Live MB OC Report :: ASUS Rampage Extreme

[3oh6]

Well tonight I found it's definitely an issue with my RAM.

simply adjusting the memory ratio doesn't just adjust the memory frequency. you are potentially running on a different strap and each ratio works in a different manner. i wouldn't be 100% sure it is your memory. it is likely, but no guarantee that it still isn't the MCH setup properly.

What bothers me though are your voltages and temps. Why can I bench my qx9650 at 485x9.5 with 1.55v NB on the 790i and you need 1.67v for 430 x 10.

you can't compare voltages across two different chipsets...different ballgames, different sports even. keep in mind, he is also running 4x2GB which is going to require more work on the MCHs part. perhaps the reason for the high vMCH. with that said, yes, a lot of those voltages in his setup are rather high.

[RLM]

you can't compare voltages across two different chipsets...different ballgames, different sports even. keep in mind, he is also running 4x2GB which is going to require more work on the MCHs part. perhaps the reason for the high vMCH. with that said, yes, a lot of those voltages in his setup are rather high.

i take your chipset comment on board. The reason I'm sticking it back in the 790i, is that I wanted to see whether my cpu had degraded or not. I had similar results on the P5E3 Deluxe to the Rampage regarding required volts and stability. For instance 4.2 ghz at 1.42v would pass OCCT on the 790i. 4.2 on the Rampage requires 1.485 to get small FTT even going without bluescreen.

Sure thing 30h6, my comments are more questions really rather than statements. Yes I haven't really worked that much with 2 x 2GB so I can see your point. We know that 1.35v for VTT is the max safe for 45nm cpus, but still want to know about PLL?

RLM

[zlojack]

simply adjusting the memory ratio doesn't just adjust the memory frequency. you are potentially running on a different strap and each ratio works in a different manner. i wouldn't be 100% sure it is your memory. it is likely, but no guarantee that it still isn't the MCH setup properly.

Same strap, different divider.

All of this is on the 333 strap so far. I will do some more testing tonight when I get my new GPU installed.

[MnM]

We know that 1.35v for VTT is the max safe for 45nm cpus

and by VTT you mean the vFSB? or FSB Termination Voltage as in RE's bios?

[zlojack]

i take your chipset comment on board. The reason I'm sticking it back in the 790i, is that I wanted to see whether my cpu had degraded or not. I had similar results on the P5E3 Deluxe to the Rampage regarding required volts and stability. For instance 4.2 ghz at 1.42v would pass OCCT on the 790i. 4.2 on the Rampage requires 1.485 to get small FTT even going without bluescreen.

Sure thing 30h6, my comments are more questions really rather than statements. Yes I haven't really worked that much with 2 x 2GB so I can see your point. We know that 1.35v for VTT is the max safe for 45nm cpus, but still want to know about PLL?

RLM

Has this been proven definitively, or is it a theory?

I know BenchZowner was running some tests on this to give us more data.
http://www.xtremesystems.org/forums/...d.php?t=198614

[RLM]

Has this been proven definitively, or is it a theory?

Good point, maybe should have worded that differently. I was lead to believe there had been tests done on Anandtech, and they found or suspected atleast that using 1.4v+ on VTT or FSB termination ran a risk of killing 45nm chips.

I know personally, and yes taking the different chipsets into account, I have never needed more that 1.4v VTT. That's with my Qx running up to 4.9+ghz

Maybe I'm wrong on that, but I was under the impression this was common knowledge.

RLM

[RLM]

As for the Rampage overclocking issues and voltages I may well have been hasty in my suspicions. It was certainly worthwhile putting the chip back in my EVGA.

Firstly I under estimated the heat output from my 4870x2. Previously I had my QX and volt modded 9800 GX2 in the same loop. However I had two Radiators the PA 120.3 and a Swiftech MCR220.

This time I have the QX and 4870x2 in the same loop, but with just the PA 120.3. It seems that at load I'm about 5-8 degrees hotter with this setup. This is obviously going to have a bit of an impact on my overclocking. I'm also suspecting that my QX isn't in as good shape as it used to be.

Doing a bit more testing now but in XP

RLM

[zlojack]

Good point, maybe should have worded that differently. I was lead to believe there had been tests done on Anandtech, and they found or suspected atleast that using 1.4v+ on VTT or FSB termination ran a risk of killing 45nm chips.

I know personally, and yes taking the different chipsets into account, I have never needed more that 1.4v VTT. That's with my Qx running up to 4.9+ghz

Maybe I'm wrong on that, but I was under the impression this was common knowledge.

