Official Water Cooling Thread of the Stock Maximus SE ROG NB Club:

[Talonman]

Very through and well written review.

Well done raju!

I will be using your info to help me when I get up and running.

I was glad to hear your findings that NB temps over 47°C are prone to failures during longer Prime torture test runs.
I am willing to call that a good line not to cross.

I thank you for the warning that the stock hose clamps, and high pressure pumps may be a bad combo to run togeather.
So one hose flat out, shot off on ya. Wow!
I am glad I opted to not use them on my rig.

Can we post a direct link to the review? It's a good read.

I will be including your 47°C findings in my chart. I hope you don't mind?

Your data has been assimilated.

So far:
ROG Member ----- NB volts -- NB Loaded --------------- CPU ------------------
Renegade5399 ---- 1.65v ------ 34°C ------------ Intel Core2Duo E6850 G0 @ 4GHz
bill d ------------- 1.30v ------ 34°C ----------------------------- Q6600 Stock
ragge86 ---------- 1.65v ------ 36°C ---------------------------- Q6600 @ 3.8GHz
BulldogPO -------- 1.65v ------ 40°C -24/7 QMC crunching clocks - Q6600 @ 3.8GHz
weescott --------- 1.74v ------ 43°C --------(Q9550 444x8.5) 1.49v 62c under load
j0nnyr0773n ------ 1.54v ------ 44°C ------------------------- Q6600 GO @ 3.6GHz
Clump ------------ 1.62v ------ 46°C (After 1HR or Prime 95) ----- Q6600 @ 3.6GHz
raju -------------- 1.65v ------ 49°C (Prime95 Torture Temp) ----- Q6600 @ 3.6GHz
Note: After Raju remounted, max temp went down to 39°C
Note: NB temps over 47°C are prone to failures during longer Prime torture test runs.

[Jedda]

Guess the second board will result in further comments, Raju?
No TIM, poor connections etc make an impression of low quality LC of the TT or Gigabite ilk. All bling and no performance.

[bill_d]

moved from bios 505 to 701 still all auto nb volts jumped to 1.46 and temp to 33c idle and 35 load w/occt

[Raja@ASUS]

thank you for the comments, once we get the revised board we will do a small write up on it using one of the newer bios releases...

Talonman, yes you may link if you wish no problems..


regards
Raja

[Talonman]

Thanks buddy!

raju's Maximus SE review:

http://www.anandtech.com/mb/showdoc.aspx?i=3149

[Talonman]

moved from bios 505 to 701 still all auto nb volts jumped to 1.46 and temp to 33c idle and 35 load w/occt

Just let me know if you are keeping thoes settings, and would like the chart updated...

Odd on the v jump. Thanks for the post.

Mabey they figure the NB does better with a tad more juce?

[ElEctric_EyE]

Thanks buddy!


raju's Maximus SE review:

http://www.anandtech.com/mb/showdoc.aspx?i=3149

Heh, Cheers to that review, especially about the stock WC NB TIM and how it's not much better than air unless you remount with a real paste. I read it all last night, very nice job with that Raju!

[Jedda]

Eddy's X38 mosfet blocks look nice.
Link

[kimandsally]

Heh, Cheers to that review, especially about the stock WC NB TIM and how it's not much better than air unless you remount with a real paste. I read it all last night, very nice job with that Raju!

Same here, nice to see attention to detail which some can't manage, very interesting about the GTL etc.

Great jov Raju

[Raja@ASUS]

thank you guys, I appreciate the support.. Bingo13 helped out as well..
regards
Raja

[jonny_ftm]

Eddy's X38 mosfet blocks look nice.
Link

For people looking to cool their mosfets on air, I'll advice these ones

http://www.thermalright.com/a_page/m...duct_hr_09.htm

They are passive, but you can manage to fit (with some home made stickers) a 60mm silent fan depending on the CPU cooler you have (clearance)

[Talonman]

Eddy's X38 mosfet blocks look nice.
Link

They need to list the Maximus SE on that web page.

