GeForce GTX 980 Ti
NVIDIA used the same cooler design for the reference version of the GeForce GTX 980 Ti graphics card as the GeForce GTX 970, GTX 980 and GTX Titan X. The green company used the same cooler for the GeForce GTX 980 Ti graphics card GPU, as in the case of the GeForce GTX Titan X - GM200. But now they use the GM200-310-A1 stepping, which is a cut-down version of GM200. NVIDIA still makes its own GPU at TSMC factories using the 28 nm process technology – the chip is impressive, considering its 8 billion transistors.
Unlike the GeForce GTX Titan X's 3.072 stream processors, the GeForce GTX 980 Ti only boasts 2.816 stream processors. They are distributed across 6 Graphics Processing Clusters (GPC) and 22 Maxwell streaming multiprocessors. 22 SMM x 4 SMM x 32 ALU units provide 2.816 GeForce GTX 980 Ti stream processors. There are eight texture units connected to each SMM, giving a total of 176 texture units. 16 raster pipelines provide a 64-bit interface, resulting in a total of 384-bit interface with 96 ROPs.
GPU The GM200 of the GeForce GTX 980 Ti operates at a base frequency of 1.000 MHz, with a minimum Boost frequency of 1.075 MHz. This is similar to the GeForce GTX Titan X, which operates at the same clock speeds. The performance difference between the two cards may be due to the different number of stream processors, some differences in architecture, and a smaller memory configuration.
GPU The GM200 of the GeForce GTX 980 Ti graphics card provides the full feature set of the "Maxwell" architecture. So you get almost all the functional blocks present in the GeForce GTX Titan X. This includes 3 MB of L2 cache, as opposed to 2 MB in the GeForce GTX 980 and GeForce GTX 970, where due to memory bandwidth limitations we get 1.792 KB. The cache bandwidth remains 512 bytes per clock.
GeForce GTX 980 Ti specifications
|
|
||||||
|
Chip
|
||||||
|
Frequencies
|
||||||
|
Memory
|
||||||
|
Interface and TDP
|
6 GB of GDDR5 memory are connected via a 384-bit interface. The memory operates at a clock frequency of 1.750 MHz. For some reason, NVIDIA decided not to install 2.000 MHz memory, which is already produced by SK Hynix and Samsung. NVIDIA also left High Bandwidth Memory for the next generation, which will be released under the code name “Pascal” next year. With a clock speed of 1.750 MHz and a 384-bit bus, the GeForce GTX 980 Ti memory operates with a bandwidth of 336,5 GB/s.
NVIDIA specifies the maximum power consumption of the GeForce GTX 980 Ti at 250 W. Of course, the video card is compatible with SLI multi-GPU, allowing you to combine one, two or three video cards into one system.