RLM

As far as I know, Anandtech based their conclusion on one chip. It's a good theory by them, but wasn't conclusively proven. Nonetheless, you're right in that if you don't need more than that to acheive your clocks, the point is moot.

Anyway, I'm going to try for some more high FSB tonight with the RAM downclocked and see what results I get.

[eternal_fantasy]

As far as I know, Anandtech based their conclusion on one chip. It's a good theory by them, but wasn't conclusively proven. Nonetheless, you're right in that if you don't need more than that to acheive your clocks, the point is moot.

Anyway, I'm going to try for some more high FSB tonight with the RAM downclocked and see what results I get.

Precisely. Anandtech based the whole article on a single CPU failure, and concluded it was absolute and conclusive prove that subjecting your CPU with over 1.4V VTT and IT WILL DIE.

Failure indeed.

[zlojack]

Precisely. Anandtech based the whole article on a single CPU failure, and concluded it was absolute and conclusive prove that subjecting your CPU with over 1.4V VTT and IT WILL DIE.

Failure indeed.

What's worse it that it's been taken as gospel and spread around to the point that many now accept it as a proven fact.

I'm not the most knowledgable or experienced overclocker around, but I do know enough not to accept one result as a hard and fast rule.

[RLM]

Precisely. Anandtech based the whole article on a single CPU failure, and concluded it was absolute and conclusive prove that subjecting your CPU with over 1.4V VTT and IT WILL DIE.

Failure indeed.

So 1.4vtt+ is okay then?

RLM

[zlojack]

So 1.4vtt+ is okay then?

RLM

I haven't seen a conclusive answer one way or the other.

[eternal_fantasy]

So 1.4vtt+ is okay then?

RLM

As with all overvoltage, the risks increases as you add more, but there are no definite and absolute "safe" voltages in any category of overclocking. I've put through 1.5V VTT through my QX9650 on the Maximus Extreme for over 6 months before I realised it wasn't needed, and it is still as meh as the first day I received it.

[3oh6]

I had similar results on the P5E3 Deluxe to the Rampage regarding required volts and stability. For instance 4.2 ghz at 1.42v would pass OCCT on the 790i. 4.2 on the Rampage requires 1.485 to get small FTT even going without bluescreen.

Sure thing 30h6, my comments are more questions really rather than statements. Yes I haven't really worked that much with 2 x 2GB so I can see your point.

you may not be too off on your assessment of the 790i...i too have found the 790i to require less voltage for equal overclocks to other boards. it might be the 790i reporting lower volts but even when measured with a DMM, it seemed to clock higher at less voltage than other boards. my experiences are all with dual cores but it might be the same type of thing you are seeing with quads. nothing crazy but enough to notice type thing.

as for 2gb sticks, i haven't spent a lot of time with them on the RE but he is running 4x2GB which has to be the reason for the high vMCH needed. again, just an assumption based on known information from others that have. i have no need to ever run 4x1Gb let alone 4x2GB so i won't be spending any time testing any time soon to find out definitively. in my 2x2GB testing though, higher vMCH definitely helped clocks so i can only imagine 4x2GB would require even more at those types of clocks.

[Mezzro]

you may not be too off on your assessment of the 790i...i too have found the 790i to require less voltage for equal overclocks to other boards.

I agree. What I don't understand is why even with less voltage, Nvidia chipsets run to much hotter (a good 10C in my experience) than their Intel counterparts.

[3oh6]

I agree. What I don't understand is why even with less voltage, Nvidia chipsets run to much hotter (a good 10C in my experience) than their Intel counterparts.

that is like comparing Intel to AMD CPUs temps at the same clocks or diesel fuel mileage to gas fuel mileage. they are completely different. in no way are they similar. they both accomplish the same goal but get there in completely different ways. you just can't compare the two, simple as that.

not to mention the fact that the NVIDIA chipset doesn't have an internal thermal probe, so temps on that chipset are arbitrary at best.

[Expat GriZ]

Well, here is where I am at now. Have been reading the thread and will try some of the other settings tonight.


