Stren's Titan Water Block Roundup

[stren]

Finally got the full write up and video done:

write up: http://www.xtremerigs.net/reviews/wa...p-coming-soon/

video: http://www.youtube.com/watch?v=sDyYliSo-Hw

Card became unstable at the operating point after 20ish mounts so I couldn't run as many data points as I'd like.

Bitspower, Alphacool and Phobya agreed to send blocks but I gave up on waiting.

The review contains AquaComptuer, EK, EVGA/Swiftech, Koolance, Watercool, XSPC

[LeoAndrade]

That's nice. Looking foward to see the results, specially koolance and aquacomputer blocks

[mbreslin]

Anecdotal but after many hours of benching/game playing with my max oc (1202mhz@1.212v) during the hottest day we've had so far with the ek blocks I got max 42c on 2 cards in parallel. Titan seems to run very cool and I can't wait to see just how low the best block in your review's temp is.

[stren]

Anecdotal but after many hours of benching/game playing with my max oc (1202mhz@1.212v) during the hottest day we've had so far with the ek blocks I got max 42c on 2 cards in parallel. Titan seems to run very cool and I can't wait to see just how low the best block in your review's temp is.

Yup I'm expecting most blocks will perform very well with good flow. We'll see though, I want to try some low flow testing too for those who run many gpus in parallel. I'll also try and compare as best I can vs air and an AIO cooler too.

[paulbagz]

Honestly didn't know half those manufactures were making Titty (lol) blocks

-PB

[tiborrr]

Alright! EK's model should be in your nest already, correct?

[Martinm210]

Look forward to your work, I just hope it ends better than Skinnees 480 roundup which resulted in a dead 480. Unlike a CPU it seems all the testing abuse on GPUs is less tolerated. The word "Loan" would bother me but hopefully the GPU's have gotten better since the 480 days. That has been my hesitation with trying it, have seen too many issues with GPU blocks having little issues here and there and just feel it's a bit risky to put a card through the ringer like that unless someone was going to sponsor potentially losing the card. I've killed enough motherboards on my own by now just with CPU testing, 2011 is the first to seem durable enough to handle it with the build in back plate. GPU i would be very cautious with. Maybe do some visual TIM and thermal pad contact checks before hard benching or something and also ensure standoffs avoid bending the pcb at all and verify backplates do not make any circuit contact somehow...etc.

Would be awesome to see some low flow data. With the AIO units and with triple parallel, I don't think it is unusual to see really low flow rates down in the .3gpm range.v With most DIY CPU blocks falling off at .5gpm with their more dense fin structure, I would think GPU blocks suffer more rapidly and the practice of triple parallel is something that should be questioned more. Not much GPU flow sensitivy data out there right now, so it would be a huge hole filled.

[.::NeO::.]

Awesome news Stren

Can't wait to see what Watercool wb will look and perform like.

Again, a big thank-you for all your hard work. You rock!

[stren]

Alright! EK's model should be in your nest already, correct?

It's on it's way, but it hasn't arrived yet I have a feeling Koolance might beat you to my door just because they have a 6000 mile advantage lol. Also Aquacomputer's is already in the country so that might show up today. I hadn't seen a picture of it here but a quick google gave me this:



I have to give it to aqua and watercool they really pull off the two tone metal look well. No one else has shipped yet though and it doesn't sound like most will be ready for another few weeks.

Honestly didn't know half those manufactures were making Titty (lol) blocks

-PB

Neither did I until I asked them, most of them have been pretty quiet about their blocks! Liquid Extasy would have participated too but they need a loaned Titan themselves to design it in the first place.

Look forward to your work, I just hope it ends better than Skinnees 480 roundup which resulted in a dead 480. Unlike a CPU it seems all the testing abuse on GPUs is less tolerated. The word "Loan" would bother me but hopefully the GPU's have gotten better since the 480 days. That has been my hesitation with trying it, have seen too many issues with GPU blocks having little issues here and there and just feel it's a bit risky to put a card through the ringer like that unless someone was going to sponsor potentially losing the card. I've killed enough motherboards on my own by now just with CPU testing, 2011 is the first to seem durable enough to handle it with the build in back plate. GPU i would be very cautious with. Maybe do some visual TIM and thermal pad contact checks before hard benching or something and also ensure standoffs avoid bending the pcb at all and verify backplates do not make any circuit contact somehow...etc.

