Sign in with
Sign up | Sign in

Conclusion

GeForce GTX 285 On Water Cooling: Zotac's Infinity Edition
By

The Zotac GeForce GTX 285 Infinity Edition uses a lot of memory and moderate GPU overclocks to achieve notable performance gains compared to reference-spec cards. But the solution is most certain to grab the attention of performance fanatics who would have otherwise considered an overclocked GeForce GTX 285 and separate liquid-cooling water block.

Here's the rub: the card hasn't shown up yet at e-tail. As the availability story changes, we'll update this space to reflect Zotac's value-proposition for the enthusiasts looking to stack two or three of these in a high-performance SLI configuration, where heat is public enemy number one.

Initial estimates from Zotac peg the card around $519. Stock GeForce GTX 285s go for somewhere around $360 online. And the Danger Den water block sells for $145. In other words, buying pre-configured from Zotac will cost close to the sum of its parts, netting you a guaranteed overclock and two-year warranty (that's upgraded to "lifetime" coverage if you register the card immediately after purchasing it).

Our biggest reservation in outright recommending it over the competing model out there with an identical cooler is that the competition provides its card with a single-slot bracket. While we won't directly compare the two, since we don't have that other board on hand to test thermals, power, or performance, its single-slot mounting could open up three additional expansion slots when used in a 3-way SLI configuration. Zotac offers superior clock speeds, but buyers must weigh that advantage against any lost slot access.

Our previous 3-way SLI articles have shown good reasons to choose liquid cooling when multiple cards are installed back-to-front, but it’s up to the buyer to decide which liquid cooled solution is best for his or her needs.

Display all 71 comments.
This thread is closed for comments
  • -5 Hide
    acasel , March 30, 2009 7:17 AM
    It would be better if we could see how well the zotac overclocked card overclocked.. In other words pushed to the limits.
  • 3 Hide
    cangelini , March 30, 2009 7:47 AM
    Page 2, 5th paragraph down. It's there ;-)
  • 3 Hide
    Anonymous , March 30, 2009 8:29 AM
    54 deg C above ambient for W/C 285GTX makes no sense.
    The temperature of my 8800GTS 512 running GPU at 800Mhz (from 650MHz) + W/C southbridge and 2 X 120 Radiator. My rig maxes out at 27 Deg C above ambient.
  • 5 Hide
    Crashman , March 30, 2009 9:45 AM
    gothminst3r54 deg C above ambient for W/C 285GTX makes no sense. The temperature of my 8800GTS 512 running GPU at 800Mhz (from 650MHz) + W/C southbridge and 2 X 120 Radiator. My rig maxes out at 27 Deg C above ambient.


    Perhaps the chosen liquid cooling system doesn't have enough pressure.
  • -3 Hide
    MikePHD , March 30, 2009 12:38 PM
    Wow! Was this supposed to fool us into thinking they knew anything about watercooling... this arcticle is a joke. Why does the water cooled card not get overclocked, while on the same token, the "air" card gets overclocked the hell out of.

    This is obviously not the limit of water cooling, it's either the product of a very lame product, or the testers ignorance of water setups.
  • 4 Hide
    neiroatopelcc , March 30, 2009 12:39 PM
    I wonder why you bothered doing this article? As we can all agree this card only makes sense if you plan on putting it in a very tight place, or have 2-3 of them in a chassis. So why didn't you request 3 cards and compare it to a standard 3 card setup? The interesting thing would be to see if it actually helps enough to make a 3sli setup doable longterm. As it is, all the article shows is that the watercooling used is inadequate for the task it was given.
  • 5 Hide
    rubix_1011 , March 30, 2009 1:01 PM
    It's really sad that you chose to use the Big Water kit to push the cooling on the GPU instead of incorporating the CPU loop with the GPU using the far superior 655 pump and just adding an additional radiator instead. Your temps of 56+ C are way to high for a watercooled component...you shouldn't see more than 40C at load, especially with that waterblock. You are crippling the temps and the basis of your entire article by using sub-par components in your tests. Anyone who watercools knows that today's GPUs need at minimum a 220 or similar surface area radiator to expend the heat being produced...same goes for your i7 CPU...but it looks like you have at least a 320 on it. You should have run a series loop and just added an additional 120 or 220 radiator/fans and been done with it. You'd have had much happier temp results.
  • 3 Hide
    grevaeg , March 30, 2009 3:03 PM
    Would make alot more sense to put in a decent watercooling setup... this test doesnt give anyone ANY idea how a watercooling kit can perform...

    Currently running two q9550 and a 4870+ a 4870x2 on my loop.. never gets anything near ur single rad setup on a single card and cpu. Have never seen the temp going past 30c at home.
  • 5 Hide
    Kaldor , March 30, 2009 4:00 PM
    Good card and cooler. Bad article.