Tweaker Menu (Bios version 0601)
Ai Overclock Tuner: Manual
OC From CPU Level Up: Auto
OC From Memory Level Up: Auto
FSB Frequency: 467
CPU Ratio Setting: 9

CPU Clock Skew: Normal
NB Clock Skew Normal
FSB Strap to North Bridge: 266 or 400, both work
PCIE Frequency: 100
DRAM Frequency: 1869
DRAM Command Rate: 2N
DRAM Timing Control: manual
1st Information: 8 8 8 21 5 82 8 6

2nd Information: Auto

3RD Information: Auto

DRAM Static Read Control: Enabled
DRAM Dynamic Write Control: Enabled

>>DRAM Skew Control
>>DRAM CMD Skew on Channel A: Normal
>>DRAM CLK Skew on DIMM A1: Normal
>>DRAM CLK Skew on DIMM A2: Advance 250 ps
>>DRAM CTL Skew on DIMM A1: Normal
>>DRAM CTL Skew on DIMM A2: Normal
>>DRAM CMD Skew on Channel B: Same as Ch A
>>DRAM CLK Skew on DIMM B1:
>>DRAM CLK Skew on DIMM B2:
>>DRAM CTL Skew on DIMM B1:
>>DRAM CTL Skew on DIMM B2:
Ai Clock Twister: Moderate
Ai Transaction Booster: Manual Perf Level 7, 6 possible but not stable yet
EPU II Phase Control: Auto
CPU Voltage:1.43750
Load-Line Calibration: Enabled
CPU PPL Voltage: 1.56
FSB Termination Voltage: 1.4
CPU GTLVerf (0): Auto
CPU GTLVerf (1): Auto
CPU GTLVerf (2): Auto
CPU GTLVerf (3): Auto
NB GTLVerf: Auto
North Bridge Voltage: 1.52422
DRAM Voltage: 2.05422
NB DDRVref: +12.5
DDR3 ChannelA Vref: +50
DDR3 ChannelB Vref: +50
South Bridge 1.5 Voltage: 1.53756
South Bridge 1.05 Voltage: 1.10014
CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled

[Expat GriZ]

I couldn't resist and had to have a go. Everything is loose, will try and tighten up the timings but made it!!!

[zlojack]

Those sticks are looking pretty sweet!

[RLM]

Well, here is where I am at now. Have been reading the thread and will try some of the other settings tonight.

You've been busy Expat Griz. Well done that man

i haven't spent a lot of time with them on the RE but he is running 4x2GB

30h6, I completely missed that. No easy task, from what I gather.

As with all overvoltage, the risks increases as you add more, but there are no definite and absolute "safe" voltages in any category of overclocking. I've put through 1.5V VTT through my QX9650 on the Maximus Extreme for over 6 months before I realised it wasn't needed, and it is still as meh as the first day I received it.

Well that holds a bit of weight and certainly backs up your skepticism.

RLM

[AndreaEG6]

Hi guys, I'm having BIG issues while OCing my Q9650 on this board, this is getting me insane because I can't even reach 450x9 to boot into Windows, and this is really strange, imho. I read in the Q9650 that it might be because of the low VTT settings, and that Q9650 owners shouldn't be scared of raising it. Do you have any ideas about my issue? I'm afraid I bought a bad Q9650 batch, and I'm thinking of asking for a new one, but I'm also tempted to buy a E8600, which, from what I've seen, suits better the Rampage Extreme. The ONLY way I can OC is setting FSB on 400x9 and ram @ 1600 MHz, it won't work with any other configuration. I've also changed my RAM to OCZ Platinum DDR3, but results doesn't seem to change.

My usual settings are:

Tweaker Menu
Ai Overclock Tuner: Manual
OC From CPU Level Up: Auto
OC From Memory Level Up: Auto
FSB Frequency: 450
CPU Ratio Setting: 9

CPU Clock Skew: Normal
NB Clock Skew Normal
FSB Strap to North Bridge: 266/333/400 they all don't work
PCIE Frequency: 100
DRAM Frequency: 1803
DRAM Command Rate: Auto
DRAM Timing Control: manual
1st Information:7-7-7-20
2nd Information: Auto

