***Asus Rampage III Extreme Owners Thread***

[Johnmark]

I'm a bit late to the party, but I joined the ranks of you R3E owners today. It replaced my R2E which has been going strong for over a year.

Installed the EK-FC R3E and leak tested, looks good so far. The block is quite a restriction it seems (either that or its dumping a bit of heat into the loop) as my idle temps have increased from what they were with just the CPU and VGA in the loop in the previous setup. Load temps aren't too different though.

I'm running 0901 BIOS as that's what it shipped with, will have to read up on what's going on with them at this stage. Also have some things to discover and play around with like PMW freq, currently set it to 500KHz, but even at that level its allowing my CPU to run stable at lower vcore than the R2E. First impressions are I like it.

Some pics of getting it ready:

Prepping the board:




This kind of shocked me. There appeared to be grass and debris on the stock thermal pad! It's also a hack job in terms of pad cutting precision. Nice QC asus.



Block on the board



Now I'm ready to do the loop

Mosfets make lots of heat! You now have that heat load as well as your NB + SB heat in your loop!

It is perfectly normal. If you increase the PWM Freq from 500 to 1000, you will notice even more heat.

Do you have any clearance issue with the EK-FB-RE3 and your first PCIE x16 slot lock? Mine is very tight.

Yes there is a lot of extra heat load, but it is also true the block is very restrictive. Just the nature of this large W.B.
I had a NB block only and after installing the full coverage EK block I could see the water was now very slow when entering the res.
But hey it works great and my over all temps are great.

No clearance issues with first slot what so ever.

[CryptiK]

Mosfets make lots of heat! You now have that heat load as well as your NB + SB heat in your loop!

It is perfectly normal. If you increase the PWM Freq from 500 to 1000, you will notice even more heat.

Do you have any clearance issue with the EK-FB-RE3 and your first PCIE x16 slot lock? Mine is very tight.

Yeah the chokes at the CPU socket run hot, can't leave you finger on them for more than a second or two. Would like someone with an IR temp gun to get some temps. I wasn't planning to run 1000KHz PWM freq, I saw what that did with the Classifieds in terms of temps, 500KHz seems fine and is giving better results than the full analogue PWM on the R2E.

Idle temps have gone up maybe 2 degrees, load temps have risen 2-3*C. I run a MCP655 and have turned it up from 3 to 4 to compensate for the reduced flow. I also had to re-orient my CPU block vertically which is not optimal, and block order is Pump->Rad->Mobo->CPU->VGA->Res so may revisit tubing path etc to improve it (people say order has little effect though).

No clearance issue with the 1st PCI-E slot lock, it's damn close but doesn't touch it.


4.3GHz @ 1.230v | 26*C Ambient | Quad rad + fans @ 1000rpm

[CryptiK]

This board has some weird quirks, not sure if its just mine........doesnt like 3710MHz uncore and anything less than 1.35Vtt BIOS (~1.296v real - I only needed 1.257v real for that uncore on the R2E). Less than that is stable in all tests, but wont cold boot, just hangs on DRAM. This is not ram timing/voltage related. Same thing happens with 2 different kits. It also just hates having uncore at 2 x ram, very unstable, if I add a tiny bit more voltage than needed to stabilize 3560MHz uncore I can easy do 3710MHz probably more (yet to try). Also needs more voltage to clock the ram than my R2E did (DDR3-1780MHz 7-7-7-21 1T took 1.495v on my R2E, need 1.56v on the R3E), so that coupled with the Vtt I'm using higher voltages all round except CPU.

[Raja@ASUS]

This board has some weird quirks, not sure if its just mine........doesnt like 3710MHz uncore and anything less than 1.35Vtt BIOS (~1.296v real - I only needed 1.257v real for that uncore on the R2E). Less than that is stable in all tests, but wont cold boot, just hangs on DRAM. This is not ram timing/voltage related. Same thing happens with 2 different kits. It also just hates having uncore at 2 x ram, very unstable, if I add a tiny bit more voltage than needed to stabilize 3560MHz uncore I can easy do 3710MHz probably more (yet to try). Also needs more voltage to clock the ram than my R2E did (DDR3-1780MHz 7-7-7-21 1T took 1.495v on my R2E, need 1.56v on the R3E), so that coupled with the Vtt I'm using higher voltages all round except CPU.

Looking a my old review results here, the R3E loses about ~0.1V from the VTT last cap on the high side of the power plane to the under socket MLCCs (the gap widens with increasing current - so higher frequncy = more sag). 1.35VTT set is going to be very close to 1.25V~1.27V under full load. The measurement points on the board are a little off, and report the voltage a little higher than it actually is at the north side of the socket.

