Nvidia GeForce RTX 4080 Founders Edition Review: 4K performance and efficiency champ that deserves sub-US$1,000 pricing

The RTX 4080 is already priced above US$1,000, which leads one to wonder why Nvidia didn't simply use a nerfed AD102 with a few more CUDA cores. A likely explanation is that Nvidia didn't have plans to make as many AD102 dies in the first place, given that Titan-class cards are still a niche market. 

The AD102 in the RTX 4090 is already 11% deficient in SM count compared to the full AD102 GPU. So, it probably makes better sense to just have a new die made in high volumes targeted at the gaming crowd, instead of just binning down flagship chips.

In this review, we put the Nvidia GeForce RTX 4080 Founders Edition through its paces and see if the performance uplift on offer really justifies the US$1,200 ask.

At first glance, the RTX 4080 seems to have cut corners when compared to the RTX 3080 Ti's specifications. The Ada mainstream gaming flagship GPU has lower CUDA, Tensor, and RT core counts along with a cut-down memory bus and bandwidth. What Nvidia apparently cuts in numbers makes up in other aspects of the architecture.

For one, the AD103 in the RTX 4080 crams 45.9 billion transistors in a relatively smaller 379 mm² die area compared to the 28.3 billion in the RTX 3080 Ti's 628 mm² GA102 die. The fourth gen Tensor cores are now capable of outputting 780 TFLOPs with Sparsity, compared to the 273 TFLOPs with Sparsity of the RTX 3080 Ti. RT TFLOPs have also gone up to 113 in the RTX 4080 compared to 66.6 in the RTX 3080 Ti. 

The Nvidia GeForce RTX 4080 Founders Edition is pretty much identical to the RTX 4090 Founders Edition in terms of dimensions and build quality. Therefore, we refer you to our review RTX 4090 FE for more information on the design and connectivity options.

The test system is identical to what we've used for our review of the RTX 4090 FE. This includes the following components:

• Cooler MasterFrame 700 test bench
• Intel Core i9-13900K
• Gigabyte Z790 Aorus Master motherboard
• Gigabyte Aorus Gen 4 2 TB NVMe SSD for benchmarks and games
• 2x 16 GB Kingston Fury Renegade DDR5-6400 RAM at DDR5-6000 speeds with timings 32-38-38-80 at 1.35 V
• Gigabyte M28U 4K 144 Hz monitor
• Cooler Master MasterLiquid ML360L V2 ARGB AIO cooler
• Cooler Master MVE Gold 1250 V2 fully modular PSU

A fresh installation of Windows 11 22H2 with the latest patches was used together with Game Ready driver version 526.72.

We would like to thank Cooler Master for supplying the test bench, AIO, and PSU, Gigabyte for the Core i9-13900K, Z790 motherboard, M28U monitor, and NVMe SSD, and Kingston for the memory. Due to some reasons, we had to change to a new DDR5-6400 memory kit. We used a DDR5-6000 XMP profile to match our previous RTX 4090 FE test system as much as possible.

The RTX 4080 offers a fairly significant 23% cumulative performance uplift over the RTX 3090 Ti in synthetic benchmarks, whilst putting up a similar show against the RX 6950 XT. This delta increases to 58% when compared with the RTX 3080 average. 3DMark Fire Strike Graphics is the only test where we see the RX 6950 XT to have a slight edge over the RTX 4080.

Unigine benchmarks see Nvidia GPUs leading handsomely throughout. The RTX 4080 FE is able to outperform the AMD Radeon RX 6800 XT by more than 60% in most Unigine tests.

When compared to the RX 6900 XT, the RTX 4080 shows a very impressive 76% uplift in the Heaven 4.0 OpenGL and greater than 50% gains in Unigine Superposition.

The RTX 4080 features 40% less CUDA cores compared to the RTX 4090, but the performance deficit generally lies in the 20% mark in most cases. V-Ray 5 CUDA and RTX tests see the RTX 4080 trailing behind the RTX 4090 by 27% and 29%, respectively. Nevertheless, we still see greater than 40% benefits when compared to the RTX 3090 Ti, which is impressive.

We see 70% leads when compared to the RTX 3080 Ti in V-Ray 4.1 GPU and LuxMark Room GPU tests. However, AMD cards dominate as usual in LuxMark Sala.

We notice that Blender 3.3 CUDA scores are way off the mark, similar to what we had observed previously with the RTX 4090, which doesn't make sense. This could be a driver optimization issue, so we have asked Nvidia for a clarification and will keep you posted once we hear from them.

The RTX 4080 is squarely aimed at gamers — there's no option to enable ECC memory — but performance in pro apps is not entirely abysmal. The AD102-based RTX 4090 leads by 17% in SPECviewperf 12 and 13 and by 27% in SPECviewperf 2020 v1, which is on expected lines given the AD103 GPU's architectural constraints.

Unsurprisingly, the RTX A6000 gets a huge 148% lead thanks to driver optimizations. Even the laptop versions of the Quadro RTX 5000 and RTX A5000 show significant gains in SPECviewperf tests overall.

For those who aren't looking at a Titan-class card like the RTX 4090, the RTX 4080 is still amenable for workflows such as 3Ds Max.