2 Xeons, ATI 5970 on a single Feser 240

[MrBean]

Hello Guys,
Mmm...I guess I have advocated in several threads these last months about what really is required in terms of hardware to cool a system proper. Over several forums, quite a few members have responded, but, it was pretty obvious my results were not....err.....let's say, accepted......

Well, just a little info:

System setup -> refer to Project Log for more details, but I have to update to reflect current status....
2x E5430 Xeons (watercooled Fuzion V2), clocked up to 3GHZ
SuperMicro X7DWA-N motherboard
8GB 800mhz FB-DIMM Memory (quad channel)
8x 2.5" SAS 146GB drives, cofigured in a HP SAS Drive Cage
1x ATI 5970 (Watercooled by EK)

Single 240mm Feser Rad, 2x Coolermaster Excalibur fans
2x MCP355's (purely for redundancy)
mCubed T-Balancer controller

Please refer to attached screenshot-:
I have pre-warmed system, running IntelBurnTest, as well as Furmark for 1/2 hour.

Then, Stopped everything, and restarted IntelBurnTest, as well as Furmark - took a screenshot 18 minutes into the run. Water temperature reach equilibrium after around 6-7 minutes, and both quad-core Xeons, as well as the dual-gpu 5970 temps were stable.

Ambient around 24 deg C, and cpu's went up on average 16 deg C from idle, while gpu's went up around 12 deg C. Please do not concern yourselves over the differences in cpu core-temps, they were like this from day 1 -> on air I would observe same deltas, so....

People should take note that very rarely you will have all your cores, as well as GPU loaded 100% for extended periods of time - while gaming, at most you will see is 100% load on 2 cores, while GPU may fluctuate a fair bit too.

So, as mentioned in the past, for those (newbies) out there after info, and what is required, don't fall into the multi-loop multi-rad trap - > by all means, if you're building a system and nothing less than above-mentioned scenario will satisfy you, and you have the money to burn, go for it.

If you want a fully funstional, decent wc loop, you don't need to go silly though

[WaterFlex]

very nice temps for this setup

[Vetalar]

fan speed would be nice. and set gpu-z sensors to MAX. anyway its great temps for setup like this.

[MrBean]

Heya Guys,
Ta for the feedback.

@ Vetalar: I will do some runs again, and play with fanspeed (and indicate it too, as well as water-temp via T-Balancer). For this test, fanspeed was at 1600rpm, and very quiet. Water got luke-warm to the touch, at least according to the radiator when you touch it.

I will also set the GPU sensors to max, but those speeds you see was the highest observed for the 2x cores, the same on the 2x trends on Furmark. The Vregs would fluctuate up/down by about 1-2 deg C max.

As said, very stable, reaches equilibrium very quickly, within 10-15 minutes, and will maintain those temps very nicely. Even should we have drastic changes with ambient, the total temperatures will be much lower than (stock) air-cooled setup, and the temps will also be much better dampened (less high/low transitions) tha with aircooling.

So yeah, end of this month, this rig is up and running for 2 years, and it has never missed a beat - I primarily use this as a simulator unit, running a massive Pulp and Paper Control system demo on it, with close to 4000 process IO points being simulated, data historised, process controlled from User Interface (Operator Stations), controlled by Process Control Stations, simulated Paper (in both machine direction and cross direction) data captured and historised, displayed on 3d waterfall maps, et al.

So yes, saddest part is, it can do all of this, plus run prime95, and I can still play FarCry2 without it missing a beat - or overheating. This would not be sustainable on high-end air even, with ambients in Melbourne sometimes reaching 47 deg C - yes, you read correct, 47 deg C.

We have periods here where ambient will exceed 35 for a couple of weeks, and this system copes with it.

So, one of the main reasons for my statement about 99% of watercooling rigs out there are over-engineered.

Yes, I know this is XS, but then again, watercooling (even with 5 loops/10 rads) is hardly extreme?



But, guys, this is not an effort to flame anyone, merely an attempt to provide newcomers with some base information, and implementing practical solutions without breaking the bank.

Would appreciate feedback from everyone, and your (extreme) experiences

[Movieman]

Personally I think you ought to take that beast and see if it's stable enough to crunch WCG with!..
Nice setup my friend.

