• Nvidia GeForce RTX 5090

  Architecture

  Blackwell

  Shader Units

  21,760

  Ray Accelerators/Cores

  170

  Base Clock Speed

  2,010MHz

  Boost Clock Speed

  2,410MHz

  Memory Capacity

  32GB GDDR7

  The Nvidia GeForce RTX 5090 is an absolute beast with 32GB of speedy GDDR7 VRAM and 21,760 shader units to chew through graphical tasks with aplomb. It's also super expensive, but ultimate power always costs.

  Pros & Cons

  • 32GB of GDDR7 VRAM
  • Tons of CUDA cores and ray tracing cores
  • PCIe 5.0

  • 575W power draw
  • Extremely expensive

  $2000 at Newegg $2000 at Best Buy

• Nvidia GeForce RTX 5080

  Memory Clock Speed

  1,875 MHz

  Architecture

  Blackwell

  Process

  5 nm

  Shader Units

  10,752

  Ray Accelerators/Cores

  84

  AI Accelerators/Cores

  336

  The Nvidia GeForce RTX 5080 is half the price of the RTX 5090, which makes sense since it's half the CUDA cores, half the RT cores, and half the VRAM amount. It's still a powerful GPU and will carry you through games with the power of DLSS 4.

  Pros & Cons

  • Support for multi frame generation and other DLSS 4 features
  • Gorgeous flow-through cooler
  • PCIe 5.0

  • Still expensive
  • Not much of an improvement over the RTX 4080

  $1000 at Newegg $1000 at Best Buy

The first two graphics cards from Nvidia's RTX 50 series are out. Well, they were out, and then they ran out, but stock is being replenished in the coming weeks. If you were lucky to get either the GeForce RTX 5090 or 5080 from the launch stock, you've got one of the best GPUs around and you might want to know what the best games for you to play are. If you missed the first drop or are still undecided on which of Nvidia's beastly graphics cards you're after so you can use multi frame generation, let's dive into the specs and help with that.

RTX 5090 vs RTX 5080: Price, specs & availability

Against the price of the flagship, the RTX 5080 looks like a relative bargain

The Nvidia GeForce RTX 5090 and 5080 launched on January 30, 2025. The flagship RTX 5090 has a $2000 MSRP, and the RTX 5080 has a $1,000 MSRP. Prices go up accordingly, with AIB models costing significantly more in some cases. Both GPUs are aimed at 1440p and 4K gamers, partly because of the graphical prowess but also because of enhanced DLSS frame generation that can generate up to three AI-created frames per raster frame from the traditional processing pipeline.

Both graphics cards use the Blackwell architecture, have PCIe 5.0 connectivity to the motherboard (the first time on a consumer GPU), and have speedy GDDR7 memory. While there are some minor differences, the rough cut of the specifications means the RTX 5080 is essentially half the RTX 5090, with just under half the cores and half the memory. It's also on a half-width memory bus that generates half the bandwidth overall. But, it's half the price, so that makes a little more sense.

• Nvidia GeForce RTX 5090

  Shader Units

  21,760

  Base Clock Speed

  2,010MHz

  Boost Clock Speed

  2,410MHz

  Memory Capacity

  32GB GDDR7

  Memory Bus

  512-bit

  Memory Bandwidth

  1,792GB/s

  Power Draw

  575W

  Architecture

  Blackwell

  Price

  2000

  Ray Accelerators/Cores

  170

• Nvidia GeForce RTX 5080

  Shader Units

  10,752

  Base Clock Speed

  2,295 MHz

  Boost Clock Speed

  2,617 MHz

  Memory Capacity

  16 GB GDDR7

  Memory Bus

  256-bit

  Memory Bandwidth

  960 GB/s

  Power Draw

  360 W

  Architecture

  Blackwell

  Price

  $999

  Ray Accelerators/Cores

  84

Architecture

Big vs mid Blackwell battle it out

As we said, both of these graphics cards use the Blackwell architecture, which has a common core of technologies. Nvidia's biggest push is DLSS 4, which, along with better upscaling, a new transformer model for better quality, and better antialiasing, comes with multi-frame generation for the first time. This is unique to Blackwell and enables up to three AI-generated frames per raster frame, when before, this frame generation only created one additional frame.

But we also get GDDR7 memory for the first time, PCIe 5.0 for the interface to the motherboard for the first time, and a solid cooling solution that can tame the 575W of the RTX 5090, so it'll have no issues with the 360W of the RTX 5080. Plus, each GPU is now two slots in width, making it easier to fit into more cases. The bandwidth that GDDR7 enables will mean these cards will be powerful at AI workloads, as well as the gaming loads they're being marketed for.

Performance and power

The RTX 5090 is king, with the 5080 getting half of everything

If you were going purely on specifications, you might expect that the GeForce RTX 5090 would be exactly twice as powerful as the RTX 5080 when you play games or run synthetic benchmarks. Except, it never quite works out like that, and for whatever reason, the RTX 5090 is roughly 56% more performant at 4K resolution (from Digital Foundry). The performance gap narrows as you go to 1440p and 1080p, but nobody should be buying these two GPUs for 1080p unless they're playing esports competitively and need the best frame rates they can get at any cost.

In any case, those who want the ultimate in performance will want to find the RTX 5090 when it's in stock. Those who want a better value will likely want to look at the RTX 5080 out of these two cards. Both are 4K gaming monsters, and DLSS 4 makes them even more so.

RTX 5090 vs 5080: If you can afford it, there's only one winner

The RTX 5090 is twice the GPU of the 5080

If money is no object, you're always going to go for the best and most expensive GPU because that will give you the most performance. That's the GeForce RTX 5090, and it's a beast. But, the enthusiast level has no competition this generation, with AMD effectively ceding in the last generation and slinking to the mid-range, and Intel is only just competing at the budget level. But this generation also has no competition at the RTX 5080 level because AMD gave up there too.

That means Nvidia can charge what it likes for both the RTX 5090 and 5080 and make consumers pay because it's the only viable option. Add the double-whammy of stopping production on the RTX 4080 Super and RTX 4090 in 2024, and there is no viable alternative for high-end PC gamers who are looking for graphics cards.

For anyone looking for a good value GPU right now, neither of these cards is that. The RTX 5090 is super powerful but also super expensive, and the RTX 5080 is only a relative value proposition based on the RTX 5090's price and the RTX 5070 Ti's price of $750. That said, enthusiasts are always going to buy the best GPU available, and that's the RTX 5090.

The Nvidia GeForce RTX 5080 is a relative value compared to the RTX 5090, with half the GPU for half the cost. It's powerful right now, but the 16GB of VRAM is going to run into issues in the near future as games get more resource-intensive.