A powerful architecture
NVIDIA’s announced the TITAN Z in March and finally put it on sale in late May. We got our hands on one and put it through the paces to see how a $3000 graphics card performs.
In March of this year, NVIDIA announced the GeForce GTX Titan Z at its GPU Technology Conference. It was touted as the world's fastest graphics card with its pair of full GK110 GPUs but it came with an equally stunning price of $2999. NVIDIA claimed it would be available by the end of April for gamers and CUDA developers to purchase but it was pushed back slightly and released at the very end of May, going on sale for the promised price of $2999.
The specifications of GTX Titan Z are damned impressive – 5,760 CUDA cores, 12GB of total graphics memory, 8.1 TFLOPs of peak compute performance. But something happened between the announcement and product release that perhaps NVIDIA hadn't accounted for. AMD's Radeon R9 295X2, a dual-GPU card with full-speed Hawaii chips on-board, was released at $1499. I think it's fair to say that AMD took some chances that NVIDIA was surprised to see them take, including going the route of a self-contained water cooler and blowing past the PCI Express recommended power limits to offer a ~500 watt graphics card. The R9 295X2 was damned fast and I think it caught NVIDIA a bit off-guard.
As a result, the GeForce GTX Titan Z release was a bit quieter than most of us expected. Yes, the Titan Black card was released without sampling the gaming media but that was nearly a mirror of the GeForce GTX 780 Ti, just with a larger frame buffer and the performance of that GPU was well known. For NVIDIA to release a flagship dual-GPU graphics cards, admittedly the most expensive one I have ever seen with the GeForce brand on it, and NOT send out samples, was telling.
NVIDIA is adamant though that the primary target of the Titan Z is not just gamers but the CUDA developer that needs the most performance possible in as small of a space as possible. For that specific user, one that doesn't quite have the income to invest in a lot of Tesla hardware but wants to be able to develop and use CUDA applications with a significant amount of horsepower, the Titan Z fits the bill perfectly.
Still, the company was touting the Titan Z as "offering supercomputer class performance to enthusiast gamers" and telling gamers in launch videos that the Titan Z is the "fastest graphics card ever built" and that it was "built for gamers." So, interest peaked, we decided to review the GeForce GTX Titan Z.
The GeForce GTX TITAN Z Graphics Card
Cost and performance not withstanding, the GeForce GTX Titan Z is an absolutely stunning looking graphics card. The industrial design started with the GeForce GTX 690 (the last dual-GPU card NVIDIA released) and continued with the GTX 780 and Titan family, lives on with the Titan Z.
The all metal finish looks good and stands up to abuse, keeping that PCB straight even with the heft of the heatsink. There is only a single fan on the Titan Z, center mounted, with a large heatsink covering both GPUs on opposite sides. The GeForce logo up top illuminates, as we have seen on all similar designs, which adds a nice touch.
Titan Z cards will take up three slots in your system which also means that many mini-ITX cases will not be able to accept installation of the Titan Z. The R9 295X2 has a similar problem with its length and of course the added radiator. The NVIDIA Titan Z will likely fit in cases built for micro-ATX cases though without issue.
The rear of the card has a very thick and strong back plate that is used both has a heatsink and for design. It protects the back side components while adding strength to the unit for installation and even shipment.
Output options on the Titan Z are identical to other reference NVIDIA designs and include a pair of dual-link DVI connections, a full size HDMI and a full size DisplayPort. There is plenty of room for air exhaust as well which is important to get as much movement OUT of the chassis as possible.
I do think it's odd that NVIDIA chose to continue with just a single DP connection as this limits the support for 4K panels to one per Titan Z. If you were planning on picking up a set of three of the new ASUS PB287Q monitors for example, you need one NVIDIA card per monitor!
With a TDP of 375 watts, well under that of the R9 295X2, the NVIDIA GeForce GTX Titan Z requires a pair of 8-pin power connections. Now, the R9 295X2 also only required a pair of 8-pin connections but it was drastically going over the recommended spec for PCIe power draw.
Hey look, an SLI connection! That's right, for those of you that are feeling REALLY froggy you can get a pair of Titan Z cards and run in them in parallel, giving you four GK110 GPUs worth of performance!
I have heard people calling the Titan Z a 2.5 slot card, and from this image you can see where that line comes from. The cooler itself only takes up about 2.5 slots worth of space but with the bracket, you are going to be using up all three-slots.
Also worth noting is that the back plate on the Titan Z is thicker than others we have seen and it actually did cause the card to make contact with the very bottom of the IO panel on the ASUS Rampage IV Extreme X79 motherboard. It did not cause any specific issues, but just wanted to note it.
Under that shroud you'll find that single fan and a large array of heatsink fins that are more than capable of keeping the two GK110 GPUs cool during gaming.
Without its clothes, the Titan Z shows the immense complication of technical design that mashing a pair of huge GK110 GPUs on to a single card requires. You have a PCIe bridge chip in the middle, power management hardware resting between the two GPUs and 12GB of GDDR5 memory running at 7.0 GHz.
Now, let's see how it all stacks up to a pair of GeForce GTX 780 Ti cards in SLI and an AMD Radeon R9 295X2!
more accurately result when
more accurately result when putting 780 ti from 1020 palit
or ghz cause 295×2 is already factory overclocked
Two OC’d Titan Blacks if you
Two OC’d Titan Blacks if you need compute. Two OC’d 780 ti s for gaming. Check out the inno ichill 780 ti which runs OC in the high 50s on AIR!!! AMD can only dream with their power hungry space heaters. Titan Z is for the I don’t care what it costs crowd and Nvidia has been pretty good holding that segment.
2 titan blacks stock is fast
2 titan blacks stock is fast then 1 titan z stock. And the titan z is basically 2 titan blacks in one package.
Any info on usable ram? its
Any info on usable ram? its advertised as 12GB, but is all of that usable or is it merely 6GB mirrored?
I don’t like AMD going
I don’t like AMD going outside the power specifications, but that is the only thing that really makes the product make any sense. Nvidia stayed within the 375 W limit, so they have lower performance than the 780Ti in SLI. Given the high price, there isn’t much of a reason for this card to exist.
These dual gpu cards seem to really only be for marketing, rather than a real product. AMD made dual gpus closer to a real product by actually enabling full speed by using water cooling. It still doesn’t make much sense (higher price for same performance) unless you are building a space constrained system. The Titan Z fails at being a marketing/publicity stunt since it obviously can not compete with the 295×2 in performance. AMD and Nvidia want to stay in the public consciousness, even when they do not have a (real) new product to release. These cards will be super low production volume, but they obviously generate a lot of publicity. Since AMD beat them to market with the 295×2, they do not want any publicity, since this has turned to negative publicity.
These dual gpu cards will not really make sense until they actually connect the gpus together to allow them to share memory and act more like a single gpu. To share memory, they need really high bandwidth interconnect (~100 GB/s), which is probably doable between chips so close together. I have been wondering if they could just shrink the memory controller and use the pins for interconnect instead, or even make them programmable to allow the same chip so be used in a single gpu card with full width memory interface or in a multi-gpu card with narrower memory interface. This would also allow use of less memory. Designing and producing such a design may not be feasible right now due to the low volume nature of the super high-end gpu market though. I don’t think it is that useful for the compute market due the way compute gpus are being used (parallel independent task).
So i am glad i built my dream
So i am glad i built my dream machine with 3 gtx blacks that cost me as 1 z. i know it is overpriced, compared to 780ti, which is also overpriced, but running them underwater gave me huge potential for oc and smooth 4k. it was just a dream machine that i don’t plan to swap for the forthcoming years. To my opinion, graphic cards should not be priced more than 400$ i don’t believe there is no room for profit even for the z if this was the price target.. anyway, just an opinion.
Ahaa, its nice conversation
Ahaa, its nice conversation on the topic of this piece of writing at this place at
this web site, I have read all that, so now me also commenting here.
My site … karnisze nowoczesne
Actually after Watchdogs came
Actually after Watchdogs came out it has shown 3gb at Ultra Settings just doesn’t cut it anymore when developers push the limits.
I’m ready to build my Rig soon but tired of waiting for GTX880 with 8gb. If the 880s are not due out until Christmas I may just go ahead with ASUS Strix GTX780 6gb in SLI.
good test!
good test!
I don’t know who made the
I don’t know who made the decision to price this card at $3000. It can’t even outperform AMD’s $1500 card. This card should be sold for $1500 if not less.
As much as I hate to admit it, I think Nvidia would be unstoppable if it capitalizes on the Maxwell GPUs. The 750Ti is an impressive GPU and that was just the tip of the iceberg
They are just about equal
They are just about equal while the 780Ti’s are running at 876Mhz. Too bad you can easily get cards that run 1300 or even 1400 on water. Just imagine the performance gap they would have with +50% clockspeed on the nvidia cards 😀 What a joke
Note: FLOPS are primarily
Note: FLOPS are primarily important to people mining for Bitcoins or for protein folding. It means very little in the way of gaming.
With that being said, I think the only way that the TITAN-Z would pull ahead of other SLI/Crossfire setups is at 8K resolutions because of its immense 12GB of VRAM and 768-bit bus. Even then, I’m not sure that the ROPs, shaders and texture procs on the TITAN-Z would be able to keep up with that resolution.
I honestly think these cards’ target audience is those planning on mining, and that’s why they justify the exorbitant price tag despite the lackluster gaming performance but with a notable increase in teraFLOPS performance.
I’ve seen a bunch of vendors
I’ve seen a bunch of vendors dropping the price significantly on the TitanZ, Alienware had a TitanZ half off promotion and now CyberPower is giving 500.00 off the price. Has this made this card a better value? Trying to price a new system, have about 4g to spend. I do not want to put it together myself or support it, I am laaaaaazy.