[MrBean]

@ Dave

I will d/l the application on the weekend and do a few hours worth of crunching - one thing I need to do is to vid-mod the cpus, and push fsb to 400 - this will then have the ram running at native speed, and cpus at 3.20 gig vs stock 2.66.

I might push vid up to safe watercooling levels, and see if I can get 3.5-3.6, hopefully the FB-Dimms will keep up - as you know, these SM boards are super-stable, but not a single OC setting in the BIOS....so we have to revert to dirty tactics, hehe. I still owe you big time for sending me this board, Dave :flowers:

But, all truth be told, this is a great board, Dave, yes, even for a 2-yr old system - it plays every title I can throw at it, and the 5970 drives the 30" Dell just wonderfully. Overkill for my racing sim though, but hey, it's fun

[Movieman]

@ Dave

I will d/l the application on the weekend and do a few hours worth of crunching - one thing I need to do is to vid-mod the cpus, and push fsb to 400 - this will then have the ram running at native speed, and cpus at 3.20 gig vs stock 2.66.

I might push vid up to safe watercooling levels, and see if I can get 3.5-3.6, hopefully the FB-Dimms will keep up - as you know, these SM boards are super-stable, but not a single OC setting in the BIOS....so we have to revert to dirty tactics, hehe. I still owe you big time for sending me this board, Dave :flowers:

But, all truth be told, this is a great board, Dave, yes, even for a 2-yr old system - it plays every title I can throw at it, and the 5970 drives the 30" Dell just wonderfully. Overkill for my racing sim though, but hey, it's fun

You will run into a lack of voltage issue at 3.5-3.6..
I'll bet you lunch it won't do those speeds 100% load.
To get the 400FSB with those chips you do a BSEL mod.
I think it's the G30 pad that you toss a tiny piece of electrical tape over.
That changes the 333 to 400 and golden.
That you need VERY good eyes for and the smart way is tohave an extra pair of hands with a 5x magnifying glass to look thru and a tiny pair of tweezers to hold the piece of electrical tape with.
I've done it but it's patience time if you follow me.

[MrBean]

Heya Dave,
Yes, that's why will do BSEL mod (and adjust VID)

It is stable at 377fbs, around 3.05ghz, but not anything higher with stock vcore. So, on the tape will go Missus can hold the mag glass, I wil, handle the tweezers and tape, hehe. But I will aim for the golden 400mhz, as this will get ram to it's native speed, and will give me a decent boost in performance. Will keep you posted!

[Movieman]

Heya Dave,
Yes, that's why will do BSEL mod (and adjust VID)

It is stable at 377fbs, around 3.05ghz, but not anything higher with stock vcore. So, on the tape will go Missus can hold the mag glass, I wil, handle the tweezers and tape, hehe. But I will aim for the golden 400mhz, as this will get ram to it's native speed, and will give me a decent boost in performance. Will keep you posted!

ahh, there are two types of BSEL mods..
One for FSB and one for voltage.
I never did the one for voltage so good luck!

[MrBean]

Yeah, somehow I doubt this system will be stable with only FSB BSEL done - with softfsb, I can get it to boot into windows at 400fsb, but it's very flakey - so I will need VID adjustments too - no worries, I will grab the Intel specsheet and do adjustments accordingly - I might bump it to 1.275 or such, and just leave it there - should give me some headroom should I wish to go higher than 400fsb/3.2ghz......

Somehow though, it is fine as it is at 2.66 even....but, as an overclocker, I just have to - after all, this is XS

[Vetalar]

MrBean

i think key to such good temps in Your setup is pumping power... could it possible to run loop with one pump?

[MrBean]

Heya Mate,
I had one MCP 655 before, same result

Only reason I went to the dual setup, is due to space-limitations within the Blackbird case - the new dua setup is slightly lower, and fits above the SAS HD cage in some leftover space, and out of eye sight.

If anything, I think the system ran a little cooler with the single pump though, due to slightly less heat dumped - but, overall, pumping power won't effect cooling efficiency at all - unless you use a very restrictive loop - which is not the case here.

But thanx for your comments though, it's always good to debate these issues

[Zeus]

There i am, worrying if my 480 can cope with my CPU-2x5870 and fullboard block. I have my 1450rpm GT's at 7V, running them at 12V gave less than 1ºC watertemp difference under load!

Thanks for this thread.

