Printable View
The 480 air cooler is made so the card almost reaches thermal limits at stock clocks.. i doubt there's any headroom on air ;)
That's the problem, so far with the cards i've had, on straight watercooling the temperature didnt change the OC limit (again as long as the card doesnt reach thermal problems). And from what i have seen that limit only really moves at sub zero temps...Quote:
Originally Posted by mk-ultra
let's go ahead and lock this thread down. not going anywhere anytime soon...
well or course sub-zero is another world, but on mine the other i was playing dirt2 with my card OCed as usual, suddenly the game crash. looked at the temps they were way high, turns out i forgot to set my rad fans at 12v, i never had a crash before of after that.Quote:
Originally Posted by gmat
especially with your comments...Quote:
Originally Posted by resident1509
Well yeah as i said, if you reach thermal limits that's entirely different (and maybe the 480's are hot enough beasts to do that ?), and that would of course fall back to the same priorities as for CPUs - but with watercooling i never had such problems on a GPU even when torturing them with furmark and such. I found out CPU's are way much more sensitive to temperatures (and that i can reach the thermal limit on my i7)Quote:
Originally Posted by mk-ultra
Hey skinnee,
Do you think the chip that fried was slightly depressed?
Seem like a common issue with trying to cool mosfets. They are small little chips that don't usually sit very flat and hard to cool with anything but thermal pads that have some ability to fill the gaps. It kind of looks like the TIM spread picture you posted back a few pages seems to hint that could have been part of it..?
Sorry for the loss..:(
Sorry to hear the 480 died :(. When will the funeral be? I'd like to pay my last respects.
The chip was well beyond depression, the PCB did not like the role of rented mule it wound up in and committed suicide. :D I don't think the fets were sagging, and most of the TIM print shots show 'good enough' contact and better with tape.
I went through and checked all the cards for contact on the VRM's before running load tests, that with a generous dollop of TIM and the height variation did not play much if any of a role in the GPU seppuku... at least I am not pointing it to that. There had been physical stress on the cap for those fets from the majority of blocks (notice the hot glue for added durability), add that into the rigors of thermal testing and that area was just prime for failure. GPU's are much more 'delicate', and separating the card from the block is a violent process. I can't recommended lots of remounts for a GPU block.
http://skinneelabs.com/assets/images...omtimprint.jpg http://skinneelabs.com/assets/images...omTIMPrint.jpg
But yeah, losses happen, I'm just glad all of the important tests were complete when it happened.
maybe people envied you for having all these Stuff :D so the gtx 480 blew up,anyway Sorry for your lossQuote:
Originally Posted by skinnee
blew up because of envy? :) (trying to imagine vax voodoo doll in form of 480, full of needles and partially burnt by candle somewhere in vrm area) :D
Sorry for Yer loss skinnee, But all things considered, You at least have the data saved.
Judging by the URLs that you're using to show off those images with, it looks like the article is coming together quite well.
i was just trying to tell a joke but i think i went mad :rofl:Quote:
Originally Posted by churchy
Yeah they are pretty flimsy, I remember how bowed my 8800gtx got too. I've also had at least one older video card die from what I think was just static electricity. I guess we never know with electronics..just had a $4,000 GPS unit crap out on me at work a few weeks ago...bad digital board was almost as much as just buying a new unit...and it was supposed to be a really rugged unit. I guess I give up on having confidence in electronic parts. I've had $20 dvd players last for ages and really high end stuff DOA. Who knows...:)Quote:
Originally Posted by skinnee
Anyhow I'm sure all that data you're putting together will be greatly appreciated. Thanks!!
Skinnee what kind of time frame are we still to expect before we can see your results for ourselves? I know your busy..I would accept any answer within the space time continuum :)
I'm trying to wrap it all up this weekend, I want to be done as well. :)
a little piece of what I just finished...
http://skinneelabs.com/assets/images...X480_pdrop.png
TEASE :cool: :confused::ROTF:
Looks good skinnee. :up:Quote:
Originally Posted by skinnee
the graph is so purdy! can't wait! the suspense is killer!
