Computex 2016 is set to be a climax of the long-standing battle between NVIDIA Pascal and AMD Polaris. Both companies have come up with novel architectures for their latest graphic cards, and we have been covering in detail both Pascal and Polaris to keep you updated on the latest developments. News has it that both the architectures will launch this June.
Polaris, for those who are still not aware, is the fourth gen GCN architecture manufactured by AMD for their latest Raedon series of GPUs. It comes up with an all new memory controller, multiple and newly designed multimedia cores that have been built using the 14-nm finFET technology and a processor of an all-new architecture. The GPUs that have hence been consolidated are expected to be massive improvements over the GPUs that are prevalent today, including faster speeds, lower power consumption, and better leakage properties.
However, a catchy competition to the Polaris is only the NVIDIA Pascal, it would seem, with the 16-nm finFET technology developed at the Taiwan Semiconductor Manufacturing Corporation, NVIDIA is coming up with its new flagship, the GP100. This monster is set to turn the tables of the GPU industry with two pairs of 4-Hi HBM2 stacks that help the device achieve sixteen gigabytes of VRAM.
The Pascal is coming with twelve TFLOPS of single precision performance and four TFLOPS of double precision performance which will be achieved with a memory bandwidth of 1024 GB/s. The device includes seventeen billion transistors, albeit it is smaller than any of its predecessors.