    You guys need to invest in a decent radiator and pump if you want to test liquid cooled components. A D5 and a MCR 220 isnt that expensive.

    At least Zotac used a good cooler. Much better than the garbage EVGA uses on their cards. Al+Cu=lose.
  • 1 Hide
    marraco , March 30, 2009 4:46 PM
    A bit off topic:

    The first picture shows a case fan stating to get dusty.

    The fan flaps already are white-brown.

    On my city, fans become very dusty really fast. They end occluding radiators, and air vents. I even cutted out all the metal grid on the case, to reduce dust retention. but still it is catched on the metal borders.
    I wish there where something to do about it. maybe some liquid or painting, or some trick to repel dust, or make it less adhesive.

    (my father have a German shepper dog, and the last time I opened his case, it was completely full of dog hair.... fun)

    If somebody read this, and know a fixing (not involving dogs), please, post.
  • 0 Hide
    marraco , March 30, 2009 5:00 PM
    I already occluded the unused memory slots with paper, because the output of the CPU radiator fills them with dust.

    Other thing I hate, is when I clean the fan cutting edge. It ends all scratched and dented. I don't know why. The rest of the flaps are not eroded, so why the flaps get dented?

    I suspect that they get dented at colliding with some heavier particles, and then start catching the dust, so maybe the solution involves metallic flaps.
  • 0 Hide
    icepick314 , March 30, 2009 5:01 PM
    "Our biggest reservation in outright recommending it over the competing model out there with an identical cooler is that the competition provides its card with a single-slot bracket."

    whoelse overclocks and adds watercooling with single-slot card?
  • 4 Hide
    funkjunky , March 30, 2009 5:20 PM
    ya this article is messed up. He starts the article by saying watercooling solutions are more importantly necessary with multiple video cards, yet never tests this setup.

    Also I've never done water cooling before, but i would take these other comments word over the articles, that water cooling can keep temperatures a lot cooler.

    A question. Does the strength of the pump havea significant effect on the cooling of the product? I would assume their is a large diminishing return after a certain amount of water pressure... hmmm thinking about it maybe not.

    What were you thinking Thomas Soderstrom?
  • 1 Hide
    Annisman , March 30, 2009 5:37 PM
    kinda sick of articles like this, wish Second Take with Ben and Rob was still around...
  • 0 Hide
    baddad , March 30, 2009 6:09 PM
    Water cooling my GPU’s drop my idle temps from 70*C to 45*C and Under load from 90*C to 60*C.
    I’m not understanding how 3 OC 285’s can be out preformed by 3 stock 8800 GTX’s. Am I missing something here? Below is my system and the Bench marks I’ve run on it.

    Thermaltake Tower Case Armor+
    Thermaltake 1000W PSU
    ASUS Striker II Extreme
    Intel C2D Q9450 OC 3.2GHz
    Corsair 4x2GB 1600 DDR3 Set to 1333 FBS 9,9,9,24 t2
    3 EVGA 8800GTX 768MB Two SLI one PhysX
    1 Seagate 250GB 32MB Cache SATA Boot Drive Vista 64 Bit
    5 Seagate 250GB 32MB Cache SATA Drives Raid 5 1TB
    1 Termaltake external E-SATA enclosure with Seagate 250GB 32MB Cache for Ubuntu
    1 Seagate FreeAgent Extreme 1TB E-SATA to back up the Raid set.
    LITE-ON SATA DVD-CD Burner
    Gyration Ultra Cordless RF Mouse and Keyboard
    3 Koolance 282GXT water blocks
    Koolance 330 CPU cooler
    Koolance External EHX-1020SL 3 fan radiator
    Koolance Internal RP-1000 reservoir and pump
    24” Acer Monitor
    Turtle Beach Ear Force 5.1 Headphones
    Momo Racing Force Feedback Wheel and Racing Pedals
    Ideazon Fang Gamepad
    Logitech Dual Action Gamepad
    Patent Pending Computer Gaming Table

    I’ll start with the FutureMark Bench marks all in default settings, and very depending on the graphics driver your using at the time.

    3DMark03 71647
    3DMark05 19292
    3DMark06 16322
    3DVantage 20520

    All game benches are in 1920x1200 max setting except for Crysis which I will start with. Crysis, Crysis Warhead, Lost Coast, Lost Planet, UT 2004, UT3, Farcry, Farcry 2, are 64 bit; I don’t know if which other games come in a 64 bit version. I tried to use bench mark software written for the game I ran bench marks for and for others I used Fraps to estimate an average.
    Crysis Very High 8xAA GPU 27.01
    4xAA 29.72
    2xAA 33.14
    0xAA 38.49
    Crysis Very High 0xAA CPU 38.02
    Crysis Very High 0xAA 34.98 Harbor
    Crysis Warhead 0xAA 45.00 Enthusiasts