Hm. . .and performance is inverse of flow rate. :stick::para::wasntme:Quote:
Originally Posted by skinnee
If I'm reading this chart correctly the EK water block has 16 psi pressure drop while the Danger Den is only 6 psi pressure drop, that's huge when considering when HWLabs pressure tests their radiators at max continuous pressure rating 24.7psi...? Huh? Something doesn't add up here because we just added what looks like a lot of restriction!Quote:
Originally Posted by skinnee
-Systemlord
at what flow rate? 16psi pressure drop is at like 4.8gpm. Rads in general have very low pressure drop, but when comparing for restrictiveness, both must have same flow rate.
Oh alright I understand, I wasn't paying any attention to flow rates. :shakes: At 1.5GPM the difference is about 1 psi difference between the DD and EK blocks, nobody runs at 5 GPM, duh! I stongly believe that the wavy pattern of fins on the EK blocks GPU area is the contributor to the restriction, this is shown on the AQ fin area restricting the flow even further! :eek: What flow rates are considered attainable in an ideal low restrictive loop? Picture below shows the AQ's restrictive GPU fin area!Quote:
Originally Posted by Boogerlad
-Systemlord
http://img810.imageshack.us/img810/8124/68443353.jpg
http://img31.imageshack.us/img31/338/32157760.jpg
aquacomputer's beautiful block lost a round here
Interesting graph :) .... Still waiting patiently :)
On the restrictive side yes, but how well does it actually cool...? Soon time will tell, and I thought my HWLabs GTX480 Radiator was on the restrictive side! I will not be adding anymore restrictive blocks to my soon to be loops!Quote:
Originally Posted by mk-ultra
-Systemlord
that's why i said it loose a ROUND
well I guess if I ever went quad sli I will be needing a few more pumps lol.
Put the GPUs in parallel :cool:Quote:
Originally Posted by zeroibis
I'll be cooling three GTX 480s with three of the winning block (or reasonably-priced block within 1-2C or so of the winner), setup is:
MicroRes > MCP655B > XSPC RX360 > GTX 480s > XSPC RX360 > Repeat
Does anyone see any problems with restriction with this setup?
Put the GPUs in parallel and even most restrictive blocks won't mean much in terms of overall restriction. Not that it really matters since your GPUs are the only things in the loop anyway. :shrug:
Yes, sir, was looking at those. I have them in my favorites as possibilities.Quote:
Originally Posted by Alexandr0s
Skinnee, OT question, but hows the fixing of the layout of the Skinneelabs site going?
Chrome or Webkit browser User? :shakes:
It seems like you're not a chrome fan :rolleyes:?
Hey, Skinnee, I figure the EK and Danger Den blocks are in the top, and about 1-3C close to each other, so, I went ahead got 3 of the cheaper. :D
Not a bad decision, right?
Just the extra work needed to be Webkit compatible too. Plowing through saving some charts and then I'll take a look at it... hopefully its a quicker fix than I am estimating.Quote:
Originally Posted by Alexandr0s
You'll be fine. :)Quote:
Originally Posted by UrbanSmooth
What I thought you needed them in serial so it looks cool :rolleyes: ;) lolQuote:
Originally Posted by Vapor
I can't remember the last time Danger Den made a block that actually performed as well as its competitors. :/Quote:
Originally Posted by UrbanSmooth
900 watts under load on that rad? Hope your not wanting quiet. You should go bigger on the rad or get a Black Ice GTX and forget quiet.:up:Quote:
Originally Posted by UrbanSmooth
No flow restriction worries. I see your temps climbing to silly levels though.
I was hoping that two of those radiators would be fine.Quote:
Originally Posted by Conumdrum
2x rx360 is perfectly fine for 3x480 IMO
I certainly agree with that.Quote:
Originally Posted by mk-ultra
I think a single 3x120 rad is fine for three GTX480s :sofa:
Not ideal, but acceptable.
well a gtx 360 with a san-ace push pull setup could cool those to some pretty acceptable level i would say :DQuote:
Originally Posted by Vapor
Uh oh, I have 2 aquacomputer blocks on order that haven't shipped yet. Can't wait to see the results.
Aquacomputer's FC-GPU blocks traditionally have performed quite well, just not with this level of restriction. Should be more than acceptable.
