There will be a lot of criticism for Nvidia’s new RTX 4080, but it’s important to point out that there’s nothing wrong with the Founders Edition we’re reviewing today in terms of build, features and overall quality. Much like the RTX 4090, this is a quality product from a company that is arguably at the top of its game, a company at the cutting edge of graphics technology. Pricing is another matter, however. That’s just too high, even at its base MSRP and especially with some of the third-party partner cards we’ve seen. The product is far faster than all previous Nvidia Ampere offerings, including the RTX 3090 Ti, but the performance boost doesn’t always match the price charged, to the point where data points suggest the $1599 RTX 4090 actually offers a better deal.
The Founders Edition we’re reviewing today arrives in the same packaging as the RTX 4090 and in the same excellent case. The only difference is that while the same controversial 12VHPWR socket is on the board (yes, the melting one – and we’re still awaiting definitive answers on why this happened), but this time you get an adapter dongle with only three eight-pin PCIe inputs. That’s still more than enough, bearing in mind that the much more power-hungry RTX 3090 Ti comes with the same adapter – and the RTX 4090 also worked with it. The uniform design across both Ada Lovelace cards also means you get the same video options: one HDMI 2.1 port and three DisplayPort 1.4a ports.
Nvidia GeForce RTX 4080 – Digital Foundry’s video review.
As this is a new generation of GPUs, there is a certain suite of features common to the RTX 4090 and RTX 4080 – even though they use entirely separate chips. This means that Ada Lovelace’s improved video encoding options are preserved – much faster h.264/HEVC encoding, as well as dual AV1 encoders. After the RTX 4090 review, we took a look at AV1 functionality – and it’s impressive. For our high quality video downloads at digitalfoundry.net, we found that hardware AV1 encoding provided better quality than software h.264 with great saving in file size and while software HEVC encoding provided smaller files, Nvidia AV1 provided higher quality. For streamers and content creators, RTX 4000 is a big deal. Features like this should persist across all RTX 4000 products as new, cheaper cards appear.
In the chart below you’ll see how the RTX 4080 compares to its flagship partner and it’s fair to say the cuts are significant – wider and deeper than the spec differential we saw between RTX 3090 and RTX 3080. You’re moving 59% of the computing power, 73% of the memory bandwidth, and two-thirds of the VRAM. And yet you’re paying 74% of the flagship’s cost price, so realistically you should expect at least 74% of the performance of the RTX 4090. Ideally, it should be much higher, as users expect at a better price/performance ratio the cheaper the product.
RTX 4090 24GB RTX 4080 16GB RTX 4070 Ti (TBC Name) CPU AD102 AD103 AD104 Transistors 76.3B 45.9B 35.8B Die Size 608mm² 379mm² 295mm² CUDA Cores 16384 9728 7680 Boost Clock Memory 2.52GHz 2.51 GHz Interface 2.51 GHz 2.51 GHz Memory Bandwidth 192 bits 1018 GB/s 742 GB/s 557 GB/s TGP 450 W 320 W 285 W Recommendation PSU 850 W 750 W 700 W PSU Cables 4 x 8 spindles 3 x 8 spindles 2 x 8 spindles Base price $1,499/£1,649 $1,199/£1,269 TBC 12 October 2022 16 November 2022 TBC
You’ll notice the table above contains data for the “cancelled” RTX 4080 12GB, which is based on an entirely different Ada Lovelace processor and originally went on sale for $899. Its cancellation is actually more of a delay as the card gets rebranded into a different product line (our money is still on RTX 4070 Ti) and is rumored to arrive in early January.
Before we jump into the battery of gaming benchmarks we’ve put together, let’s take a quick look at power efficiency. Going into this generation of graphics hardware, we were preparing for extremely power-hungry devices that were due to arrive in the middle of a power crisis – maybe that’s not a good idea. Inaccurate rumors of a 900W RTX 4090 gave way to a 600W RTX 4090 and finally, the actual TGP: 450W. That’s still plenty of power, but actual in-game usage is lower . The RTX 4080’s power limit is 320W, so in that regard it’s exactly the same as its predecessor, but what’s the reality in gaming?
Power efficiency varies greatly depending on the game you’re using, of course, if we compare performance to power draw – which is exactly what we’ve done in the table below. In any case, the RTX 4080 reports power consumption below its 320W limit and compared to previous Class 80 cards, it has considerable efficiency advantages. Simply put, frame rates are vastly improved from generation to generation and especially compared to RTX 3080, the performance advantage doesn’t result in much of a power boost. In fact, as you’ll see below, the RTX 4080 can simultaneously deliver improved performance and reduced power consumption in certain workloads.
GeForce RTX 2080 GeForce RTX 3080 GeForce RTX 4080 Power Reduction Control Per Frame, 4K, High RT 213 W/17.2 fps – 12.38 Joules per frame 309 W/31.6 fps – 9.78 Joules per frame 301 W/44.66 fps – 6.74 Joules per frame 45% vs 2080 / 31% vs 3080 Dying Light 2, 4K, Ultra RT 193W/11.23fps – 17.19 Joules per frame 311W/25.3fps – 12.29 Joules per frame 290W/ 38.6fps – 7.51 Joules per frame 56% vs 2080/ 39% vs 3080 Forza Horizon 5, 4K, Extreme 181 W/47.7 fps – 3.79 Joules per frame 290 W/77.5 fps – 3.74 Joules per frame 217 W/118.0 fps – 1.84 Joules per frame 52% vs 2080/ 51% vs 3080
All of this sets the stage for a frankly colossal range of benchmarks presented on the next six pages, covering standard rasterization in modern titles, some of the most demanding ray tracing workloads we could put these cards through – more an overview of how today’s image reconstruction techniques (DLSS, FSR2) can help mitigate the performance impact of the most challenging games.
Before we start with the numbers, here’s a quick look at the new PC we’ve built to help test this new generation of GPUs. At the heart of our system is an Intel Core i9 12900K, running with P-cores at 5.2GHz and E-cores at 3.9GHz. The processor is complemented by a 6,000 MT/s Trident Z5 RGB DDR5 memory from G-Skill. The Noctua NH-D15 is our cooler of choice to handle the powerful CPU, while all titles run from PCIe Gen 3 NVMe SSDs. All components are mounted in an Asus ROG Maximus Hero Z690 motherboard, while power is provided by a Corsair RM1000i power supply.
With the preliminaries out, check out the table of contents below and find out how the RTX 4080 stacks up against its closest Nvidia rivals, as well as AMD’s RX 6900 XT and RX 6800 XT. Please note that following the release of the RTX 4090, Nvidia actually improved the performance of the RTX 3000 class via a driver update. All RTX cards have been redesigned to account for performance changes.
Nvidia GeForce RTX 4080 Review
Introduction, Hardware and Power Analysis [This Page]
RT References: Dying Light 2, Cyberpunk 2077, Control, F1 22
RT references: Hitman 3, Metro Exodus Enhanced Edition, Marvel’s Spider-Man Remastered
RT/DLSS benchmarks vs FSR2: Cyberpunk 2077, Dying Light 2, Marvel’s Spider-Man Remastered
Game Benchmarks: Control, Cyberpunk 2077, Doom Eternal
Game references: F1 22, Gears 5, Hitman 3
Game references: Forza Horizon, Red Dead Redemption 2, Shadow of the Tomb Raider
Nvidia GeForce RTX 4080: the verdict of Digital Foundry
To view this content, please enable targeting cookies. Manage cookie settings
Article source https://www.eurogamer.net/digitalfoundry-2022-nvidia-geforce-rtx-4080-review-a-powerful-gpu-with-a-big-pricing-problem