It does have :Especially X38 chipset ASUS motherboards, but if they list the Striker by name, the Maximus SE should be listed 4 sure then, with = billing!!

I think I am still sold on the stock ROG cooling so far. I also like raju's advice on using a fan/water combo to cool the Maximus.
I think my TJ09 case, with it's 'open air flowing nature' should help me to have good temps...
I am not sold on the remount yet until I see exactly what temps I get, and if it looks to be required.

Even though I kind of feel like the fan I have installed on top of my ram cooler is a tad hard on the eye's, it pointing right at the heat sink directly above the CPU is making me think it actually will be helping me out some?
I do have that tiny one on the back of my sound card too, blowing all of it's 5cfm's at the lower heat sink.
I may add more fans in down the line, if my system is having high temp issues.

What do some of you other ROG members think...

Correct me if I am wrong, but so far dosen't it appear that even when loaded, the stock ROG cooling on water, keeps most of us in the safe zone?
Even withought remounting everything?

Does anyone think with what we know so far, running the stock ROG block is impeding our ability to OC?

Don't we pritty much just have to watch our NB voltage, and see on how little we can get by with and still remain stable, whilst keeping in mind our loaded 47°C warning track temp?

[RickCain]

Just got mine up and running on water. Thus far NB is 44c and SB is 44c w/ 1.57v to the NB.

[Talonman]

Sweet!! After you get it loaded up for the first time, be sure to report a loaded NB temp, and what GHz your running your CPU at, and it's type.

I'll get you charted up then.

Good picture too. I like the reflection thing ya have going on there.

Thanks for the post.

[jonny_ftm]

I think I am still sold on the stock ROG cooling so far. I also like raju's advice on using a fan/water combo to cool the Maximus.
I think my TJ09 case, with it's 'open air flowing nature' should help me to have good temps...
I am not sold on the remount yet until I see exactly what temps I get, and if it looks to be required.

My expierience with previous boards, is that on high OC you need to active cool your motherboard NB and mosfets
The X38 can't be an exception. The vital settings are:
- C2D CPUs will never, even past 4GHz stress your mb like a quad will do
- CPU clock and vcore: on a quadcore overclocked above 50% (that is 3.6GHz and above) will make your mosfets burn. They will in all cases be lower performers (heat and electric stability are just opposed)
- Memory speed/timings, NB strap and NB voltage, CPU clock: these will burn your NB. Now, we're talking about the most sensitive part in an overclocked PC: RAM management

So, with all this in mind, the NB and mosfets should be the coolest to be stable. If 47°C limit meets Prime errors, I will say a 5-10°C lower temperature is needed for the long run (after all, it is just a chip like a CPU, so what applies to CPUs will apply to any chipset)

Personally, when I receive a motherboard, I remove its stock cooling and at least reapply thermal compound before first run. After all, we all do this for our CPUs and debate what is the best compound (AS5, MX-1, MX-2...). SO we should do it for our chips. Any gain in temperature will lead to better OC performance like with a CPU

Just an advice for future owners of this board: before mounting it, just tune your heatsinks, especially if you plan to get>50% OC from a quadcore

Also, like with CPUs, there are poor and bad overclockers in the motherboard department, as we OC the X38, some of them will be better than others (not all quads reach 4GHz stable OC, so not all X38 would be able to stabilize it)

Wish I could help with testings, but really disappointed to see no one getting his board prime stable above 3.6GHz on the main maximus thread. And if I do remember, no one except Raju tried reseating the heatsink and cooled his motherboard well with fans. And, yep, he's quiet the best performer on high FSBs with a quadcore

At Raju: could you please, in your future X38 motherboards big review mention not only the maximum FSB on a Q6600, but the maximum stable clock speed on a Q6600. In fact, having a benchable 480MHz FSB will be unusable if the system can't get stable above 9x400 MHz on a quad core. The total speed of CPU/RAM will have a stronger stress on the NB than the FSB alone (8x400 much easier to stabilize than a 9x400, but the end user needs the 9x400). It is really strange that quiet no one is getting a Prime / F@H stable above 3.6GHz on his Q6600+Maximus

[Talonman]

Wow! Thanks for your input. I appreachiate the post.

