Tag: NVIDIA

Nvidia’s new RTX 5080 graphics card isn’t as exciting as I was hoping it would be. While the sleek new Founders Edition redesign dramatically shrinks the size of the card compared to the RTX 4080 and RTX 4080 Super, you’re getting the same 16GB of VRAM and only small performance improvements over the previous generation of cards.

The RTX 5080, which arrives January 30th for $999, is about 15 percent faster than the RTX 4080 at 4K without the use of any AI-powered upscaling tricks. While the RTX 4080 managed to beat the RTX 3090, and the RTX 3080 dethroned the RTX 2080 Ti, the RTX 5080 isn’t enough to topple the RTX 4090. Nvidia has built up an expectation that the 80-series card will surpass the previous generation’s flagship, and that’s simply not the case this time around.

The RTX 5080 does comfortably beat its only competition at this price point: the $999 AMD RX Radeon 9700 XTX. That makes the RTX 5080 a good card for 4K if you’re willing to drop settings in some games and more GPU than you probably need for 1440p. For meaningfully better 4K performance, you’ll have to spend $2,000 on the RTX 5090.

If you’re upgrading to the RTX 5080 from the RTX 3080 or RTX 3080 Ti, expect to see average performance gains of more than 50 percent at 1440p and 4K. If you want more than that, you’ll have to rely on Nvidia’s new DLSS 4 and Multi Frame Gen to create more frames with AI models. As we work our way down the RTX 50-series lineup, it increasingly feels like Multi Frame Gen is going to be the main talking point.

The RTX 5080 Founders Edition model looks identical to the RTX 5090. It’s dramatically thinner than the RTX 4080 it succeeds, shrinking from a three-slot design down to two slots. I really like this redesign, which includes double flowthrough fans that exhaust air above the RTX 5080 into the case instead of out of the top and back of the card.

Just like the RTX 5090, Nvidia has slightly angled the power connector on the RTX 5080 so it’s easier to fit into a variety of cases. The included dongle, which converts three eight-pin PCIe power cables to fit the card’s 12VHPWR connector, has slightly more flexible cables than the one that shipped with the 40-series cards. You can also ditch the dongle and get a 12V-2×6 or 12VHPWR cable that connects directly to your power supply.

It’s disappointing to see Nvidia has stuck with 16GB of VRAM on the RTX 5080. AMD’s RX 7900 XTX offers 24GB, and while the RTX 5080 delivers better performance for now, it may well hit video memory limits in 4K gaming in the future.

For both my 4K and 1440p testing, I’ve paired Nvidia’s RTX 5080 with AMD’s latest Ryzen 9 9800X3D processor and Asus’ 32-inch 4K OLED PG32UCDP monitor. I’ve put the RTX 5080 up against the RTX 5090 — Nvidia’s current flagship — as well as the RTX 4090, RTX 4080 and 4080 Super, AMD’s Radeon RX 7900 XTX, and the RTX 3080 Ti from 2021.

Without DLSS or ray tracing enabled, the results are largely consistent across both 4K and 1440p: the RTX 5080 is about 15 percent faster on average than the 4080 at 4K and 12 percent faster at 1440p. It’s 12 percent faster than the Radeon RX 7900 XTX at 4K and 11 percent at 1440p, and it beats the RTX 4080 Super by 11 and 10 percent, respectively.

That’s enough to make this the best $999 graphics card, but by historical standards, it’s disappointing. I really wanted to see the RTX 5080 beat the RTX 4090, but it lags the older card by about 18 percent. Maybe it was a big ask for a $999 GPU to beat a $1,599 last-gen flagship. The $1,199 RTX 4080 was faster than the $1,499 RTX 3090, but there was only a $300 gap there, not $600. Still, it’s the first time in a long time we’ve had such a paltry increase in an 80-series card.

Nvidia’s top GPU, the RTX 5090, is 50 percent faster than the RTX 5080 at 4K resolution, but it’s also double the price. That certainly leaves the door open for an RTX 5080 Ti, priced between the 5080 and 5090, that can beat the RTX 4090.

The RTX 5080 makes the most sense if you’re upgrading from the RTX 30 series or earlier. At 4K, without DLSS or ray tracing, it’s nearly 54 percent faster than the RTX 3080 Ti (56 percent at 1440p), and you’ll also be able to make full use of DLSS 4 and Multi Frame Gen if you upgrade.

DLSS 4 and Multi Frame Gen

Nvidia claimed the RTX 5080 would be twice as fast as the RTX 4080. For those kinds of numbers, you’ll need to turn to DLSS 4 and Multi Frame Gen.

Nvidia’s new Multi Frame Generation tech uses the latest AI graphics models, powered by an updated transformer architecture, to generate up to three additional frames per traditionally rendered frame, pushing the RTX 5080’s frame rates beyond what it’s normally capable of at 4K.