Single loop of course, seems if one is going with three GTX 480's external radiator mount seems ideal. There's not many cases that hold three 360 radiators, the third would be for the CPU loop.Quote:
Originally Posted by mk-ultra
Mountain Mods. :DQuote:
Originally Posted by Systemlord
I second that.Quote:
Originally Posted by Vapor
Yes a triple does in fact work just fine. I have a HWLabs GTX 480 but have yet to swap it out with my PA 120.3 yet in part because I don't think my temps will get much better. With my cards at 850/2100 at stock voltage they idle at 33C and fully loaded never get over 55C.Quote:
Originally Posted by Vapor
Can you explain why you proclaim lower to be better? This seems like an assumption on your part.Quote:
Originally Posted by skinnee
In a way, lower could not necessarily mean 'better'... those of you gifted with endless pumping power care less about restriction then those on a single or dual pumps.Quote:
Originally Posted by resident1509
With any fixed amount of pumping power, lower restriction equates to higher flowrate.
That's incredible temps for three overclocked GTX 480's pushing more than 250 Watts per card! :eek: Have you tripped any circuit breakers with your Silverstone 1500W power supply?Quote:
Originally Posted by Section8
-Systemlord
Funny you should ask. It kept tripping the 15A circuit breaker I had in the breaker box so I just made a trip to the home depot and bought a bunch of supplies to beef up the wiring going to my office including a 20A circuit breaker.Quote:
Originally Posted by Systemlord
The are many reviewers and people that have encountered the same issues tripping their 15A circuit breakers, most American homes only have 15A circuit breakers in the bedrooms, studies and/or dens. It's more common than you might think.Quote:
Originally Posted by Section8
-Systemlord
So after all there are some upsides living in the old world ;)Quote:
Originally Posted by Systemlord
Quote:
Originally Posted by skinnee
Skinnee, I am sure that the above question echoes among many other users.Quote:
Originally Posted by Systemlord
May I suggest that you superimpose the PQ curves of the 350/355 and 655 on your graph? This would give users an accurate idea as to the flow performance enveloppe. You could do this relatively painlessly with photoshop. If you are short of time, and with your permission, We'll be happy to do this for you.
This is why I am looking at buying a 2000v+ bbu b/c I know that in the future I am going to need 1500W+ support anyways. Now if only it was not 120lbs...
Good to know the temps for the triple sli, I bet you get some great scores and fps on that beast!
Yeah, system flow rate is one of those topics that comes up continually, and I have loop flow rate numbers the data table for each block. Every block stayed above the 1GPM mark, which for me is the low side of where I personally aim to keep flow rates.Quote:
Originally Posted by gabe
Yeah, I'll get another chart with 355 and 655 PQ overlayed... but don't have 350 numbers. I'm on my laptop atm, and the Excel workbook just kills this machine with all of the linked workbooks.
if that was true, the bitspower block got second.Quote:
Originally Posted by Waterlogged
If thats true, then 2012 is really coming to an end.
I'm running 3 GTX470's and a GTX480 (so 4 Fermi's alltogether) off one PA120.3 temps top 55/60°C with ambient around 30°C
Thats 3 Aquagrafx and 1 EK block.
Great! adding the pump PQ curves will really give an instantaneous view to ppl who are familiar with graphs. Thanks Cam, and sorry if it's more work for you, but as I said, if there is anything you need, we'll be happy to lend a hand.Quote:
Originally Posted by skinnee
with Dynamic Head Pressure from the D5-S5 with EK V2 Top (Top Outlet) and the DDC3.2 with XSPC V3 top.
http://skinneelabs.com/assets/images...480_pdrop2.png
I posted that to try and stir up a bit of trouble, why is it your the only one to bite? :rofl::p:Quote:
Originally Posted by NaeKuh
I don't know where to begin to understand the two gray lines? :confused: Neither pumps can go past 2.3GPM so I'm :confused:Quote:
Originally Posted by skinnee
-Systemlord
Where each gray pump curve meets each respective pressure drop curve, that will be your approximate flow rate for that water block and pump setup.Quote:
Originally Posted by Systemlord
For example, where the DDC3.2 w/ XSPC V3 top curve meets the the EK FC480GTX curve, you get an approximate flow rate of 2.5GPM.
