NVIDIA GeForce RTX 4080 Super Founders Edition 16GB vs. MSI GeForce RTX 4080 Super Expert 16 GB Review – Super-homeopathic differences on all fronts, but cheaper

1 - Introduction, technical data and technology 2 - Unboxing and technical details of both cards 3 - Test system and measuring equipment 4 - Teardown: PCB and components 5 - Teardown: Cooler and fans 6 - Material analysis: NVIDIA RTX 4080 Super FE 16GB 7 - Material analysis: MSI RTX 4080 Super Expert 16GB 8 - Gaming-Performance WQHD (2560 x 1440) 9 - Gaming Performance Ultra-HD (3840 x 2160) 10 - Gaming Performance DLSS vs. FSR 11 - Gaming performance with Frame Generation 12 - Latencies and lags 13 - Workstation graphics and rendering 14 - Details: Power consumption and load balancing 15 - Load peaks, capping and power supply recommendation 16 - Temperatures, clock rates and infrared analysis 17 - Fan curves, noise level and audio samples 18 - Summary and conclusion

Today marks the last round of the green super upgrade weeks and with the new GeForce RTX 4080 Super, there is a chip that is on the one hand more of the same and yet is supposed to be something different. With the RRP of 1109 euros (old model 1469 euros), the company naturally wants to create a sales impulse and at the same time suggest a completely new product that it wants to offer at a lower price without perhaps losing face by changing the price of the already well-known GeForce RTX 4080. More graphics performance is promised at a lower price, which of course makes a direct comparison a real challenge. However, the retailer prices are a completely different matter, because apart from a few so-called MSRP cards, the financial freewheel will probably be used again for the individual and better equipped models. MSI has set an RRP of 1249 euros for the MSI GeForce RTX 4080 Super Expert 16GB and it will be interesting to see whether and how retailers react.

Important foreword

This time everything went a little chaotically, so that we were unable to present any MSRP cards or a Founders Edition yesterday due to a logistical delay (how could we do it without a card?), but today we can deliver the full broadside. That’s fair enough, but it still presents us with completely new challenges. Unfortunately, the Founders Edition didn’t arrive until Monday afternoon, so this time there’s a slight change in the design of the content. Of course, I also ran the Founders Edition through all the benchmarks, but compared it to the MSI GeForce RTX 4080 Super Expert (which had already been tested previously), I only noticed such marginal differences in performance that a direct comparison of the two cards is completely superfluous due to the measurement tolerances that need to be taken into account. A deviation of < 0.6 percent on average across all games is not something I consider presentable. That’s why I use the results of the Founders Edition as an example (I could also have used MSI) and therefore prefer to compare the things of both cards in detail for which there are really visible and measurable differences to report. And there are quite a few.

The comparison with the GeForce RTX 4080 Founders Edition without Super is also not entirely without merit and requires a brief preliminary remark. NVIDIA has, not entirely coincidentally, included various improvements for individual games in the respective press drivers for the new cards without communicating this visibly in the changelog. If one were to assume that, due to time constraints, the reviewers would not test all NVIDIA cards once a week with the apparently still current driver from the previous week, then the reviewer has done the math without NVIDIA. Sure, in the sum of all games this is only more or less visible (depending on the game selection), but even in my case there was a difference of around 0.5 to 0.7 percent, with one game in my selection shining in terms of general performance and another with at least significantly better DLSS values. However, the metrics had all already been created and uploaded before I also tested the GeForce RTX 4080 Non-Super with the new drivers for plausibility reasons, which suddenly also performed better. Therefore, I will write something about this on the benchmark pages and put the few affected individual results, as well as the sum of all games, into perspective. However, since the majority remained plausible and I ran out of time, I can live with the “old stock” quite well. But it has to be mentioned.

Of course, as usual, there are many benchmarks for gaming and workstation, the comprehensive teardown, a very elaborate board, cooler and material analysis with some reverse engineering, as well as the analysis of the power consumption and load peaks, including a suitable power supply recommendation. As I know that many of my colleagues will also repeat all the technical details and theory that have already been presented in various snacks, I’ll save myself the trouble today and just briefly refer to the data that is already known. After all, you want to see real figures today and not PR fireworks. The specs will come later, of course. But back to the chip. Interestingly, both the Founders Edition and some other cards still use subsequently labeled chips from 2022…

… while the chip on the MSI card comes from week 44 in 2023 and is therefore a whole year newer. In the end, however, this also shows that chips have probably been collected here for over a year, which are now sold as RTX 4080 Super. This is certainly interesting, as it perhaps also explains why different chip grades require a lowest common denominator, which ultimately results in a little more voltage. But more on this later in the conclusion on the subject of power consumption. I really don’t want to speculate wildly here, because one way or another, we will probably never know.

The AD 103-400 and the new Ada architecture

On January 8, 2024, NVIDIA presented the GeForce RTX 4080 SUPER at CES 2024, an enthusiast-class graphics card based on the 5nm process. This card, which uses the AD103 graphics processor in the AD103-400-A1 variant, supports DirectX 12 Ultimate, which ensures compatibility with all modern games. The GeForce RTX 4080 SUPER, with a die area of the AD103 GPU of 379 mm² and 45,900 million transistors, offers important features such as hardware ray tracing and variable rate-based shading, which will be important in future video games.

The card is equipped with 10240 shader units, 320 texture mapping units and 112 ROPs, as well as 320 tensor cores, which increase speed especially in machine learning applications. It also has 80 ray tracing cores. NVIDIA has equipped the card with 16 GB GDDR6X memory, which is connected via a 256-bit memory interface. The GPU of the GeForce RTX 4080 SUPER operates at a base frequency of 2295 MHz, which can be increased up to 2550 MHz, while the memory runs at a speed of 1438 MHz (23 Gbps effective).