1. OCAU Merchandise is available! Check out our 20th Anniversary Mugs, Classic Logo Shirts and much more! Discussion in this thread.
    Dismiss Notice

Radiator experiment - to MOARRAD, or not?

Discussion in 'Extreme and Water Cooling' started by mrbean_phillip, Jan 2, 2014.

  1. mrbean_phillip

    mrbean_phillip Member

    Joined:
    Jun 18, 2002
    Messages:
    5,192
    Location:
    Doreen (3754) Vic
    Hello,
    Generally, there is a lot of confusion wrt rad-sizing, when it comes to watercooling computers.

    A few questions:

    Would any of you like to see what temps a system like this would be running at when idling, normal day-day use, as well as extreme gaming sessions, and then 24 hours of severe CPU/GPU benchmarking? Running a 240, or 280 vs a 480 rad?

    Would you want to see me employ only standard software measurements, via OS, or actually use some high-end industrial temperature sensors and trending equipment to measure the results?

    What would be constituted good temperatures, and why?

    Do you believe a single 240 rad can very successfully cool the over clocked 3570k, as well as the 780GTX.

    What about running triple 780GTX's, as well as a 3930k, on a single Alphacool UT60 280mm rad, with fans only spinning at 1200rpm?

    Let's have a debate, let's refrain from getting personal, and let's keep the emotions out. Just your input, expectations et al.

    I will gather the correct equipment in the meantime, and will conduct the tests if that is what people wants to see.

    Let me know which you would like to see, but the majority of my testing will be conducted on the 3930k with triple 780's, this is probably as extreme as most serious gamers will go at the start of 2014, so any lesser system will be happy with the radiators specified if the system is running lower specs.

    I will also conduct some testing on my Wife's setup, running an Alphacool UT60 240mm rad, 3570k and 670GTX, all overclocked.

    Let's start the discussion, hopefully at the end the tests can help watercooling enthusiasts with what can be achieved with such a small WC setup....

    Links to both this Discussion and the Results thread in my sig...

    Cheers,
    Beano.
     
    Last edited: Jan 12, 2014
    Ubermarshall likes this.
  2. Acesi7

    Acesi7 Member

    Joined:
    May 25, 2009
    Messages:
    2,148
    Location:
    Melbourne
    Love it.

    Healthy debate and practical testing.

    I've got a spare 560 rad lying around that you can use for testing if you like beano.

    Also got a bunch of fittings and other items that might come in handy.

    Let me know if you need anything and I'll see what I can wrangle up.

    This is the thread that we've needed for a long time.

    I'm also happy to put my system up for any testing that people want to see.

    3960x
    3-way Classified 780's (non-Ti)

    My idea of a "successful" water cooling setup is anything that operates both Quieter AND cooler than it's air cooled counterpart under 100% load for at least 1 hour.

    So I'd say that this should be considered the initial success/failure point for a WC system.

    Sound fair?
     
  3. Sabien

    Sabien Member

    Joined:
    Jun 27, 2001
    Messages:
    1,347
    Location:
    Brisbane 4178
    Enclosed in a case or test bench?

    Push/pull setup or just one way?

    I have a very similar Prodigy setup to your proposed test bench but also with a 200mm phobya but both with only a single set of fans on each. I have mine O/C'd plenty so I'm throwing my support behind you Beano.
     
  4. bobster

    bobster Member

    Joined:
    Nov 11, 2011
    Messages:
    1,027
    Location:
    Sydney
    :thumbup: Look forward to this

    I have spoken to mrbean_phillip a few times regarding this matter and from the advice I did some testing and you can see it in my R4 build:-
    http://forums.overclockers.com.au/showthread.php?t=1090711&page=2

    I cooled a 7970Ghz and 4670k at very reasonable temps on Swiftech H220(thin 120mm rad and small pump) on crappy enermax fans.

    The R4 was rebuilt for my friend that bought it from me with D5, Noctua NFF12's and Alphacool Monster 240 and got a 10-15 degree drop in temps.

