A Closer Look
From a first glance, it looks identical to the RTX 5090, but there is a small difference, and I’m not just talking about how it says RTX 5080 on it now, instead of 5090, but it’s actually a slightly different shade of colour, which in all honesty, I prefer the colour of the bigger brother, but I also wouldn’t be willing to drop double the price on it, just for aesthetic reasons, though of course, you get what is arguably a lot more power, though we will see what that means when we look at benchmarks shortly.
Due to that being the other real difference between the two, we still have a dual-slot card, which you could claim is easier to cool on a less powerful card, though they’ve still kept the dimensions the same at 304mm in length and 137mm in width, so there’s still a lot of surface area to help spread the heat to dissipate.
Much like the RTX 5090, NVIDIA with their Founders model are relying on their “fan pass-through” cooling system to help push airflow through the heatsink fins, and exhaust upwards, though there has been talk about what this means for CPU temperatures, and it’s something we’re definitely going to look at on both the 5080 and 5090 Founders cards in the very near future, because a lot of outlets do test outside of a chassis, and I honestly believe, that could be a huge difference in terms of performance, especially in terms of 1% low figures, which we saw on the 5090, were actually lower than expected.
In terms of branding, there’s not really too much going on, short of just that really small NVIDIA logo once again, and I really like that. I know who makes the GPU. I know what I’ve bought. I don’t need reminding, so I’d definitely like to give NVIDIA some brownie points on that one for keeping it nice and clean.
The card, like the 5090 has some understated LED lighting on the middle of the card and the GeForce RTX logo on the side, which sits next to the angled 12V-2×6 connector though the 5080 has considerably less total graphics power of just 360W which is just over 12% more than its predecessors, the RTX 4080 and 4080 SUPER, which both came in with a TGP of 320W and due to this, NVIDIA do recommend an 850W power supply with 3 available PCIe 8-pin connectors for the included adapter or a single PCIe Gen 5 cable capable of supplying at least 450W of power.
In terms of display outputs, just like the 5090, it packs a single HDMI and 3 DisplayPort connectors, and while I’ve seen comments on the 5090 review we did with consumers upset that it only has a single HDMI, I’m personally ok with this as DisplayPort 2.1b offers much more bandwidth over HDMI 2.1b meaning higher resolutions and refresh rates are possible, so maybe that’s a nod to 8K gaming being possible in the near future.
The PCB on the Founders 5080 is still extremely small and is classed to some as a work of art and design engineering, especially when you factor in that the GB203-400-A1 GPU is packing 10752 CUDA cores which is an 11% uplift over the RTX 4080 and a smaller 5% increase over the 4080 SUPER. Other specs include the same uplift of Texture Mapping Units, now at 336, the same for Streaming Multiprocessors, now at 84 along with the same for RT cores and the same uplift for Tensor cores, which now come in at 336. For Render Output Units, which are now at 128, we see 14% more when compared to both the RTX 4080 and SUPER variants and both the L1 and L2 cache remain unchanged from the Ada Lovelace predecessors.