[neonatas]

@ mrbean,

I took you advice and ripped out my second loop, and i now have my 875K @ 4.0Ghz and a 480GTX running on a feser480 with 2 noctua's @ 5volt and a d5 pump @3. Max temp is 60 degrees on cpu, en 55 on gpu.
Ofcourse temps where better in the old situation, but getting rip of 6 fans + second pump. makes alot it alot more silent!

[MrBean]

You guys are welcome

Glad things are going better (more silent, but still respectable) for you, neonatas!

[Cookiesowns]

Damnit, I shouldn't have bought a quad to install in my 800D, almost dremeled myself., haha nah, its still working fine. Well at least now I know that Water cooling my next rigs won't cost me an arm and a leg.

Thanks for this..
But I guess my temps are wack, maybe 980x's are just much hotter than the xeon 54series?

[Church]

Cookiesowns: that + higher overclock.
From googled out specs of E5430 Xeons it's 80W cpu. I'm guessing minimal overclock to 3ghz can be gained with no voltage rise. For your 980X it's already 120W at stock clocks. If you overclock it to 4.4+ghz/1.4+V your cpu will dissipate 225-300W. Tiny bit more then those xeons . I'm guessing also that you have overclocked your two 480 fermi-s aswell? Then probably those also will dissipate more heat from 300W@stock clocks each to some 400-450W.

[wez]

As always, it depends on what you expect from your loop, and how noise sensitive you are

[bundymania]

Yep, i remember a guy @forumdeluxx.de which said, that a Tripleradi is enough to cool a i7 920 4.2ghz + 2x GTX 480...then we asked about his temps and fan speed: 52° Watertemp and Fans @ 1700 RPM:

[Church]

bundymania: exactly. One can think of that tripple rad with such fans like that of double tripples, as by skinneelab tests going fan rpm wise 1000>1800 aproximately doubles rad cooling capacity. So physic laws are still in effect , and if one wants quiet cooling and hot overclocked hardware in same package, no escape from lot of rad sections.

[MrBean]

Hello Guys,
I think the numbers you're using are estimations, based on what's out in the good old www, and not measured. The only way you can determine wattage of the CPU would be to measure it.

The problem you're facing wit overclocking the modern processor reminds me of the old Prescot days - similar issues with leakage current, and then, to further complicate things, a significantly smaller die - so in reality, you are having issues getting the heat away from (inside) the die, channel it to the surface, where it can be dissipated into the "condensor"

I don't beleive those "hypothetical" numbers of 225-300W for a single quadcore vs the "theoretical" 80W for my single quadcore.

And then again, how often is your system at full load? Something else to consider.

I will run tests on the weekend again, and provide water temps too - and I do not believe they are even 50 deg C - and yes, if they are, it means I am getting the rad to do something useful, ie what it was designed for. Why ever would you want to have water temperature remain at ambient - at the end of the day, it means the rads are not doing anything - read over-engineered......




But, I am rambling, and there are good arguments both ways, so I will log watertemps on the weekend. Now, I'm off to the gym to go and do a nice workout.

[Church]

"what's out in the good old www, and not measured" - so majority of LC crowd when relying on test results like from skinneelab and other reputable testers are plainly all mistaken?
As for hipothetical power usage numbers, tell testers/writers for articles like this, that they also are gravely mistaken and should stop lying to people.
"And then again, how often is your system at full load? Something else to consider." - are you saying we should compare temps at idle?

I just pointed that i see no wonders in your results that your hardware with given heat dissipation and cooling capacity with that fast fans give such temps. I stumbled upon this thread from link in other forum - what i don't want to see newbies getting impression that they will get same temps with that small rad even in cases where they heavy overclock and want silent fans at same time.

[MrBean]

@ Churchy: As mentioned, there are very good arguments in this thread, and, in many other threads like this As much you don't want to see noobs being brought under wrong impressions - well, I am in total agreement with you. Point is, my fans are running 1600rpm, it is a relaitvely quiet system, and 1600rpm is not that fast - or noisey.

The "wrong impression" argument goes the other way too - I don't want noobs telling newer noobs that you need 2 quad rads and push-pull configuration to cool a single cpu and 1 or 2 GFX cards - no matter what they are, and what there ceiling is on water

As an electronics engineer, I can safely say that unless you measure direct current consumption of the procesor, and know the efficiency, as well as the percentage of current converted into heat due to very small litho process (leagage current) used in manufacturing process with modern cpu's, (and gfx gpus for that matter), we can only guestimate the heat released by these components, and it is not an exact science.
Hence my comment about the 225-300W estimated heat dissipation of the said overclocked cpu.