3RD Information: Auto

DRAM Static Read Control: Disabled
DRAM Dynamic Write Control: Disabled

>>DRAM Skew Control
>>DRAM CMD Skew on Channel A: Auto
>>DRAM CLK Skew on DIMM A1: Auto
>>DRAM CLK Skew on DIMM A2: Auto
>>DRAM CTL Skew on DIMM A1: Auto
>>DRAM CTL Skew on DIMM A2: Auto
>>DRAM CMD Skew on Channel B: Same as Ch A
>>DRAM CLK Skew on DIMM B1:
>>DRAM CLK Skew on DIMM B2:
>>DRAM CTL Skew on DIMM B1:
>>DRAM CTL Skew on DIMM B2:
Ai Clock Twister: Moderate
Ai Transaction Booster: Manual Perf Level 7
EPU II Phase Control: Auto
CPU Voltage:1.40 max, I've tried almost all settings up to 1.40
Load-Line Calibration: Enabled
CPU PPL Voltage: 1.55
FSB Termination Voltage: 1.32
CPU GTLVerf (0): Auto
CPU GTLVerf (1): Auto
CPU GTLVerf (2): Auto
CPU GTLVerf (3): Auto
NB GTLVerf: Auto
North Bridge Voltage: 1.4
DRAM Voltage: 2.0 for Corsairs, 1.90 for OCZ
NB DDRVref: Auto
DDR3 ChannelA Vref: Auto
DDR3 ChannelB Vref: Auto
South Bridge 1.5 Voltage: 1.5
South Bridge 1.05 Voltage: 1.06
CPU Spread Spectrum: Auto
PCIE Spread Spectrum: Auto

Here's my setup:
Case: Cooler Master Stacker | PSU: Hiper HPU-5B880 880W | CPU: Q9650 Cooled By Thermalright Ultra 120 Extreme | Motherboard: Asus Rampage Extreme | RAM: 4 GB DDR3 OCZ Platinum | VGA: Sapphire ATI HD 4870 X2 | HDD: Velociraptor 300 GB + 2x Seagate 7200.11 750 GB | Monitor: Samsung 2493HM | Keyboard: G15 Mouse: G9 OS: Vista Ultimate 64bit

[RLM]

Well it was very worthwhile going back to the 790i. Couldn't clock on it any where near my previous clocks. A number of thoughts maybe CPU degradation, inadequate cooling (temps were certainly up) and the fact the board's shot. Obviously hope it's not the first thought. In fact it seems the 790i has croaked it. Shutdown and couldn't boot again. Stuck on a 25 error. Tried a q6600 in there and still no joy.

The positive is that the RE is looking more like a decent choice Got it all setup again and feeling a lot happier(for now anyway). First boot at 410x10. Vcore set to 1.55 I think, crazy droop all the way down to 1.41v. Running small FTTs no worries. I've clocked the graphics card back to 500/500 for the timebeing and i think in part this is helping with temps in the loop.

I'm going to pick up another 4870x2. I'll take the present one out of the loop and run the new one on air for a bit. Would like to see how the cpu scales without the Graphics card in the loop. I'm obviously hoping that previous barriers were temp related. Eventually I'll get onto the vapochill. Well sometime. A thumbs up.

RLM

Edit: Just had a thought why things might be running smoother. Forgot I'm now in a clean install of Xp not Vista 64. Additionally My Vista 64 really needs a re-install

[RLM]

AndreaEG6,

What bios are you running?

Have you tried to loosen up the ram timings and performance levels, to try and focus on whether it's memory or cpu holding you back. You know maybe 9-9-9-30 2N TRFC 100 and perf on 9. maybe even try a divider.

Another thought maybe try without load-line calibration enabled. you have to account for this though, so for 1.4v load you may need to enter 1.54v into bios.

i'm sure others will have suggestions. just a few thoughts though.

RLM

[Bullshooter]

This Board have Problems with 2x2GB RAM.
And when you are the 1:2 dividier for the RAM have it was very strong for this MB.
With the new Bios 0601 i can run with my E8600 and 2x2 GB CSX Diablos FSB@540x8 and the RAM works @1800 (3:5 400 Strap) fine with 7-7-7-21 2T.
My NB works at 1,46V.

[cheapseats]

@ AndreaEG6

not sure which OCZ Plats you have but my OCZ 2x1GB 1800's need a fair bit more VDIMM (1.987V) for 900Mhz / CL7. It's taken me a while to get stable with CL7 - finding the 'right' DRAM CTL and NB skews was the key. No stability for me without them.

Just finished Priming for 900 / 7-7-7-20 / PL6, so will post up screen and settings next. I guess you'll need more VNB than I did with 2x1GB and E8600 (though mine seems highish @ 1.444V). I doubt my skews will work for you either (but they might).

Agree with RLM that getting the mems stable first, looser or lower clocked, will be best. DRAM Skews will help at some point but I've found you need to be 'near' stable before you'll see any benefit.

Generally OCZs on this board seem quite hard work but with enough time, should be OK

HTH