What Is A Teraflop: Xbox Series X 12 Teraflop

A few days ago Microsoft’s Phil Spencer dropped more details on Xbox Series X specs and features. The biggest takeaway from the Xbox Wire article is that Xbox Series X will indeed be powered by 12 Teraflop & RDNA 2 based GPU. Please see the details below:

“

• Next Generation Custom Processor: Xbox Series X is our most powerful console ever powered by our custom-designed processor leveraging AMD’s latest Zen 2 and RDNA 2 architectures. Delivering four times the processing power of an Xbox One and enabling developers to leverage 12 TFLOPS of GPU (Graphics Processing Unit) performance – twice that of an Xbox One X and more than eight times the original Xbox One…..”

This news has got a lot of people talking about the power of the new console and how will it compare to previous consoles and PC. The simplest way to compare is to do it based on a number of teraflops, but what is a teraflop?

FLOPS stands for Floating-Point Operations Per Second. FLOPS are units of measure for the numerical computing performance of a computer or a GPU in this particular case. As with most measure units we have kiloFLOPS (thousand flops), megaFLOPS (million flops), gigaFLOPS ( billion). So a teraflop means one trillion calculations per second. We calculate it by multiplying the “boost clock speed” x “number of Cuda Cores or shading units” x 2 (two operations per clock). So the more teraflops GPU offers the more mathematical calculations it can perform and the more raw GPU it offers. Also the more Cuda Cores (NVIDIA) or Shader Units (AMD) and higher clock speed the more teraflops it offers meaning more floating calculations can be performed per second. However, these calculations are just a part of the GPU pipeline.

The GPU consists of a few parts. Shading Units in AMD GPUs or Cuda Cores in NVIDIA GPUs are responsible for all of the calculations on shaders in the rendering pipeline (geometry, vertex, pixel, etc.). But for GPU, this also consists of TMU’s (texture mapping units) and it is the TMU’s job to apply texture operations to pixels or ROPs (render output units). All data rendered has to travel through the ROP in order to be written to the framebuffer, from there it can be transmitted to the display. These extra GPU units do have an impact on the GPU performance too. With not enough number of ROPs or TMUs GPU may get bottlenecked. on top of that, there is also GPU memory which plays a crucial role in delivering good performance. GPU has to fetch a huge amount of data from memory before it can do any calculation so the more and faster memory is the better

I hope I have shown that number of teraflops on the GPU is very important as the more shading units it’s equipped with and higher the clocks it runs at the more calculations it can perform but other parts of GPU could potentially cause a bottleneck which would lower final performance in a game.

We also have to remember that there is a number of other GPU tech features that can boost performance like Variable Rate Shading or Rapid Packed Math without increasing the number of teraflops.

SOURCE: Xbox Wire