    Looking back at boths setups using 240mm rads I would have just put the Noctua NF-F12's on the swiftech for slightly better noise and temps and been happy. Realistally as long as the CPU is under 70 degrees who cares
     
  5. OP
    OP
    mrbean_phillip

    mrbean_phillip Member

    Joined:
    Jun 18, 2002
    Messages:
    5,192
    Location:
    Doreen (3754) Vic
    Hey Mate,
    Would be good to separate fact from fiction floating around on the Watercooling Interwebs, and yes, that includes my 'fiction'

    I may still be wrong :)

    I have a very nice Feser 480 sitting around doing nothing, another Feser 240, as well as Swiftech 240 and 120's, 2x Coolwave 140's, then Alphacool UT 240, 280 and 360 rads, as well as some others I have forgotten about.

    I will shout if I need help, thanx for that. I am ok with fittings/tubing et al, so no worries there either.

    I guess the one thing we need to get consensus on, is what constitutes a decent gaming rig?

    I always go overboard with hardware, so I am not a good example in this instance...

    I guess we can assume something like a wildly clocked 3570k and 780GTX.

    But then again, I am so firm in my conviction that we can also use my iRacing rig....dual 780's, 3930k, single 280 rad, all in a very compact media pc box? Ouch, am I setting myself up for failure :lol:

    Thanx for the offer, will keep it in mind - I will need to get a reliable meter to do db measurements with.

    @Sabien: I will do a fully functional enclosed 'daily' PC, as your wife, mom or pop might want to have, so no open test bench type testing - whilst this is great for epeen, it is bad for everything else, especially with small children and cats around...

    Fans will be normal pull or push (preferred), not both....again diminishing returns with both...I will rather use a slightly thicker rad, and fans with a better static pressure/flow characteristic....like the Gentle Typhoons....in my opinion the best 120mm fan out there...

    @Bob: your results were very good with the hardware you had. Wrt the NFF-12, reason why heat pipe coolers are so good, is it takes a huge amount of energy to convert liquid to gas, that phase change principle is what draws the 'heat' more efficiently from the CPU/GPU...

    I think next generation coolers will have both heat pipe evaporator, and watercooled condenser (radiator)...but certainly, work needs to be done in CPU/GPU cores to reduce the culprit of heat generation, leakage currents.....I guess some years into the future we might see better semiconductor technologies preventing this.


    Cheers,
    Beano
     
    Last edited: Jan 2, 2014
  6. sjp770

    sjp770 Member

    Joined:
    Apr 23, 2009
    Messages:
    1,724
    Location:
    Sticks, NSW
    I love the idea, I personally think it can be done, but the main issue will be fan noise. Whatever testing you do make sure there is some way to measure dB.

    The MOARRADTM idea is not that it cant be done with less, but for the most part more is better because pumps are quiet and you can have quieter fans at lower RPM than a smaller setup. At least that my idea for going WAY over the top with rads. I'm running 2x 560's for 2x 290X and an OC'd 3930k and will be cooling the mobo soon as well. Would I try it on a 480? maybe if I was desperate, but it would have to be a great performing rad and push pull with loud fans as the 290X cards pump out a lot of heat. I'm also aiming at more cards in the future.

    I think what a lot of people forget is once you go watercooling your expectations for a max CPU temp is lower. I hit 60c tops but I could be running a lot less cooling hardware and hit 85c top and still be fine.
     
    Last edited: Jan 2, 2014
  7. Acesi7

    Acesi7 Member

    Joined:
    May 25, 2009
    Messages:
    2,148
    Location:
    Melbourne
    I'm running a significantly higher heat load than that with approximately half as much rad as you are, and my fans run @ 1100rpm (would go lower, but my crappy fan controller turns them off after that).

    I think people put a little too much stock in this also, I don't see much of a difference from 1100rpm to 1900rpm... I don't know the exact figures, I'll check them when I get home if you like.