I never had an R2E here so don't know what the offset between the measurement point and the north side of the socket is. The only way to check is either to flip the board while it's running (tricky) or solder a small wire to one of the MLCCs and check the offset under full load.

Of course, this was don one an earlier BIOS, there is the chance that ASUS have shifted the voltage table to compensate for the delta somewhat.

I don't think this will cure all your quirks but might explain some of the issue.

-Raja

[Johnmark]

Yeah the chokes at the CPU socket run hot, can't leave you finger on them for more than a second or two. Would like someone with an IR temp gun to get some temps. I wasn't planning to run 1000KHz PWM freq, I saw what that did with the Classifieds in terms of temps, 500KHz seems fine and is giving better results than the full analogue PWM on the R2E.

Running an 980x at 4.1 Ghz and 1.3v results in cool chokes, that's with the machine crunching 24/7 and all cores fully loaded. Using an IR gun the temps are between 31c and 38c depending on the choke.

[CryptiK]

Looking a my old review results here, the R3E loses about ~0.1V from the VTT last cap on the high side of the power plane to the under socket MLCCs (the gap widens with increasing current - so higher frequncy = more sag). 1.35VTT set is going to be very close to 1.25V~1.27V under full load. The measurement points on the board are a little off, and report the voltage a little higher than it actually is at the north side of the socket.

I never had an R2E here so don't know what the offset between the measurement point and the north side of the socket is. The only way to check is either to flip the board while it's running (tricky) or solder a small wire to one of the MLCCs and check the offset under full load.

Of course, this was don one an earlier BIOS, there is the chance that ASUS have shifted the voltage table to compensate for the delta somewhat.

I don't think this will cure all your quirks but might explain some of the issue.

-Raja

Thankyou, that's interesting. I have at least preliminarily corrected the issue by increasing Vtt to 1.35625v in BIOS, which according to everest is 1.296v under load, and as per your findings this is higher than the real Vtt. I will have to measure at one of the MLCC's as you suggested to see what my board does. I did that with my R2E (I have a TJ07 so the socket area is cut out making access easy) and made a table of BIOS vs Software vs Real voltages.

Thing is I thought I'd sorted it last night by upping it to 1.350Vtt BIOS but this morning it hung on DRAM again on cold boot after PSU power off.

Did you investigate what enabling Vtt LLC did to the voltage? I don't usually run any LLC as I want to avoid transient overshoot conditions. However if it corrected the sag a little but still allowed some droop under load I'd be temped to use it.

Running an 980x at 4.1 Ghz and 1.3v results in cool chokes, that's with the machine crunching 24/7 and all cores fully loaded. Using an IR gun the temps are between 31c and 38c depending on the choke.

31-38*C for the chokes under full load? (Just to make sure we're talking about the same thing - the ferrite blocks with R68 written on them to the IO side of the CPU socket).

Mine are way over that. I'd say 45-50*C by guessing and touching. I'm running a W3570 at 4.3GHz @ 1.237v and have my PWM freq set to 500KHz, what do you have yours set to? You on water or air? Is there any airflow over that area in your case?

[CryptiK]

Ok no go, I can get it to boot into windows and be stable, but the second cold boot will get past the point during POST where it would DRAM hang with lower Vtt, but then it will either BSOD loading windows or just hang requiring reboot, whereupon it will say OC failed blah blah. I can reboot it multiple times after this and it will never make it into windows.

The alternative scenario after raising Vtt is it boots into windows but only with 4GB of ram. Rebooting and changing voltages will not rectify it, but a power off then on cycle will.

This is rather frustrating I must say, everything is set correctly but it goes from stable to can't load windows or detects 4GB of ram in 2 reboots. Never had any issues like this with my R2E.

I have tried reseating CPU and ram, and cleaned all the contacts with iso before putting them back in. I have tried every BIOS and tried using the second BIOS chip. I've tried setting manual ram timings (loose safe ones too), manaully setting QPI lower, different combinations of uncore, enabling Vtt LLC, changing ram ref voltages and vdimm, cpu and IOH skews etc. Nothing makes any difference so far. I'll keep at it, hopefully I find the trick.

[zsamz_]

cryptic raise the ioh to 1.26 and ioh/ich to 1.6

[CryptiK]

No difference zsamz, tried that before too. I am actually kinda stumped. Never had an issue this bad (actually with the R2E same thing happened until they released BIOS 1504 then it fixed the cold boot problem).