My expierience with previous boards, is that on high OC you need to active cool your motherboard NB and mosfets
The X38 can't be an exception.

Well, we have the NB covered as far as active cooling goes running water through the block, so your saying our mosfets around the CPU will probably be our weak spot correct? I guess whatever we have flowing through our heat pipe, going to our mosfets heat sink isn't getting us credit for being 'active'.... Fusion System or not? I do have some extra fans pointing at my PWM's(?).

Remember: The Maximus SE boxes official stance:
Fusion Block System:
H2O Ready: The next step in thermal solution. The Fusion block system combines the ROG's renowned heatpipe design with the ability to conduct to a water cooling system for a 47% increase in thermal performance compared to other heatpipe designs.

I do think this is a new thing just on our SE correct? If so, we have no data on how well it actually cools our mosfets yet do we?

So, with all this in mind, the NB and mosfets should be the coolest to be stable. If 47°C limit meets Prime errors, I will say a 5-10°C lower temperature is needed for the long run (after all, it is just a chip like a CPU, so what applies to CPUs will apply to any chipset))

Gasp! If we go down the full 10°C and use 37°C for our line not to cross under max load, were going to have allot of people on the chart in trouble. I am getting more and more curious to see some of the sexy high end NB block temps, under max load.
Backing down 5°C would put us at 42°C and still most would be on the warning track. Ouch!
I still would like to know if Intel has an official stand on our NB max temp...

ROG Member ----- NB volts -- NB Loaded --------------- CPU ------------------
Renegade5399 ---- 1.65v ------ 34°C ------------ Intel Core2Duo E6850 G0 @ 4GHz
bill d ------------- 1.30v ------ 34°C ----------------------------- Q6600 Stock
ragge86 ---------- 1.65v ------ 36°C ---------------------------- Q6600 @ 3.8GHz
BulldogPO -------- 1.65v ------ 40°C -24/7 QMC crunching clocks - Q6600 @ 3.8GHz
weescott --------- 1.74v ------ 43°C --------(Q9550 444x8.5) 1.49v 62c under load
j0nnyr0773n ------ 1.54v ------ 44°C ------------------------- Q6600 GO @ 3.6GHz
Clump ------------ 1.62v ------ 46°C (After 1HR or Prime 95) ----- Q6600 @ 3.6GHz
raju -------------- 1.65v ------ 49°C (Prime95 Torture Temp) ----- Q6600 @ 3.6GHz
Note: After Raju remounted, max temp went down to 39°C
Note: NB temps over 47°C are prone to failures during longer Prime torture test runs.

I would want to know for sure there were some lower NB temps than our current range, before I jump.

Personally, when I receive a motherboard, I remove its stock cooling and at least reapply thermal compound before first run. After all, we all do this for our CPUs and debate what is the best compound (AS5, MX-1, MX-2...). SO we should do it for our chips. Any gain in temperature will lead to better OC performance like with a CPU.)

Yeha, except my D-tek didn't require me to melt it off of the CPU to start with. Make no mistake, this would be a big task for me getting all the ROG's cooling off of the motherboard and not hurting anything. Then buying all the correct fittings, blocks, and backplates, and re-mounting....
I wish I could pretend it was a 'why not' situaction for me, but I find it rather a major decission to tear everything off my mobo. It's not even close to a freebie in my mind. Everything is connected too, so were talking heat sinks and block.

Just taking all the stock stuff off, and putting it all back on is involved, and I really don't know how much better I could do than ASUS did?
Mabey their thurmal glue does a decent job?