    But my 3x 780's all overvolted and 3960x overclocked pull over 1500w from the wall (hence the second PSU), and this is being happily cooled from my 360 + 360 setup, and honestly I think could get away with 1 UT60 360 rad.

    Yes my temps would rise, but would they rise above aircooled. Highly doubtful.


    Edit: I think a 3570k heavily OC'd with a 780 is a very good representation of what kind of heatloads most PC are running. I tend to agree that most SLi based gaming systems are normally 760's 770's and the like. Not sure about AMD to be perfectly honest, so open to suggestions in that arena, however in my experience, most people tend to either have 1 very powerful GPU or 2 less GPU's that equate to the approximately 1 high power GPU.
     
    Last edited: Jan 2, 2014
  8. sjp770

    sjp770 Member

    Joined:
    Apr 23, 2009
    Messages:
    1,724
    Location:
    Sticks, NSW
    My current heat load doesn't include the motherboard and was being cooled fine on a single 560, but the fans would ramp up to full and I would see a difference in air temps from intake to exhaust of at least 4c. The fan speeds would make a big difference on the water temps, and a small difference on the cpu temp. I had one night when the room heated up so much from the PC, the ambient temp went up and my water got to near 90c 65 c on the 1x 560 (EDIT CPU hit 90, system locked up). The big difference with water vs air is that if your water heats up past a certain point you wont be able to bring the temps down with a smaller rad. You're running warm water over a heat source and it warms up further.. an aircooler never re introduces heat to a CPU, it always travels away unless something is seriously wrong. marked as BS.. I now agree

    For me it was more about what can I fit into the case.. I dont think that anyone can say its making things worse, but I completely agree when it's stated that most of the time adding more radiators is unnecessary and expensive.
     
    Last edited: Jan 2, 2014
  9. zero_velocity

    zero_velocity Member

    Joined:
    Sep 16, 2010
    Messages:
    2,957
    Location:
    QLD
    Ah, this seems to fall inline with what we have all been saying!

    How does that even happen? What happened to your pumps? last I looked they had a max water temp rating of 60 degrees before they esploded :confused::confused:

    Nah I disagree with this, the big difference with water vs air is increasing the heatsink surface area.

    Just like the water temp, undissipated heat will remain on the heatsink, which is why they can get so blooody hot.... otherwise by this logic heatsinks would remain at ambient temperatures too (absolutely not the case).

    WC vs air cooling is as follows;

    waterblock == copper block in contact with CPU.
    water/coolant == heatpipes for heat transference
    radiators == heatsinks to remove heat from the source

    So basically what I am getting at is that air cooling and water cooling theories are identical
     
  10. Acesi7

    Acesi7 Member

    Joined:
    May 25, 2009
    Messages:
    2,148
    Location:
    Melbourne
    ...

    Calling BS on that one mate...

    90 degree water temperature???

    Your components would fail if the water was that hot.

    Do you mean 90 degree component temperature?

    The water would have no cooling effect what-so-ever on the components if it was 90 degrees and the parts would all overheat and shutdown.

    Also, an MB block will add a minimal amount of heatload when compared to a 290x/780 or a CPU...
     
  11. sjp770

    sjp770 Member

    Joined:
    Apr 23, 2009
    Messages:
    1,724
    Location:
    Sticks, NSW
    BS called.. bad memory. 65c

    :wired: whoops

    yep, that sounds better..

    It was 90c CPU temp, it locked up and put me into panic mode.

    I expect this thread to be good for clarifying various assumptions and destroying ideas with no basis. i'm learning already :)
     
    Last edited: Jan 2, 2014
  12. shane41

    shane41 Member

    Joined:
    Nov 10, 2008
    Messages:
    8,487
    Location:
    dice que no sabe
    This all sounds very interesting. Re the temp testing part.

    I'm not sure what a gpu block will add to overall temp.
    As I only have an AIO running atm.

    But.............a 3570K does not get super hot.
    Even under a $40 air cooler, it is cooled quite well.
     