Again, I also mentioned that if you're shootting for the stars wrt clockspeeds, go for LN2 or similar -> 99% of blokes, even on this forum here, use there PC's for something else than just OC numbers - or am I wrong?

And wrt your question about idle temps - and I say this with a bit of tongue-in-the-cheek -> you are probably more correct than you think, as most modern PCs sit at idle most of the time, hehe.

So I am merely trying to dispel the myth about over-engineering - nothing wrong if you need all the cooling capacity you feel necessary to keep your system cool - all I am saying is that most people do not need this - yes, even on XS.

edit:

As for hipothetical power usage numbers, tell testers/writers for articles like this, that they also are gravely mistaken and should stop lying to people.

ninja-edit: I commend the author of the article for the thoroughness of his testing - great result. I would like to remind you though, that one should not confuse full system power numbers with that of the CPU alone - nor should one assume that all the electrical power is converted into heat.

At most, I "assume" your 4.2ghz overclocked i7 cpu will consider approximately the same power as both my Xeons.


Also refer to my results, I used the same artificial situation, by running all 8 physical cores at 100%, as well as running Furmark in dual-gpu mode - so, max power.

In real life, you won't ever get there - unless you're using your system for something very special - which, again admittedly, some folks on here do, and thus this thread "exempts" them from reality.

ninja-edit 2:

Please note, that in the following quote, all credit goes to the Author of the article at xbitlabs!

I would like to kindly point to the average overclocker, gamer and cruncher, the wisdom of the conclusion below - there is a point with any cpu, where you reach the point of diminishing returns, and, for crunchers, or guys running massive simulators, or, photo-editing applications et al, you're almost always going to be better of, at some point on the overclocking curve, to get another processor, ie go for dual quads or whatever, as opposed to pushing a single cpu past that point - you will get better overall numbers from dual processors running at say, 3.8ghz, at maximum efficiency, as saying a single processor running at crappy efficiencies at 4.2 ghz plus.

Thank you for everyone's input though, much appreciated!

Conclusion
In our articles we always talk about processor overclocking potential and even benchmark overclocked systems. However, it is important to understand that CPU overclocking is not only an affordable way of increasing the system performance without any additional investments. Although overclocking has gone down to the consumer level these days and doesn’t require the user to possess any specific skills or knowledge, there are a lot of hidden obstacles hiding behind simple adjustment of the BIOS Setup parameters. Today we tried to reveal one of these obstacles that many computer enthusiasts face: CU overclocking increases its power consumption. Moreover, it is fairly easy to push the processor power consumption far beyond its calculated TDP and everyone should be prepared for that. Processor cooling system should be able to dissipate a lot of heat. Processor voltage regulator circuitry on the mainboard should have more than twice the “current reserve” available. And the system power supply should have at least 1.5 times the power capacity of the system working in its nominal mode under maximum load.

So, it turns out that you can’t really overclock without any additional investments. Moreover, you have to make them beforehand, even when you are just putting the system together. And unfortunately, we can’t really neglect them: we always experience power consumption increase during overclocking, and it doesn’t matter what platform and what processor we have. However, we don’t want to overestimate the power consumption increase during overclocking, either. As our tests showed, there is always a cheaper” alternative.

If we look at the power consumption graph during overclocking of almost any processor, we will see, that it consists of two very characteristics parts. The first one is a relatively flat part of the graph that increases slowly, and the second one is the part where the graph starts growing very abruptly. The turning point on this graph doesn’t occur spontaneously, but happens at a very specific moment of time – namely, when it comes to raising the CPU Vcore. In other words, you don’t need t fear that the processor power consumption will skyrocket, until there is real need to adjust any of the system voltages to ensure stability. As for overclocking at nominal voltages, even though it doesn’t allow you to hit very high performance levels, it is not so useless after all. For example, several processors from our today’s test session worked at 3.6-3.8 GHz frequency and at their default core voltage. And by the way, overclocking like that doesn’t really stress the cooling and power systems in any way, but also allows to save some power when working on some resource-hungry tasks.