Would be awesome to see some low flow data. With the AIO units and with triple parallel, I don't think it is unusual to see really low flow rates down in the .3gpm range.v With most DIY CPU blocks falling off at .5gpm with their more dense fin structure, I would think GPU blocks suffer more rapidly and the practice of triple parallel is something that should be questioned more. Not much GPU flow sensitivy data out there right now, so it would be a huge hole filled.

Yes I hope I don't break the card too - these are not cheap cards lol. I myself run triple parallel on my 580s and it has been on my mind when I was thinking about adding a fourth card. Now I've been thinking about splitting my cpu and gpu loop so it's also caused me to reevaluate the decision to go parallel. Before I valued my cpu temps more so I chose higher flow even if it compromised the GPU temps. Now I don't have to compomise so it'll be fun to find out what I should be doing Parallel seems to be way more common these days than the past and I've seen plenty of 4 way parallel setups with weaker pumps that worried me so that was definitely the inspiration behind the super low flow testing.

I may have to pick your brain on the best way to test these too as this is my first GPU roundup

Awesome news Stren

Can't wait to see what Watercool wb will look and perform like.

Again, a big thank-you for all your hard work. You rock!

Me too - I have something else coming from Watercool soon too that will be interesting too I haven't done any real work apart from lots of emails yet though!

[bds71]

...particularly EVGA who are loaning a Titan

i've always had really high opinion on EVGA. had a couple issues with some of their cards, and their customer service was absolutely great. good on 'em!! glad someone was able to get a card like this for testing - should be interesting to see what this thing can really do under water....

[Yaka]

heh cant wiat for the round up

[Church]

stren: slight offtopic, but for 4-way it's best to have semi-hybrid (2parallel-serial-2parallel) = you get both not too high pressure drop and more reasonable 1/2 of flow per card instead of 1/4. IIRC EK thought about this as well and has released such bridge. Well, can be done simply using fittings too.

[stren]

i've always had really high opinion on EVGA. had a couple issues with some of their cards, and their customer service was absolutely great. good on 'em!! glad someone was able to get a card like this for testing - should be interesting to see what this thing can really do under water....

+1

heh cant wiat for the round up

+1

stren: slight offtopic, but for 4-way it's best to have semi-hybrid (2parallel-serial-2parallel) = you get both not too high pressure drop and more reasonable 1/2 of flow per card instead of 1/4. IIRC EK thought about this as well and has released such bridge. Well, can be done simply using fittings too.

Yup I agree but for my older cards they didn't have the option it was parallel or serial I believe. Or use SLI adapters and do it yourself. I'm glad EK finally did it though. So I guess triple is probably the worst case now.

Also the Aquacomputer box just arrived, so expect photos tonight

[Church]

But old ones probably had that beautiful old EK bridge design . Still sad about them not being usable on new cards <sniff>
Still, it's good thing that you will test at lower pump settings too. For most vendor many-ways bridges it might be important nfo. BTW, if you will test only at 2 preset pump or flow settings, maybe it's worth one of blocks of pin design, one of microchannels design, and one of big channels design test at more flow testpoints, so that line chart can be drawn for estimation, and for different types of designs to see which designs are more low flow agnostic. Probably can be done once per one mount, as imho relative temp drop because of flow changes should be same for all mounts? I'm somewhat interested in what lowest flow might be for gpu blocks without overly temp worsening because of low flow exactly because too-many-ways parallelizing issue. I've heard that untill 0.25gpm you should be good, but it would be nice to see full very wide flow range even below that to draw up some generalised conclusions.

[Shoggy]

By the way: on your website you wrote "Aquacomputer aquagraFX Titan Full Cover Full Cover Block" which is wrong. It must be kryographics and I also think a full cover block which is fully covered again is just a bit over the top

edit: ahhh, even worse - please link to www.aquacomputer.de

[stren]

By the way: on your website you wrote "Aquacomputer aquagraFX Titan Full Cover Full Cover Block" which is wrong. It must be kryographics and I also think a full cover block which is fully covered again is just a bit over the top

edit: ahhh, even worse - please link to www.aquacomputer.de

Fixed

[musicfan]

Thanks stren. Subscribed.

[stren]

Alright I took some quick pics though I didn't unseal the block yet as there seemed little point until I have the card from EVGA!

Aqua also sent some parts for the pump/top/reservoir roundup that I'm also working on concurrently. I won't show you the rest of the pics of those parts here because I want to stay on topic, however while we're briefly off topic, I will post the pump/top/res stuff on XS in some form, I just haven't worked out exactly how I want to do it yet!