I think raju may have got a bad NB install the board he received? After he did his remount, his max temp under load fell back in line with the other 'standard production model' Maximus SE's.
I know his temps dropped 10°C, but then again his NB came loose all by itself too, and his initial loaded temp was 49°C. That was our highest temp reported so far. I am betting the next Maximus they send him won't have a loose NB, nor would he probably realize a 10°C drop in temp if he remounted it himself?

Just an advice for future owners of this board: before mounting it, just tune your heatsinks, especially if you plan to get>50% OC from a quadcore.)

Tuning means dismounting everything off of the mobo correct, then reapplying AS5 or such?
Again, not an easy decission for me...

Also, like with CPUs, there are poor and bad overclockers in the motherboard department, as we OC the X38, some of them will be better than others (not all quads reach 4GHz stable OC, so not all X38 would be able to stabilize it)

True enouigh! I hope I got a runner.

It will be fun to see the boys that lay the long green out first, and buy all the best blocks, if their systems will be able to run much faster in the end, than the boy's who just stuck to running the stock ROG Fusion cooling.

I would love to see the air cooled Maximus boys start a temp chart, us stock ROG'ers on water have this one, and a high end water block temp chart. That would be some fun numbers to look at for me!

Let the games begin!

[jonny_ftm]

It will be fun to see the boys that lay the long green out first, and buy all the best blocks, if their systems will be able to run much faster in the end, than the boy's who just stuck to running the stock ROG Fusion cooling.

On this one, you could be damn true. In fact, if you read through the P5E-P5E3, the Maximus and the P5E3 Deluxe main topics, there is one common point: memory deviders are broken and people can't get their RAM sticks to run at the specified speeds for PC-1200 and even PC-1000 for many. So for now, maybe it is the most limiting factor

Really hope Raju in his next coming big review of X38 motherboards he can test the maximum stable clock a Q6600 can achieve rather than stuck on maximum FSB speed. OCCT RAM or a long F@H are the reference for testing the NB-RAM-CPU subsystem while small FFT is for brute CPU test mainly (roughly, small FFT for vcore tweaking and the OCCT RAM/F@H for all the subsystem)

[Raja@ASUS]

If Intel send me a Q6600 I'll be happy to test one, we are at Intel's mercy. I run Prime95 Torture Test (not small fft or even large fft only but BLEND) and state 8 hour test pass limits and provide the bios info (just as I did in the review on page 11). Everyone has their own stability regimen, I run Prime95 for 8 hours on blend for that stuff. The rest is for users who are not concerned about Prime95, like many of the more extreme guys, so it's all about doing a little for everyone. That's all I have time to do (I can't work more than 20 hours a day), with the reviews being so comprehensive. There's only so many people we can keep happy. Sorry if this does not please you, but we have to draw the line on each section of testing based upon time, I think we are already exceeding most. Not only that, but we also realise that no matter how much we do there will always be someone who wants more or someone who wants it in faster time with less overall info (yes we do get requests both ways).

regards
Raja

[jonny_ftm]

If Intel send me a Q6600 I'll be happy to test one, we are at Intel's mercy. I run Prime95 Torture Test (not small fft or even large fft only but BLEND) and state 8 hour test pass limits and provide the bios info (just as I did in the review on page 11). Everyone has their own stability regimen, I run Prime95 for 8 hours on blend for that stuff. The rest is for users who are not concerned about Prime95, like many of the more extreme guys, so it's all about doing a little for everyone. That's all I have time to do (I can't work more than 20 hours a day), with the reviews being so comprehensive. There's only so many people we can keep happy. Sorry if this does not please you, but we have to draw the line on each section of testing based upon time, I think we are already exceeding most. Not only that, but we also realise that no matter how much we do there will always be someone who wants more or someone who wants it in faster time with less overall info (yes we do get requests both ways).

regards
Raja

I do fully appreciate your work, no doubt. Prime95 Blend is a good test for all the subsystem stability. Thought you were on small or large FFT which won't enlighten the weakness of the RAM-board at high overclocks. So it is just great.