In conclusion I would like to stress that when you overclock different processors up to about the same frequencies, they still consume different amounts of power. It is quite obvious that quad-core processors consume more power during overclocking. However, this is not the only dependence that we observed. Namely, LGA775 CPUs proved to be the most energy-efficient during overclocking. Also, Athlon II processors demonstrated very modest power appetite. I would also like to point out Core i3-540 processor. This is the only CPU tested today that is manufactured using the most advanced 32 nm technological process, which had to affect its results. It obviously consumed the least power of all at its nominal frequency as well as after overclocking.

As for the power consumption “leaders” they are Phenom II and Core i7 CPUs. Overclocking these processors causes a much more substantial increase in the system power consumption than in all other cases. And I am not talking only about the absolute values here. Even in relative prospective, overclocking these CPUs may cause about 4—50% increase in system power consumption. Therefore, if you are looking for a suitable power supply for an overclocker system, make sure that you will have at least that much capacity in reserve.

[Church]

MrBean: i can give few more pointers for going less rads aswell:
- most common usage is gaming, and it .. well, never gonna heat hardware that much as stresstests/crunching do, i've even noticed few guys telling "it's stable when i game", "i don't care if box doesn't pass linpack/furmark";
- even such underrad is comparable to efficiency at least to that of stock air cooling, or even bests it a bit (arguable, as yes, most probably more efficient then very constructively/dimension wise limited gpu cooling, but then again does it make sense to get only slightly better cooling by investing that much in buying all LC components? I kind of compare best top air coolers to 1.5x120 rad/and top gpu coolers to 1x120rad cooling efficiency/noise wise);
- when gaming, usually people have ramped up sound on speakers or headphones and don't hear noisy fans anyway, but when they really need quietness, eg. when sleeping, box mostly idles if it's switched on at all;
- it's smart to buy/build cooling hardware that's "just enough" for whatever user needs, instead of going overkill/xtreme because that seemingly increases e-peenuz length

Now to the other side arguments
- When 1st timers decide to go to LC, they usually go for two main reasons - to get more efficient cooling or more silent cooling (or both in different ratio) (+i purposely leave out bling / interesting hobby reasons). With underrad situation that's similar cooling/noise wise to air cooling they'd be much better off investment wise to improving air cooling (eg. getting best air coolers like Thermalright Silver Arrow or Noctua DH14, moding case for direct extra airflow directed to GPUs, change to better fan selection - it all should bring similar performance as underrad LC but by spending way less and with less complexity to implement).
I've heard people saying - "i've upgraded my air cooler to Corsair H50, it's liquid cooling, so should be better, but it's as loud and i'm not getting better temps!!" - illustration what many expect from LC, and what often they get in underrad configurations. Also several newbies asking questions in LC forums/threads about building LC, because they hope to get higher overclock then on air (usually mentioning 4.5/4.8ghz - i'm not disputing about returns of such overclock, but well, that's what they want to get);
- Usually best way to lower temps or lower fan speed for lower noise in LC is to add MOAR rads[tm] , at least in majority of cases increasing rad area improves cooling more then investing in better waterblocks / more pumps, so adequate rad area is simpliest(and cheapest way within LC boundaries) way to get those two main things that many from LC-ers initially aimed/wished for;
- imho it's simplier to initially mount even more rads then is actually needed, to have cooling system capacity reserves for possible future even hotter coolable hardware upgrade and to do Big Modding properly at the beginning instead of gradually replacing/upgrading to bigger rads, leaving older rads unused and having to replan/redo moding each time. Even when it's more then needed, one can simply turn fans even slower then required, also in overall LC costs picture rads+fans is only part of cost, sometimes not even biggest (eg. compared to cost of all the waterblocks used - cpu/one or two gpu full cover/motherboard fullcover, one or two pump, some smart fan&pump controller .. rest of costs might rise rather high compared to cost of even very big rads with fans)


At the end most probably The Truth [tm] is somewhere in the middle. No need for overkill builds with insane components, yet also not much practical sense in underrad builds either.

[Vetalar]

it would be nice that LC users, who uses their PC with "typical" tasks (gaming, web surfing etc.) give us their _daily_ MAX and MEAN CPU and GPU temps (from e.g. speedfan). this allow us to make some estimations and give us some realworld info. just my 2 cents.