Cyberpunk 2077 is the only game with official support for DLSS 4 and Multi Frame Gen before the RTX 5080’s official launch on January 30th. In my testing, I’ve found big frame rate increases at 4K with full ray tracing enabled, with fewer graphical glitches than Nvidia’s previous frame-generation tech thanks to the new transformer model. Without DLSS 4, Cyberpunk runs at an unplayable 19fps average with ultra settings and full ray tracing. DLSS 4 Multi Frame Gen x4 brings that up to a far more playable 128fps, but it isn’t the same as a true 128fps.

DLSS Super Resolution — which renders the game at a lower resolution and then uses AI to upscale it — bumps the frame rate up to 38fps. Multi Frame Gen creates the three extra frames to get to that 128fps mark. While motion does look smoother, in terms of input latency, it still feels like a 38fps game. Multi Frame Gen is far less noticeable when the base frame rate with DLSS Super Resolution is higher than 60fps, so at 4K, you might need to lower the quality settings to really see the benefits.

It’s this Multi Frame Gen that lets Nvidia say the RTX 5080 is twice as fast as the RTX 4080, with the previous-gen card managing 62fps on average in Cyberpunk 2077 with full ray tracing and Frame Gen x2. The RTX 5080 more than doubles this with x4 enabled.

It’s a better story at 1440p thanks to higher base frame rates. Without DLSS 4, Cyberpunk 2077 averages 39fps at 1440p with full ray tracing and ultra settings. DLSS 4 Multi Frame Gen x4 brings that all the way up to 225fps. While it’s still not a true 225fps, it feels a lot better than the 4K equivalent because DLSS Super Resolution gets the base frame rate up to 73fps before Multi Frame Gen kicks in.

This is the ideal situation for Multi Frame Gen because you’re getting the motion clarity benefits and the game still feels more responsive. It’s obviously not as responsive as 225fps would be without Multi Frame Gen if you really dropped the settings, but in a game like Cyberpunk 2077, I think people are going to notice the motion clarity improvements more than the input latency.

When the RTX 5080 launches on January 30th, there’ll be a new DLSS override feature inside the Nvidia app that lets you force games to use its new Multi Frame Gen and transformer models. I haven’t been able to test this yet, since it’s not available, but if it works, it could be a great way to improve DLSS image quality and Frame Gen in games before developers can officially patch them.

I focus most of my GPU testing on gaming — because that’s what GeForce cards are designed for — but the RTX 5080 is also very capable of video editing or AI workloads. I found that the RTX 5080 was nearly 10 percent faster than the RTX 4080 in PugetBench’s DaVinci Resolve test, and in Procyon’s AI XL (FP16) test, it was nearly 28 percent faster.

Nvidia recommends an 850-watt power supply for the RTX 5080, which is 100 watts more than for the RTX 4080 and 4080 Super. The RTX 5080’s total graphics power is 40 watts more than the RTX 4080; it maxes out at 360 watts instead of the massive 575-watt power draw of the RTX 5090.

I’m glad to see that the RTX 5080 doesn’t significantly increase the power draw over the previous generation. I only saw it hit 360W once, in the Metro Exodus extreme benchmark. Even in Cyberpunk 2077 running full path tracing and no DLSS, it only got up to 348 watts.

More impressively, on average, the RTX 5080 drew 278 watts of power across the nine games tested without DLSS or ray tracing. That’s slightly less than the 281-watt average I found on the RTX 4080 but 12 watts more than the RTX 4080 Super.

The RTX 5080 didn’t heat up excessively in my open bench testing. The highest temperature I recorded was in Metro Exodus, where it reached 71 degrees Celsius (160 degrees Fahrenheit). The RTX 4080 reached 68C in this same test, and the RTX 4080 Super hit a maximum of 63C.

DLSS 4 is really the star of the show so far with the RTX 50 series, but there still aren’t enough games to test it with. Multi Frame Gen has shown early promise, and it makes a lot more sense at 1440p in Cyberpunk 2077 on the RTX 5080 than it does at 4K. I’m waiting to test more games with DLSS 4 and Multi Frame Gen or even the ability to force these options on with the Nvidia app.

The RTX 5080 is the best $999 card on the market right now, beating AMD’s Radeon RX 9700 XTX by an average of 11 or 12 percent across the games I tested, without DLSS or ray tracing. It’s a significant upgrade over an RTX 30 series or earlier, with over 50 percent higher frame rates than the RTX 3080 Ti, plus access to the latest Multi Frame Generation technique. The 5080 offers about two-thirds the performance of the RTX 5090 for half the price and a little more than half the power draw.

But Nvidia hasn’t delivered the pure performance gains I was expecting. The RTX 5080 isn’t the cheaper RTX 4090 many were hoping for. I wanted to see a more meaningful bump to 4K performance than just 15 percent over the RTX 4080 without DLSS. I’m left wondering whether Nvidia will introduce an RTX 5080 Ti down the line and how close the upcoming $749 RTX 5070 Ti will get to the RTX 5080.