Yum! Big box seems fitting for a "titan" of a card!




Proof:



More proof:



What you get in the box:



Underneath of the vacuum sealed block:



TITAN:



The kryographics logo doesn't come out well in the photos, I think it's because of the bag that the block is in. It looks better in person:

[Elmy]

That block is fn sexy!

[Martinm210]

Been a while since I have done any GPU testing too. I know skinnee did some installing of dallas probes on the card itself to get good VRM temps. Not sure what the titan has onboard. I would tend to probably stick to using the kit contents TIM and pads as that is probably how most will use them, but I'm not sure about mounting numbers etc. You might only get one shot at it depending on what comes with he block. Haven't a clue on the stress test these days.

Still have my 570gtx on air. Last GPU I had on water was an 8800 and my son stil has one running.

As far as pump/loop. A separate loop might help isolate heat and you could certainly dial down pump speeds and even install a valve if need to really dial down flow rates, but there is always a desire to see actual CPU + GPU loop testing too. Obviously you cant test Quad SLI parallel on all the blocks though so flow rate testing will have to do.

If you could test down to .3GPM that would be close. I would start with one and find a pump power that matches something close to .3gpm, .9gpm, 1.5gpm and let the restriction vary actual flow a little. Performance at like pumping power and restriction is what I like to see, then the blocks restriction is automatically factored in a bit.

[stren]

Been a while since I have done any GPU testing too. I know skinnee did some installing of dallas probes on the card itself to get good VRM temps. Not sure what the titan has onboard. I would tend to probably stick to using the kit contents TIM and pads as that is probably how most will use them, but I'm not sure about mounting numbers etc. You might only get one shot at it depending on what comes with he block. Haven't a clue on the stress test these days.

Still have my 570gtx on air. Last GPU I had on water was an 8800 and my son stil has one running.

As far as pump/loop. A separate loop might help isolate heat and you could certainly dial down pump speeds and even install a valve if need to really dial down flow rates, but there is always a desire to see actual CPU + GPU loop testing too. Obviously you cant test Quad SLI parallel on all the blocks though so flow rate testing will have to do.

If you could test down to .3GPM that would be close. I would start with one and find a pump power that matches something close to .3gpm, .9gpm, 1.5gpm and let the restriction vary actual flow a little. Performance at like pumping power and restriction is what I like to see, then the blocks restriction is automatically factored in a bit.

Interesting. Yes my flow meter only goes down to ~0.3gpm anyway, so below that people will have to infer. My original plan was to keep it similar to the CPU test i.e. multiple mounts with MX2. You've scared me away from the multiple mounts a bit though lol. I'd do a gpu and cpu loop with my usual 560 rad with 2150rpm Gentle Typhoons. I'll unpower one of the MCP35x2 pumps in order to reduce flow. I'll use the dallas probes to measure the water temp and log with wintest. The GPU temps I'm not sure how I'll be logging yet, I guess I'll play around with titan and see what options I have. I've heard a lot about it throttling unless you use a modified bios, so that's one concern. The other issue is the best way to stress test. Back in the day it was always furmark but that's gone out of favor it seems. I could loop a benchmark like unigene valley constantly and average the temps over a decent window after letting the system warm up for long enough. I'll have to poke around in the titan threads to find out more I suspect.

[stren]

Koolance Titan:



Mini back plate included:



Accessories



Block:



Butt:



Butt with backplate:



Inside of backplate with marked areas for thermal pads for memory:



Details:

[stren]

EK Titan:





Accesorize it:



Explosion:



Nickel Plexi:



Butt:



Details:







Backplate:



Very simple and very few scews:





How they would sit:



With EK Blood Red coolant in it:

[Church]

Stren: btw, there is one big minus to choosing titan (apart from high cost if it gets damaged during extensive testing) for roundup. It simply is not the hottest card around, unlike 480 back then at skinneelab test time . It's 250W TDP is rather reasonable even among top cards, less the heat load = imho probably less temp difference among blocks, and even worst ones will cool good enough, no? gtx480 was nice worst case scenario (especially for testing universal gpu blocks. as in - if they are able to cool THAT, they will do for everything else thrown at them ).

BTW only looks wise i'm glad for EK's vote/thinkcell movement result, transparent clear block brings back FC bling many liked EK for. From presented blocks theirs looks best in my eyes. Pitty it can be well seen only in inverted atx cases.

[LeoAndrade]

Show us the internals of both koolance and aquacomputer blocks