run orthos priority 10..
keep doing start / stop / start stop..
***56K warning.. *** if ure gtl not right.. high chance of shutdown and wipeout ure bios including ure profile... ...

so change now to must add dram nb vtt to +12.5
also with the both dram vtt at +25mv
this going to sound even more hilarious
good way to spot insufficient vdimm for the paticular cpu speed.

[Solarfall]

getting ready for tomorrows 3mark benching session with 9800gx2, pretty good E8600 and of course rampage extreme.. the pic it self it pretty explanatory..just took the heatsink out and bolted 2 D-tek fuzion blocks to the card

[sup3rfly]

dont know if this is a know issue or wat... lets say after a superpi 32m i reboot and try to go in to bios, sometime when i hit delete, it will give me no display which i assume its hang, sometime it wont happen...i even get this with all settings to default.... wonder if anyone has the same problem or not....

[cheapseats]

rob2k and Ghostleader, cheers guys. It'll be interesting to see what people can do with the 200 strap.

Ghostleader, thanks for the 400FSB results - I didn't test for lower voltages on 200 V 333, but with the same max performance across 200/333 straps (though not possible with all FSB/CL), lower voltage looks like a nice bonus on 200

rob2K, the 200 strap will be slower (same as 400 strap) for FSBs around 475 and CL7. To equal 266 or 333/CL7/tRD6 performance, you need to run CL7/tRD5 on 200.

As far as tested, CL7/tRD5 on 200strap becomes non-allowable at some point between 401 and 450FSB. You may have found a no-POST already with the 200 strap, but 266 or 333 with CL7/tRD6 are both faster than 200 (and 400) with same nominal CL7/tRD6.

(EDIT) .......the current difficulty is getting 2x1gb 2000mhz tRD 8 gtl's correct.. so far IBt can run 50 runs without failure
but all the residuals are varying...
cheapseat.. i think this was ure issue when u tried to go more than 2 hours on prime/orthos blend..
.......(EDIT)

many interesting points cstkl1 - it's hard to keep up with your speed on this

On the unstable 500/1:2/CL8/tRD7 clocks, I was able to extend this from 2 to 8 hours Blend purely on NBDDRref and A/B DDRref adjustment. More stable but still not able to get through 1 full Round in Blend. Couldn't find a better GTL combination than I had for Large FFT stability though.

DDRrefs made a good difference, but only with negative A/B DDRref for me.

I was using Custom>2048-4096k to test for NB + A/B refs and found best results with NBDDRref +12.5 and both A/B DDRref -10.0

Differences of only +/- 2.5mV for A/B DDRrefs gave me +/- 2 hours for Blend, and I think other 'working' values would vary with other vDIMM etc. Seemed like a very narrow margin for success. Positive A/B refs gave me blue screen or no Windows.

With same other settings, +12.5/-10.0 combination was (repeatably) 7-8 Hour Orthos stable, and the next best (NB+12.5/AB-12.5) would only last 4 Hours Blend then fail on a Large FFT.

Pretty frustrating - and with low memory performance too.

For my settings, positive NB GTL fails quicker in Large FFT than using negative value. In 86+, I can 'remove' GTL related Test 5 errors by using NB -40 to -60mV range, but no Windows start.

Memtest HCI passed 2000% for same 500/1:2 clocks with NB and A/B DDRrefs on AUTO and lower vNB (1.56V set), so Orthos (over 2048k FFT) was more effective to show these instabilities.

Similar to vega22, I experienced problems with PCI-e and/or graphics card at certain FSB. At 500FSB, I find more predictable stability with 103 pcie bus and card overclocked or underclocked (stock is 580/1500/900). Time for a new card anyway ?

Have a cheap new 2x1 kit today but should still be able to test for some new 500/1:2/CL8 results. Looking for decent CL7 clocks first though.

[aln688]

The S1L has 2 fittings 54.6mm and 59mm, not sure exactly what the RE hole spacings are but all you need to do is to file the overlapping holes into a single slot and position it centrally over the S/B chip.

You need to file between both holes to create a larger hole? I was going to say because as is the peg moved to the inner positions will only go in at an angle, the pushpin won't go into the board vertically straight. Do you have a picture of your CNB-S1L as filed? I'd just be interested to see the result. Thanks again.

[cstkl1]

dont know if this is a know issue or wat... lets say after a superpi 32m i reboot and try to go in to bios, sometime when i hit delete, it will give me no display which i assume its hang, sometime it wont happen...i even get this with all settings to default.... wonder if anyone has the same problem or not....

yo bro...

its a gtl issue afaik.

a lot of stability on this board in the end points to gtl being the cause.
using a few orthos custom and silly tricks and ibt...

so now trying to understand whether skews can fix this or not.
setting the cpu delay to around 200 shows there is other ways to fix this.
but still inconculsive

weird other ddr3 boards was not at fussy but other ddr3 boards cant do 10500 on everest read for 400fsb at 1.25vnb.

but for 4x1gb i think will just give up

this board at 500fsb 1600 CL7 is already damn fast at only clock skew moderate vnb 1.25v is just a bit slower than 2000mhz CL8 trd 7 clock skew
strong vnb 1.6. still cant understand this part.. cause in normally u expect the difference to be big .

[spoof]

I want to ask something here to. I'm currently using on Rampage Formula (flashed P5E X48) vNB 1.51v and 0.63 NB GTLREF. I know, that by the CPU the VTTxGTLREF must be less than 1.0 (so, the Vref must be less than 1.0). So, what is it about the NB Vref. I'm asking, because my stability depends on NB Vref. So, even for benching i have to use vNBxNBGTLREF>1.0 Is there any problem NB Vref to be more than 1.0?