Nvidia announced the Blackwell-powered RTX 50 series of graphics cards at CES 2025. This new generation of GPUs brings numerous benefits, including the latest iteration of DLSS (with multi-frame generation), improved CUDA and Tensor cores, more advanced memory, and beefier RT cores. Nvidia introduced Blackwell as an architecture for generational AI and the company's entire CES presentation was about AI. You'd be forgiven for believing Nvidia was an AI-only company if you're new to the GPU space. The RTX 5080 is a mid-range GPU, sitting below the flagship (and incredibly expensive) RTX 5090. But how does it compare to the outgoing RTX 4080?
3 Enhanced visuals with DLSS 4
DLSS 4 isn't exclusive to RTX 50 series GPUs, but you will not enjoy all the benefits it offers without upgrading to one. Nvidia will want to sell some RTX 50 series cards so it was inevitable that some features would be exclusive to this new generation of AI-powered GPUs. What's unique for RTX 50 series cards on the Blackwell architecture is DLSS Multi Frame Generation. This differs from normal DLSS frame generation in that the AI cores are used to take the one extra frame generated through DLSS and crank it up to three. So instead of two frames, you get four.
Does this equate to better performance? Not exactly. If you're struggling with low frame rates, DLSS 4 with Multi Frame Generation will not magically improve your FPS to more than a hundred, but it will smooth out the feed to the monitor, especially if you have a high-refresh-rate panel. All other improvements to DLSS, including memory optimizations, will be transferred to select older Nvidia GeForce RTX GPUs.
3 reasons DLSS is still the most revolutionary technology in PC gaming
DLSS has come a long way in six years, but I'm still excited about what the future holds
2 Better price to performance
The Nvidia GeForce RTX 5080 launched with an MSRP of $999, the same as the RTX 4080. AIB partners will likely charge slightly more for aftermarket cards, and stores could increase prices further due to demand. Because of the same price but higher reported performance, the RTX 5080 is a better deal. During its CES 2025 briefing, Nvidia promised RTX 4090 performance with the RTX 5070 through AI development instead of raw performance and rasterization. We can see this in the specifications comparison between the RTX 5080 and RTX 4080 with only slight bumps in certain areas but a considerable gain in reported performance.
According to Nvidia, the RTX 5080 offers up to two times the performance across a range of tests, including some heavier games and rendering suites. Cyberpunk 2077, Alan Wake, and Black Myth: Wukong all saw a two-times increase in performance over the RTX 4080, which is likely due to the AI boost with DLSS 4 and other advancements made through the company's Tensor cores. The slight improvement to the specifications would carry some of the weight, but these generational gains are powered by Nvidia's AI tech.
We'll have the RTX 5080 in our hands soon enough for testing to see how it fares against the RTX 4080. It should be considerably better when comparing full game tests with DLSS and other graphical settings cranked up high, but how much so will depend on the CPU used, the game, and other parts of your PC.
Best GPUs in 2025: Our top graphics card picks
Picking the right graphics card can be difficult given the sheer number of options on the market. Here are the best graphics cards to consider.
1 More powerful architecture
Nvidia used Ada Lovelace for the RTX 40 series of graphics cards and the RTX 5080 runs on Blackwell, which launched at CES in 2025. Nvidia didn't push system upgrades as hard with the RTX 5080 and other Blackwell GPUs, instead focusing on the enhancements made to its AI capabilities. For better or for worse, Nvidia has transformed from a simple GPU brand to one with heavy development in artificial intelligence and machine learning. This effort brought us DLSS, an incredible piece of kit that can boost in-game performance when enabling advanced graphics such as ray tracing.
-
Nvidia GeForce RTX 5080 Nvidia GeForce RTX 4080 Shader Units 10,752 9,728 Ray Accelerators/Cores 84 76 Base Clock Speed 2,295 MHz 2,205 MHz Boost Clock Speed 2,617 MHz 2,505 MHz Memory Clock Speed 1,875 MHz 1,400 MHz Memory Capacity 16 GB GDDR7 16 GB GDDR6X Memory Bus 256-bit 256-bit Memory Bandwidth 960 GB/s 716.8 GB/s Power Draw 360 W 320 W Architecture Blackwell Ada Lovelace Process 5 nm 4 nm AI Accelerators/Cores 336 304
The RTX 5080 has some notable improvements over the RTX 4080. In terms of specifications, we're looking at an additional 1,024 CUDA cores (shader units), 12 ray tracing cores, 32 Tensor cores (AI accelerators), GDDR7 RAM (compared to GDDR6X), higher memory bandwidth, and faster clock speeds. All this is based on the new Blackwell architecture with a slightly higher power draw of 360W vs. 320W for the RTX 4080. Though we are seeing GDDR7 RAM included, it's still 16GB, which is identical to the outgoing RTX 4080.
It's a tricky upgrade path
Upgrading from an RTX 40 series to an RTX 50 series graphics card is a difficult sell. An RTX 4080 isn't a terrible GPU. It may be viewed as a better value upgrade for other PC gamers with the introduction of the newer generation since some will want to offload their existing GPUs to make way for an upgrade. Should you buy either of these graphics cards for a fresh PC build, I'd go with the latest and greatest, the RTX 5080. It's the same price as the RTX 4080, though good luck locating stock and listings that aren't charging considerably more.