Also, the pumps can and will go past 2.3GPM with the appropriate pump top.
I don't know how the DDC3.2 18W with XSPC V3 top compares to a DDC3.25 18W with an EK X-Top V2, but is the flow rate for just the VGA block itself or a full system with rad, res etc? Thanks for the explanation. :)Quote:
Originally Posted by ilal2ielli
-Systemlord
the DDC3.2 is 18w as well afaik, i think you're thinking of the 10w DDC3.1. I think the flow rate/pressure drop is just for the block itself...
I'm dying over here skinnee! I really need those results... I'm about to pull the trigger on a koolance block and need to know whether I'm better off buying a DD instead.
:up:Quote:
Originally Posted by ilal2ielli
That would be just the block and pump. You're going to have additional restriction in your loop with a radiator, tubing and so on... but blocks are the most restrictive component in your loop.Quote:
Originally Posted by Systemlord
For instance, The PQ chart shows the EK block at 2.5GPM with a D5-S5/EK V2 Top, but our loop flow rate for our test loop was 1.74GPM with radiator, tubing, and QDC's. Each component brings restriction to the loop.
Oh yeah, and we used a DDC3.2 for the test loop, not a D5. :)
so far . . .Danger Den FTW?
It's nice to know that the blocks are the most restrictive component because some of those blocks are quite restrictive, makes me rethink my purchase of my 4x120 GTX480 radiator being on the restrictive side!Quote:
Originally Posted by skinnee
Looking at the flowcharts, I'd say hell yeah. Now let's just wait calmy for Skinnee to give us the final results....Quote:
Originally Posted by DarthBeavis
Aw hell, I can't wait! Skinnee, we needz them :eek::D.
the flow shows flow....:D temps show temps..
The DDC 3.25 is actually 20w max according to Laing's specs, not 18w as the resellers list.Quote:
Originally Posted by Systemlord
Take what we can get. We can call Danger Den Flo ridah (which I like cause I have to fly to Miami in a couple weeks for the sole purpose of doing an insane water-cooled build).Quote:
Originally Posted by justin.kerr
triple pun... D'OH! :rofl:Quote:
Originally Posted by DarthBeavis
That looks great!Quote:
Originally Posted by skinnee
You might want to clarify for the benefit of casual readers that the DDC3.2 with XSPC V3 top is a dual pump.
Also, it would be interesting to show where a single DDC falls.
Thanks for the hard work!
DDC3.2 + XPSC V3 top is single pump :)
http://www.performance-pcs.com/catal...ducts_id=22700
Quote:
Originally Posted by gabe
I don't think that's the dual DDC XSPC top, looks to be the single from the dual DDC article:
http://www.skinneelabs.com/assets/im...ual_pq-psi.jpg
ahh, I see.. that's the "unreviewed" V3 top from the dual top review that got me confused. thanks for clariyfing.Quote:
Originally Posted by Vapor
Hey now... don't make me bring up the PWM pumps. :stick:Quote:
Originally Posted by gabe
Pump top reviews are changing on this next go round... less fancy review, more data driven which should mean faster turn around. I take too long.
Just some Qs if you don't mind answering about that:Quote:
Originally Posted by skinnee
What exact aspects will change?
Will you compare a topped DDC 3.2 to a topped DDC 3.25 using all the tops you have?
How did the last roundup not turn out consistent to European reviews that all showed EK's V2 to be better than XSPC's tops, and do you expect that to change?
And a more on topic question; when should we see the GTX480 blocks roundup get published?
Based on all the testing I've seen, EK V2 top is only better for max flow, but the PQ curve isn't as good as the XSPC V3 in the important/usable portion of of the curve.
On top of that, in my own personal use (with flowmeters), the XSPC V3 is ahead of the EK V2, by a noticeable amount.