  13. Tekin

    Tekin Member

    Joined:
    Nov 16, 2002
    Messages:
    4,039
    Location:
    Elsewhere.
    I've never been a big proponent of the MOARRAD option...the physics behind it doesn't seem to stand up given the inherent heat capacity of water.

    It'd definitely be interesting to see the following if possible:

    - Reported CPU Temp
    - Reported GPU Temp
    - Water Temp
    - Ambient Temp

    Be nice to have idle (averaged over 30 minutes perhaps?) and 100% load (averaged for 30 minutes) and perhaps a 'standard usage' (heavy gaming?)

    From that we should be able to control for the basics as well as normalising for mounting differences etc.
     
  14. OP
    OP
    mrbean_phillip

    mrbean_phillip Member

    Joined:
    Jun 18, 2002
    Messages:
    5,192
    Location:
    Doreen (3754) Vic
    Hey Shane, is the air cooler a simple aluminum heat sink, or a heat pipe cooler? In the latter, it is actually a phase-change cooler :) pedantic, I know, but important to make the distinction....

    @sjp: wrt the ambient temp, and running ever hotter water over your CPU/GPU, bear in mind, that very same ambient temp will much quicker overpower an aluminum heat sink with far lower thermal capacity than water...in other words, of all known medium out there, water has the largest thermal/heat capacity, meaning under the same environmental conditions, it will do a better job than any heat sink, due to this characteristic...

    Cheers,
    Beano
     
  15. OP
    OP
    mrbean_phillip

    mrbean_phillip Member

    Joined:
    Jun 18, 2002
    Messages:
    5,192
    Location:
    Doreen (3754) Vic
    That very 1st sentence gave me goosebumps, thank you, Mate :) Especially the recognition someone finally gives to the Specific Heat Capacity of water, the highest of all known single or combined known affordable elements out there.

    I will use modern industrial Long-term trending Historian to trend temps over let's say, full 24 hour cycle, or longer if required, to see what effect ambient temp variation have over both idle and loaded temps.

    I fully expect at idle, that the CPU, GPU and water temps will maintain the delta to ambient, i.e, if ambient rise by 10deg C, the other s will follow suit after point of equilibrium is reached......

    Wrt load temps though, a totally different animal, I do not expect those same variables to follow ambient, for the same 10 deg rise, I would expect a much, much smaller delta to ambient, maybe 1-3 deg C, if that much.....this due to the Heat Capacity of water...so equilibrium would not be much different than equilibrium temp at the lower ambient, when system is fully loaded.

    Of course, it also means to be logical about this, any system will only cope with the higher ambient heat load within limits....but, the ambient temps that will have that negative impact on my test bed, will be so high I will choose not to live there. I suspect 50+ C ambient will start being problematic for such a small system....

    But only speculation, so take it with a grain of salt - feel free to share your thoughts, thanx Guyz.
     
    Last edited: Jan 2, 2014
  16. shane41

    shane41 Member

    Joined:
    Nov 10, 2008
    Messages:
    8,487
    Location:
    dice que no sabe
    Just those basic heat pipe coolers with a 12cm fan.
    Coolermaster Hyper 212 evo is one of them.

    My 2x system here, 3570K air cooler vs 4770K on Swiftech H320

    The 3570K OC's nicely on that setup. Temps still good.
    But the Nasty Haswell 85deg peak the other day
    running LinX ( Low prob size ) @ 42x

    I just think the Ivy is gonna be a limp to push the water cooler.

    Testing I think should be something most of us can achieve.

    3570K 4.5ghz
    Gpu Stock

    In a medium tower size box inclosed.

    I'm more interest in the " more Rad theory " if it can lower temps & fan speeds for the same settings.

    60deg load would make me happy. :thumbup:
     
  17. OP
    OP
    mrbean_phillip

    mrbean_phillip Member

    Joined:
    Jun 18, 2002
    Messages:
    5,192
    Location:
    Doreen (3754) Vic
    Hello Shane,
    Thanx for the comments. Points noted!

