nvidia recently introduced the new GeForce RTX 4000. The clear protagonist is the RTX 4090, a dedicated graphics card that is not only monstrous in performance: it is also monstrous in size.
The Founders Edition is huge. Those who have had access to NVIDIA’s own model, known as ‘Founders Edition’, are making it clear. It is curious, but in fact this graph is even compact compared to models from other manufacturers.
The 4090 is bigger than an Xbox Series S. pic.twitter.com/U0AtgyZapf
— Okami Games (@Okami13_) October 5, 2022
but really huge. as indicated at Okami Gamesthis graph is larger than an Xbox Series S, and it is amazing to see it next to an Xbox Series X like the one in the cover photo, which shows its thickness, but also when seen from the front.
make room. They also compared it to a Steam Deck or a Nintendo Switch – both are small – but the Founders Edition is not even the largest RTX 4090: they put it next to an ASUS TUF Gaming 4090 and saw how it was even more enormous .
Yes it’s bigger than a Switch too (and Steam Deck).
4090 FE pictured next to the Asus TUF Gaming 4090. pic.twitter.com/kgekzkEFws
— Okami Games (@Okami13_) October 5, 2022
These enormous dimensions and also its weight mean that installing it on a PC is going to be somewhat more complicated than usual: we will have to make sure not only that the box we use has enough space, but also that it is held very well and prevents the graphics from being able to cause damage to other components.
The RTX 3090 Ti on top — a little thinner, a little deeper — and the RTX 4090 below. There is an explanation to those dimensions.
They were going to use 8nm photolithography. As explained in IgorsLab, the problem is that these graphics cards were initially designed so that it was Samsung who manufactured its chip with 8 nanometer photolithography. The reason? TSMC’s 5nm chips would have been desirable, but that manufacturer had reserved those processes for upcoming chips from Apple and AMD.
With 8nm photolithography, the GPU was less efficient, and that made the initial leaked data point to consumption of up to 600 W. The graphics needed huge heatsinks, which are the ones that have ended up being used.
It didn’t have to be this way. When those heatsink designs were complete, TSMC called NVIDIA. “Hey, in the end we can make the GPU in 5nm.” NVIDIA agreed to the terms and managed to get these GPUs manufactured using TSMC’s 4N (5nm) process. Suddenly they had much more efficient chips, and the TDP, which was supposed to be 600 W, has ended up being 450 W for the RTX 4090 and 320 (instead of 420) for the RTX 4080.
But it was too late to change heatsinks. That change caused NVIDIA and its partners to be launching absolutely monstrous cards on the market when, as they comment in El Chapuzas Informático, they could have actually launched more compact models.
fresh. It will be interesting to see how these models evolve once the manufacturers can polish designs, but for now one thing is certain: the RTX 4090 are huge, but of course they will have plenty of cooling and dissipation, so we will have components that will surely work more quietly. And they get less hot. Which is not bad either.