There's nothing wrong with my components or settings, the board should not be doing this.

[l0ud_sil3nc3]

No difference zsamz, tried that before too. I am actually kinda stumped. Never had an issue this bad (actually with the R2E same thing happened until they released BIOS 1504 then it fixed the cold boot problem).

There's nothing wrong with my components or settings, the board should not be doing this.

what sucks is that I don't think they have had a bios come out for a while

I know that when I had mine I could run 4.3 stable and with lower volts that my UD7 however, above that there was a complete lack of stability.

[zsamz_]

No difference zsamz, tried that before too. I am actually kinda stumped. Never had an issue this bad (actually with the R2E same thing happened until they released BIOS 1504 then it fixed the cold boot problem).

There's nothing wrong with my components or settings, the board should not be doing this.

try another bois


man i cant beleive how high the lemon rate is with the reIII simply ridiculous
i was just too happy to take a 100$ loss after a few weeks with the board
i shoulda just takin the restocking fee lol but had too high hopes
asus

[CryptiK]

Ok testing more, doesn't do it running stock clocks, DDR3-1333 etc. I'm now on 30 x 133 @ 4GHz, ram at DDR3-1866 8-8-8-24 and its ok too...............I'll keep testing and hopefully I'll find what is causing this. I will RMA it if this keeps going, try another R3E, and if it does the same thing then I'll sell it and move try another board. Would probably give a UD7 a go and find a full cover block for it.

[zsamz_]

Ok testing more, doesn't do it running stock clocks, DDR3-1333 etc. I'm now on 30 x 133 @ 4GHz, ram at DDR3-1866 8-8-8-24 and its ok too...............I'll keep testing and hopefully I'll find what is causing this. I will RMA it if this keeps going, try another R3E, and if it does the same thing then I'll sell it and move try another board. Would probably give a UD7 a go and find a full cover block for it.

you should try a classy

[CryptiK]

That's always an option.

[Raja@ASUS]

Thankyou, that's interesting. I have at least preliminarily corrected the issue by increasing Vtt to 1.35625v in BIOS, which according to everest is 1.296v under load, and as per your findings this is higher than the real Vtt. I will have to measure at one of the MLCC's as you suggested to see what my board does. I did that with my R2E (I have a TJ07 so the socket area is cut out making access easy) and made a table of BIOS vs Software vs Real voltages.

Thing is I thought I'd sorted it last night by upping it to 1.350Vtt BIOS but this morning it hung on DRAM again on cold boot after PSU power off.

Did you investigate what enabling Vtt LLC did to the voltage? I don't usually run any LLC as I want to avoid transient overshoot conditions. However if it corrected the sag a little but still allowed some droop under load I'd be temped to use it.

I didn't get time to investigate the LLC option, unfortunately.

One thing I would do is check what the board is defaulting the RTL settings to, if they are too tight some of the symptoms you mention would manifest. Hunting something like this down can be like shooting in the dark, though.

[spoolindsm127]

@raju

Since you mentioned RTLs I wanted to ask you. Did you happen to have a RTL calculator specifically for the R3E that you use? I have been using your rtl2 calc and have noticed the board sets RTLs as if the ' Min Sep Delay ps ' field value was 0.58 instead of 0.67.

I also agree that the right RTL settings are essential in achieving a stable overclock.

[Raja@ASUS]

@raju

Since you mentioned RTLs I wanted to ask you. Did you happen to have a RTL calculator specifically for the R3E that you use? I have been using your rtl2 calc and have noticed the board sets RTLs as if the ' Min Sep Delay ps ' field value was 0.58 instead of 0.67.

I also agree that the right RTL settings are essential in achieving a stable overclock.

No, nothing specific for the R3E. The min sep delay was probably left like that when I saved the file. There's normally some drift in that setting at increased clock speeds and somewhere around 580ps is the least I've ever seen the IMC default too. The reason is that clock skews need to be advanced as you increase bus frequency.

[CryptiK]

I didn't get time to investigate the LLC option, unfortunately.

One thing I would do is check what the board is defaulting the RTL settings to, if they are too tight some of the symptoms you mention would manifest. Hunting something like this down can be like shooting in the dark, though.

It's very odd, I can get it to post from cold boot (PSU power off) now at the overclocked settings, but it can't load windows. It seems to be SATA/southbridge related - often says my bootmgr is corrupt. It's not. Sometimes it gets to the 'loading windows' splash screen and freezes.

Ive tried new SATA cables, and I did try setting RTL's manually, all ram timings in fact, I know this kit very well from my previous board.

The odd thing is, once I'm in windows from a hot boot, not problems, it will pass any stability test. If I restart mostly it will get back into windows fine. Shut it down and kill the power to the PSU though and its a cold boot nightmare that I can't seem to track down the root cause of.

[Raja@ASUS]

It's very odd, I can get it to post from cold boot (PSU power off) now at the overclocked settings, but it can't load windows. It seems to be SATA/southbridge related - often says my bootmgr is corrupt. It's not. Sometimes it gets to the 'loading windows' splash screen and freezes.

Ive tried new SATA cables, and I did try setting RTL's manually, all ram timings in fact, I know this kit very well from my previous board.

The odd thing is, once I'm in windows from a hot boot, not problems, it will pass any stability test. If I restart mostly it will get back into windows fine. Shut it down and kill the power to the PSU though and its a cold boot nightmare that I can't seem to track down the root cause of.

Could be several things. A corrupt bootmgr message (when it is not really corrupt) obviously signifies instability, and most likely on the memory end.


Try playing with the torque of your waterblocks - make sure they aren't warping the board in any way - these boards are very sensitive to that. The problem with these custom blocks is that they sometimes allow you to overtighten, which isn't good for BGAs or the transmission line traces.

Lastly, if you have the option, try another PSU.

-Raja

[CryptiK]

Found it. Can't believe it took this long. The C1 ram slot or signal path is damaged. It wont boot at all when overclocked with a stick in that slot alone, boots occasionally with 3 sticks in but rarely loads windows, and I get erratic boot errors when running stock speeds with that slot populated (eg: bootmgr is corrupt or missing etc).

Board will be exchanged for another. Hope the next one works!

Thanks for the advice guys, nice to have people to bounce ideas off

EDIT: Picture now added explaining my feelings

[fmsam]

So, has the Marvell 9128 controller been fixed now then ?
It cant be as poor as it was at launch as surely Gigabyte would not use it on a £429 M/B

http://www.novatech.co.uk/novatech/p...-assassin.html

Anyone know if a firmware update was included in a recent bios from Asus

?

[Raja@ASUS]

Found it. Can't believe it took this long. The C1 ram slot or signal path is damaged. It wont boot at all when overclocked with a stick in that slot alone, boots occasionally with 3 sticks in but rarely loads windows, and I get erratic boot errors when running stock speeds with that slot populated (eg: bootmgr is corrupt or missing etc).

Board will be exchanged for another. Hope the next one works!

Thanks for the advice guys, nice to have people to bounce ideas off

When you get your return board, run it without the full cover block (use the stock sink) for a few days first. The if everything is a-ok use the block, making sure not to go too tight.

[CryptiK]

When you get your return board, run it without the full cover block (use the stock sink) for a few days first. The if everything is a-ok use the block, making sure not to go too tight.

Will do, I don't want another saga like this happening. I nearly did that when I got this board, but the red mist descended and I wanted it all done and in the case asap

Thanks for the advice too.

[Johnmark]

31-38*C for the chokes under full load? (Just to make sure we're talking about the same thing - the ferrite blocks with R68 written on them to the IO side of the CPU socket).

Mine are way over that. I'd say 45-50*C by guessing and touching. I'm running a W3570 at 4.3GHz @ 1.237v and have my PWM freq set to 500KHz, what do you have yours set to? You on water or air? Is there any airflow over that area in your case?

Yes we are talking about the same thing. I'm WCed using the same block and loop arrangement as you. Using an Obsian 800d case (air flow sucks) so I do have a fan that is blowing cold air into the case inside of sucking out as per original design, running PWM at 500 as 1000 didn't seem to buy me anything.
Water temp in the loop is 30c at its hottest.
Glad you found the source of your boot problem! They sure don't make boards like they used to!! Recently put together five crunchers for WCG and ended up RMAing two boards, both GB's within 2 months of install.

[UrbanSmooth]

About to get the R3E up and running again after a few months off while planning the upgrade to my upgrade (went back to the backup rig for the time being). Anything that I should know about the R3E as far as improvements over running it at stock? Hardware will be:

ASUS R3E (water cooled, EK full cover)
Corsair AX1200
i7 920 @ 4+ GHz (water cooled, EK Supreme HF w/ plate #1)
6GB DDR3 Cas 6 Mushkin Enhanced Ridgeback
256GB Crucial RealSSD C300
GTX 480s in 3-Way / Tri SLI (water cooled, Koolance blocks)
Windows 7

Main purpose for this rig: Triple monitor gaming via Nvidia Surround @ 5760 x 1200.