    There are other factors to consider wrt full load CPU temps, no affordable pre water- only system will take us past the physical limitations we face.

    Plus, I take quite a few of the posted temps with a grain of salt, those thermal diodes is only an indicator of core temps, instantaneous transients are the problem father here, second to that is calibration inside the Bios to accurately read that same temp.....

    The temp we see in the OS is very seldom a true and accurate indicator of the real situation inside the core, and to often misused and abused for a larger epeen....or out of ignorance...

    But that's just me.....

    Cheers,
    Beano
     
  18. shane41

    shane41 Member

    Joined:
    Nov 10, 2008
    Messages:
    8,487
    Location:
    dice que no sabe
    AIO solution was just what I'm using.
    I mention that in regard to ( having no experience - add gpu to loop )
    What added temp will do?

    I understand some temp measurement to be unreliable.
    Software not sure if good enough ( Core temp, Real temp etc )
     
  19. fast fate

    fast fate Member

    Joined:
    Feb 2, 2012
    Messages:
    23
    Location:
    Perth
    I'll be following this thread, maybe even adding to the test data if I can replicate mrbean's testing meathods - once finalized.
    This could be great especially if test equipment is not overly expensive and meathods are easily replicated - info and data gathered over large spread of hardware would be valuable info..
    My current theory seems to contradict that expressed so far in this thread.
    I try to fit as much rad as possible inside my case with as many fans as possible running in push/pull where ever possible.
    2 x 420 and a 240 P&P near completion
    [​IMG]
    I believe the increased surface area of what others say is unessesary rad/s allows me to run the fans at lower rpm's and maintain cooling performance.
    A pretty simple test to start off might be to just turn fans off on one cooler pack, cardboard over one side so no air flow through the rad to act as passive cooler, but instead would be large res in the loop.

    EDIT - what is the correct coding for a spoiler here ??
    Mine didn't seem to work :(
     
  20. OP
    OP
    mrbean_phillip

    mrbean_phillip Member

    Joined:
    Jun 18, 2002
    Messages:
    5,192
    Location:
    Doreen (3754) Vic
    Nice build there, well executed :)

    More fans and rads will always be better (at least theoretically) than smaller rads, less fans.

    I guess the question is one of ever increasing diminishing returns, as well as the real value of contribution to the cause ......

    Consider this scenario.....you have a V8 racing engine, which will deliver a specific amount of heat at the hypothetical 650 hp produced.

    Let us say, you need a 4-Core radiator 'X' to cool the engine, keeping it at the optimum 94 deg C under both the best and worst conditions....the temperature is regulated by a regular thermostat.

    Now, ask yourself the question....adding a second radiator 'X', as well as fan, will that help at all?

    Substitute the 650 hp V8 racing engine with your CPU and the thermostat with the physical limitations in extracting the heat from the CPU core.....issues that we cannot overcome.....will adding the second rad improve matters at all? Provided the 1st radiator/waterloop was designed/sized correctly?

    The only real issue in this day and age, is the one of overclocking.

    Overclocking in itself doesn't significantly increase the GPU/CPU die temperatures, if it is pure MHZ we are talking.....what are causing problems is the indiscriminate use of vcore/vgpu to increase overclocking.....

    At the nano size of gates used in the manufacturing of modern processors, leakage current is a must, it generates a lot of heat, yet people just increase all the way, not realizing what they are doing....

    So, i guess there's a right and proper way to overclock, and to design a waterloop for that, and then the wrong way, requiring MOARRAD to cool the system, most of those extra clocks not doing anything significant seeing as PROCHOT is throttling the hell out of the CPU.....

    Not to add though, how many of the gaming systems out there will be running full bore 24/7?

    Not many, if any......i do 10 hours racing/practice sometimes, no issues....

    But yeah, rambling again, Beano.....blah...
     

Share This Page

Advertisement: