Last week during Computex 2015, NVIDIA released the GeForce GTX 980 Ti graphics card. The NVIDIA GeForce GTX 980 Ti features 2816 NVIDIA CUDA cores and 6 GB of GDDR5 memory with a price tag of $649.99. For a limited time gamers will also get Batman: Arkham Knight with the purchase of a GeForce GTX 980. That price tag might be more than some are willing to pay, but you’ll soon learn that is shares many key features with the GeForce GTX Titan X 12GB graphics card all while costing $350 less than NVIDIA’s most expensive desktop gaming graphics card solution. NVIDIA believes that both of these cards are deliver a good gaming experience on the latest games in 4K at max settings, plus next-gen experiences like VR. Good luck finding a GeForce GTX Titan X for $999 though as up to this point most are selling for over $1,100 even after the GeForce GTX 980 Ti has launched!
The NVIDIA GeForce GTX 980 Ti uses the ‘big’ GM200 ‘Maxwell’ GPU like the GeForce GTX TitanX, but it has two Streaming Multiprocessor units disabled. As a refresher the NVIDIA GeForce GTX Titan X has 24 Streaming Multiprocessor units. Each SM contains 128 CUDA cores and that is how you end up with a total of 3072 CUDA cores that handle the pixel, vertex and geometry shading workloads. The texture filtering is done by 192 texture units and you have 3MB of L2 cache and 96 ROPs. The 3072 CUDA cores in the Titan X’s GM200 GPU are clocked at 1000MHz/1075MHz. Lastly, you have 12GB of GDDR5 memory on a 384-bit memory interface running at 7010MHz (7GHz) effect memory clock.
The NVIDIA GeForce GTX 980 Ti on the other hand had has just 2816 CUDA cores since two SM units have been disabled. This also means there is a reduction in the texture units, so you’ll find just 176 on the GTX 980 Ti. The clock speeds are set to the same as on the GeForce GTX Titan X, so the card is clocked at 1000MHz for the base clock and 1075MHz on the boost clock although we have been told that the boost clocks are more aggressively set on this card. The memory on the card has been cut in half, but the 6GB of GDDR5 memory still runs on a 384-bit memory bus and is clocked at 7010MHz. The GeForce GTX 980 Ti still has 96 ROPs and 3MB of L2 cache.
|Titan X||GTX 980 Ti||GTX 980||GTX 780||GTX 680||GTX 580|
|GDDR5 Memory Clock||7,010MHz||7,010MHz||7,000MHz||6,008MHz||6,008MHz||4,008MHz|
|Memory Bus Width||384-bit||384-bit||256-bit||384-bit||256-bit||384-bit|
|FP64||1/32 FP32||1/32 FP32||1/32 FP32||1/24 FP32||1/24 FP32||1/8 FP32|
|Manufacturing Process||TSMC 28nm||TSMC 28nm||TSMC 28nm||TSMC 28nm||TSMC 28nm||TSMC 40nm|
The only people upset right now might be those that purchased a GeForce GTX Titan X as that card came out just three months ago and here is a card that has roughly 8% fewer CUDA Cores and Texture Units for 35% less cash! With an 8% reduction in the shader and texture count you might expect to see a drop like that in gaming performance, but NVIDIA said that the performance difference between the Titan X and 980 Ti will be minor.
The NVIDIA GeForce GTX 980 Ti might leave some wondering why NVIDIA is coming out with such a card so quickly after the GeForce GTX Titan X was released, but it has to do with what AMD has been not so quietly working on. The AMD Radeon Fury series of graphics cards is expected to be announced on June 16th at E3 2015. The AMD Radeon Fury lineup is rumored to be based on the companies new Fiji GPU micro architecture and will be the first desktop GPU to feature High Bandwidth Memory (HBM).
The NVIDIA GeForce GTX 980 Ti is likely the card that NVIDIA will be using to compete against the AMD Radeon Fury X and will hopefully hold NVIDIA over until they can release video cards based on the Pascal GPU architecture sometime in the middle of 2016. If you are looking to buy a video card in the summer of 2015 you might as well get one as Pascal could still be a year away and we don’t see it getting pushed forward unless the AMD Fiji GPU is better than everyone expects.
Let’s move along and take a look at the GeForce GTX 980 Ti reference card!