    Lost Coast 134.04
    HL2 150.0
    Lost Planet Snow 86.7 Cave 86.2
    UT3 238.0
    Prey 115
    Call to Juarez 31.3
    Fear 174
    Fear XP 155
    Fear PM 179
    SS2 166.3
    ETQW Capt at 60
    Quake 4 Capt at 60
    Doom 3 Capt at 60
    Fallout 3 Capt at 60
    Dead Space Capt at 30
    The next games I could only estimate using fraps.
    Bioshock 130
    Time Shift 130
    Fear 2 130
    Jericho 40
    Portal 200
    MassEffect 60
    Call of Duty 2 250
    Call of Duty 4 150

    The only game I play with AA off is Crysis and I really don’t notice any deferent’s when I’m playing it. I think it will take another huge leap ahead from Nvidia, as with the 8800 GTX, to over come AA and higher resolutions playability with Crysis. I think this machine is good for another five years or more.
  • -4 Hide
    Crashman , March 30, 2009 7:01 PM
    MikePHDWow! Was this supposed to fool us into thinking they knew anything about watercooling... this arcticle is a joke. Why does the water cooled card not get overclocked, while on the same token, the "air" card gets overclocked the hell out of. This is obviously not the limit of water cooling, it's either the product of a very lame product, or the testers ignorance of water setups.


    Article clearly states that both cards were overclocked to their respective limits. So what does that say about your comments?
  • 1 Hide
    Crashman , March 30, 2009 7:08 PM
    icepick314"Our biggest reservation in outright recommending it over the competing model out there with an identical cooler is that the competition provides its card with a single-slot bracket."whoelse overclocks and adds watercooling with single-slot card?


    The Zotac GTX 285 Infinity IS a single-slot card. It happens to be a single-slot card with a double-slot bracket.

    Zotac sells a single-slot water cooled GTX-285 with a double slot bracket, so if you use 3 of them you loose six expansion slots. BFG sells the same card and cooler combo with a single-slot bracket, so if you use three of them you still have 3 or 4 expansion slots left.

    If both cards are otherwise equal, BFG wins giving you space to put your OTHER cards.

    But it looks like both cards may not be otherwise equal, because Zotac gets some awesome memory overclocks from its card.
  • 0 Hide
    Crashman , March 30, 2009 7:36 PM
    rubix_1011It's really sad that you chose to use the Big Water kit to push the cooling on the GPU instead of incorporating the CPU loop with the GPU using the far superior 655 pump and just adding an additional radiator instead. Your temps of 56+ C are way to high for a watercooled component...you shouldn't see more than 40C at load, especially with that waterblock. You are crippling the temps and the basis of your entire article by using sub-par components in your tests. Anyone who watercools knows that today's GPUs need at minimum a 220 or similar surface area radiator to expend the heat being produced...same goes for your i7 CPU...but it looks like you have at least a 320 on it. You should have run a series loop and just added an additional 120 or 220 radiator/fans and been done with it. You'd have had much happier temp results.


    Unlike BFG, Zotac does not include 1/2" barbs with its card. The 3/8" barbs would have put a restriction on the liquid cooling system used for the CPU and left-over Swiftech 1/2" barbs had a different thread, so a separate liquid cooler was used.

    Concerning the GPU clocks being stuck, a card that operates perfectly at its shipped speed (722 MHz GPU, 1584 Shader) but cannot operate stably at the next step up, that normally indicates a "stuck" card regardless of whether or not the GPU temp reached into the 70's. That is to say, you'd think the card would go at least 732 MHz if it wasn't stuck, but it didn't. Having a card's factory overclock and a separate liquid cooler's limits be an exact match would be a coincidence too rare to consider.
  • 5 Hide
    rubix_1011 , March 30, 2009 8:11 PM
    I understand your defense, but to counter your reply, anyone building a water cooled system would simply have just purchased/used the correct sized barbs with threads for use. A quick purchase at most online stores should get you the components you need...and also allow you to criticize the vendor for omitting barbs for a large majority of watercoolers out there. The 3/8 barbs might have caused some flow restriction, but I bet far less of a performance hit than using a substantially weaker pump and radiator combo to power the cooling system for the GPU.
  • 1 Hide
    avatar_raq , March 30, 2009 8:44 PM
    The temperatures on the "reviewed" card are a joke. Why bother writing this article when you did't show us the advantages of water cooling over the reference air cooler (not even a mention of the noise levels!!), and you did't test it in SLI, which -ironically- you made all the readers think about it when reading the 1st page!!
    You also missed mentioning the fan speed of the OCed air cooled card, was it left to the auto settings? or crancked up? If so, what noise the infamous reference cooler produced?
    I can't deny it. This substandard article feels out of place when compared to other TH articles.
Display more comments