No need for a Q6600, I can understand it. But as you do, your QX6800 at 9x multiplier is the best choice you did to satisfy all the performance thursty, but poor people not affording an extreme CPU, we all are. So, it is just great how you bench and the CPU/test bed you used

The only thing, I'd like to find in your tests, is the maximum stable overclock, not only a useless unstable maximum FSB at a low CPU clock. But again, it is "what I like", and you do the best you can already. So, just take it as a constructive comment, a suggestion, and I fully respect all your decisions

[Raja@ASUS]

I do fully appreciate your work, no doubt. Prime95 Blend is a good test for all the subsystem stability. Thought you were on small or large FFT which won't enlighten the weakness of the RAM-board at high overclocks. So it is just great.

No need for a Q6600, I can understand it. But as you do, your QX6800 at 9x multiplier is the best choice you did to satisfy all the performance thursty, but poor people not affording an extreme CPU, we all are. So, it is just great how you bench and the CPU/test bed you used

The only thing, I'd like to find in your tests, is the maximum stable overclock, not only a useless unstable maximum FSB at a low CPU clock. But again, it is "what I like", and you do the best you can already. So, just take it as a constructive comment, a suggestion, and I fully respect all your decisions

8x450=3.6ghz, the same as 9x400 but harder to achieve. 8x450fsb (3.6ghz) prime stable for 8 hours. I also stated how the stability was affected as fsb was raised coming down to 1hour and under between 460 -465 fsb. 9x 400 is easier to get stable than 8x450, all Q6600's can use the 8x multiplier, there's no injustice there to Q6600 users (in the review).

Make sure you do not skim read, there's too many people guilty of this. It's all there (the info).


regards
raja

[jonny_ftm]

8x450=3.6ghz, the same as 9x400 but harder to achieve. 8x450fsb (3.6ghz) prime stable for 8 hours. I also stated how the stability was affected as fsb was raised coming down to 1hour and under between 460 -465 fsb. 9x 400 is easier to get stable than 8x450, all Q6600's can use the 8x multiplier, there's no injustice there to Q6600 users (in the review).

Make sure you do not skim read, there's too many people guilty of this. It's all there (the info).


regards
raja

I agree, just that quiet no one is getting his Maximus board stable (especially OCCT RAM) above 3.6GHz for a quadcore on the Maximus topic... You also write: the DFI is more stable at 465MHz FSB, that means the Maximus is not stable at that FSB (8x465 was tested), which is what most people noted with their board: nothing stable over a Q6600 above 3.6GHz in OCCT RAM. And th efact is that it is not a CPU limit, but a combo motherboard-RAM limit. End users are deceived by their Q6600 topping at 3.6GHz, what a P5B deluxe can achieve often
Anyway, I don't skim read, was just a suggestion, you're the author, you do what you feel is better. Also, didn't read anywhere on the use of Prime95 Blend test for 8h in your article, that's why I proposed OCCT (RAM).

Didn't like to launch such a debate in this thread dedicated to the NB cooling, maybe it should have been in the maximus topic. Anyway, thanks for your efforts, and sure we'll be waiting for the next review

[Raja@ASUS]

Also, didn't read anywhere on the use of Prime95 Blend test for 8h in your article, that's why I proposed OCCT (RAM).

Look under NB GTL voltage on page 11, it's been there all along...

[jonny_ftm]

Look under NB GTL voltage on page 11, it's been there all along...

Ohh, my bad. I've read it, but didn't look so closely at the small Prime95 screenshot to decipher the time length and test type Prime95 was running. I'm a skim watcher (didn't look well at that screenshot)
Thanks for taking the time to clarify it, and hope to see the X38 roundmap soon

[jonny_ftm]

http://www.xtremesystems.org/forums/...postcount=2429

Looks like as I stated, a reseat should be necessary

[TheSamurai]

awww, jonny, you took the time to quote me in another thread. that brings a tear to my eye.

that was a AIR : reseat and WC... so i'm thinking that the reseat gave me around 10* and the WC 4*... i was too lazy to compare them properly. sorry guys.