Are there accurate electronic flow meters that you can attach to your loops so you know exactly what flow rates you are getting? I've never heard of a XSPC V3 yet.Quote:
Originally Posted by Vapor
-Systemlord
Sure, sample one from Aquacomputer:Quote:
Originally Posted by Systemlord
Click
haha, no I wouldn't do that to you :p:Quote:
Originally Posted by skinnee
...but I am still confused by the DDC curves because the curves in the link that Eric posted above (http://www.performance-pcs.com/catal...ducts_id=22700) do not match those of Martin's shootoot http://martin.skinneelabs.com/DDC32PumpTopTesting.html
#1 Pump Top Test comparision
http://martin.skinneelabs.com/img/DD...omparison1.png
#2 Performance PC
http://www.xspc.biz/DDC32PumpTopsComparison1.png
It looks like you are using the #1 set, right? and equally important, why the difference? maybe Martin could clarify?
Completely different pumps, newer revision of the top and completely different test gear, there are going to be differences. I've even retested the same RD-30 as Martin, my amp draw and dynamic head were lower.
Cross comparing benches...
when will the good bread crumbs start falling skinnee?
Your shedding your results like a trail... we want the main course!
Thanks for clarifying. So if I understand correctly, you are using your own data? no offense to Martin, but I prefer that, as I verified your data before, and found it to coincide with ours.Quote:
Originally Posted by skinnee
There must be a point where flow rates (example 2 GPM or more) no longer play a role in water temperatures, where is that point of diminishing returns?
-Systemlord
I suspect its either a difference in the pump motor/impeller and or a difference in testbed t fittings or the manometers. I did all my testing on the same exact pump motor/impeller to ensure good relative data, but it wouldn't surprise me to see some varience in the performance levels between pumps. I saw some of that with different D5s. I also made my own 1/2" copper to 1/8" brass fittings to connect my manomter to. Theretically and pressure errors should cancel out if measuring pressure difference, but there could be some influences we are missing. Also calibration of the manometers..probably some differences there too. I never had mine calibrated..mine could be reading highfor all I know..
Bottom line..different test beds....we can hold precision and have very good comparative data on the same test bed, but I we don't know how accurate the testing is without some sort of check. Soo, results from two different test beds shouldn't be combined or compared..you would need standardized proceedures, equipment, and calibration efforts which we don't have nor is there any interest to do so...we already donate our free time and paid for the equipment.
It would take a manufacturer to develop a standardized testing specification to get to a cross test bed level of testing accuracy along with precision. As hobby testers...were making up the specs as we go and constantly changing them...no incentive for us to do otherwise.
IMHO the same thing could be done for all of water cooling testing including cpu/gpu c/w testing, but it would take several manufacturers to come to some agreements on the testing specification.
No one has stepped up to that plate yet..;):shrug:
Hopefully one day..:D
Yeah, all my own data.Quote:
Originally Posted by gabe
@ Naekuh... be happy with crumbs, main course takes a while due to my poor planning.
Skinnee, when you publish the results, do you plan on holding an hour-long primetime press conference? :D
Disregard..I'm slow.:DQuote:
Originally Posted by gabe
Hello Martin,Quote:
Originally Posted by Martinm210
At least one of the biggies would have done it but will not provide the results because of impact on the bottom line.
Wes
yes, a long time ago. Note the MCP655-B label stating P4.Quote:
Originally Posted by Martinm210
http://www.swiftnets.com/assets/imag...55-800X500.gif
Note: It was unclear to me that these two sets of curves were performed on different test beds, nor was I aware of the fact that Cam used his own data. What confused me was Eric's reference to the curves published at Performance PC.
I see, sorry hadn't checked in a while. I still think the lines are too smooth/simplified but I'm picky that way, don't really matter in the scale of things.Quote:
Originally Posted by gabe
Maybe adding another note to the max pressure written spec would help emphasize it. Just one that burned me when I first started water cooling and I can't let it go...Ill get over it..someday..:):up:
I was referencing the product to show it was a single pump top, not so much the curves, sorry for the confusion :p:
I feel like your avatar reading this thread at this point. I make sure to refresh this thread every night before I sleep and every morning when I wake up and still nothing, ahahahahhhhQuote:
Originally Posted by NaeKuh
I require results or there will be more closed space....
FWIW, this is just the working thread, I suspect results will be in a different one :lol: