The RTX 5080 didn’t blow us away in our testing, but there’s no denying that it’s the second fast graphics card in the world, behind Nvidia’s ludicrously-priced RTX 5090 halo card.
Still, if you are sitting on a last-generation RTX 4080 Super, or found one at a good price second hand and are weighing up the pros and cons, it’s useful to know how these cards compare.
Pricing
Unfortunately, we don’t have any good news to share on either card, on either pricing or availability. The RTX 4080 Super initially launched at the start of 2024 with a price tag of $1,000, boosting the original 4080 with enhanced performance and a better price (down from $1,200).
While it maintained that price for most of its lifecycle, the 4080 Super is no longer in production and we can’t find it in stock anywhere. They are still some on second-hand sites, but they’re listed for over $1,200.
The RTX 5080 launched on January 30 2025, and quickly sold out. We’ve had a number of stories in recent weeks about stock being almost non-existent, and while it should pick up in the coming months, expect it to be short for some time to come.
Its initial retail price was $1,000, but Trump tarrifs and supply constraints mean it regularly goes for over $1,500 when available, and you’ll have to spend over $2,000 if you’re buying from scalpers on Ebay or similar.
| Nvidia RTX 5080 | Nvidia RTX 4080 Super | |
| CUDA Cores | 10,752 | 10,240 |
| RT Cores | 84, 4th generation | 80, 3rd generation |
| Tensor Cores | 336, 5th generation | 320, 4th generation |
| Boost clock | 2.6GHz | 2.55GHz |
| Memory size | 16GB GDDR7 | 16GB GDDR6X |
| Memory bus | 256-bit | 256-bit |
| Memory speed | 30Gbps | 23Gbps |
| Memory bandwidth | 960GBps | 736GBps |
| TBP | 360W | 320W |
On paper, there isn’t a lot of difference between these two cards. There are a scant few percent more CUDA cores in the newer RTX 5080, a couple of extra RT and Tensor cores, comparable clock speeds, memory counts, and memory bud-widths. The only standout difference here, is the faster memory, which gives the 5080 about 30% more overall bandwidth.
The newer card is based on the more-advanced Blackwell architecture, but it uses the same process node, so there are no major advantages to find there. Arguably, the biggest generational change here is support for multi frame generation as a part of DLSS 4.
Things are a little different when we talk ray tracing, as there, even small performance enhancements count. In the most demanding games, you’ll still find added performance from the 5080 and it will get you more playable frame rates, even without DLSS doing some heavy lifting.
When you do enable ray tracing, though, and particularly frame generation, it’s worth noting just how big a difference that can make. You won’t want to use it in every game and the 5090 still stands well ahead of anything else out there, but the 5080’s ability to generate all those added frames does give FPS a big boost.
It’s a strange situation we find ourselves in at the start of 2025. Both the RTX 4080 Super and RTX 5080 are excellent graphics cards for high-end, 4K and ray traced gaming. But should you pay the exorbitant $2,000+ fees for the 5080, when that should really be netting you a 5090?
But then you shouldn’t really spend $1200+ on an RTX 4080 Super in 2025, should you?
Unfortunately, while GPU stock is so low, you really are stuck with what you can find at a reasonable price, and that’s incredibly limited. If you can wait, the landscape of new GPUs will be far clearer by mid-March, when we should have the 5070 Ti and AMD alternative, RX 9070 XT, to consider too.
