Listen. I’ve been around computer stuff my entire life. I’ve been around games stuff. I know how to build a computer. I can fix some basic errors and stuff. I know how to make the game work good. But if you looked at me dead in the eye and said ‘teraflop’ I would politely ask you to not curse at me. If you gave me the official Xbox Series X technical specs, I would read them and pretend to get it but I wouldn’t get it at all.
And you know what? As a games journalist, I feel like I should at least be able to kind of know what this means. So, let’s find out together, so we can both nod along when someone tells us something technical because we at least *kind of* get it.
On the Left:
The IBM ASCII White – A supercomputer consisting of 512 nodes that offered a theoretical peak performance of 12.3TF. It went into operation in 2001 and cost $110m to build.
On the right:
The Xbox Series X – A 12TF console releasing in 2020. pic.twitter.com/kFzePEhBMm
— Daniel Ahmad (@ZhugeEX) February 24, 2020
Teraflop? What?
First off, let’s get the big question out of the way. What’s a teraflop? First, we need to understand that ‘flop’ is the thing being measured, and ‘tera’ is the unit of measurement (think mega, giga, etc.). ‘Flop’ in this case stands for ‘floating-point operations per second’. That means absolutely nothing to me so I had to look some more, and it turns out, these floating points that are operating so many times per second are calculations your graphics card is doing to make polygons and stuff happen.
Essentially, it’s ‘how many polygons your computer can calculate to make incredibly intricate images’. Bigger amounts of flops mean bigger amounts of polygons which mean more realistic-looking games. Which is kind of concerning because I already keep getting screenshots of Forza confused for real cars. So, one teraflop, therefore, would be a trillion flops. Similarly, a megaflop would be a million flops.
To better get an idea of what 12 teraflops means, for a reference point, the Playstation 4 worked on 1.84 teraflops. The Xbox One S worked on 6 teraflops. This thing is going to throw twice as many polygons at you than before. That is more polygons than anyone needs I think, to be honest. But it’s very exciting.
Comparisons
So. Now we all know what a teraflop is and that it is a lot. Now, the only way I can comprehend this is if I compare it to something I know about. Let’s say, my personal gaming computer. It’s a mid-high range, it’s not a beast, but it’ll play most things on high.
As you can see, there’s a lot of immediate differences. Namely, I run NVIDIA and Intel, vs. Xbox’s AMD system. So right off the bat, it’s a little confusing for me because I am very simple. Where I run an Intel Xeon 4 core 3.40GHz system, the Xbox Series X runs a custom AMD 8 core 3.6GHz system. That means you get to do more stuff, have better frame rates, that kind of thing. Where the Xbox One X could do 4K gameplay, stuttering is still possible. With this? It’s basically not gonna happen. 4K, possibly even 8K games, running at a smooth 60FPS minimum, with 120FPS entirely within the realms of possibility. So you’ll get those teraflops giving you pretty pictures while the CPU is working hard to make sure they stay pretty.
Also, it’ll let you have multiple games minimised and running at the same time. Which is huge.
The RAM is another big one, with 13GB dedicated purely to games, while 3GB is for running the operating system. Especially being DDR6, when most computers run on DDR4. From what I can tell, the bigger the number after DDR, the faster it goes. So this will be very fast! Very smooth! Very pretty!
Essentially Xbox is creating a tiny little supercomputer just for video games. It’s an absolutely ridiculous technical feat that seems excessive, but at the same time, ushers in a next step for what we’ll be aiming for with future consoles. Of course, the real test is when we actually see some first-party titles designed to push this to the limit, but so long as Master Chief is there, and I have been assured he will be, I’m pretty okay with it.