Our review coverage of the GeForce RTX 4080 Super includes nine cards: ASUS RTX 4080 Super STRIX OC, ASUS RTX 4080 Super TUF OC, Galax RTX 4080 Super SG, Gigabyte RTX 4080 Super Gaming OC, MSI RTX 4080 Super Expert, NVIDIA RTX 4080 Super Founders Edition, Palit RTX 4080 Super GamingPro OC, PNY RTX 4080 Super Verto OC, Zotac RTX 4080 Super AMP Extreme Airo.

With the GeForce RTX 40 Super series, NVIDIA is refreshing their Ada-based lineup for 2024. At CES Las Vegas, the company announced three new SKUs: GeForce RTX 4070 Super, RTX 4070 Super Ti and RTX 4080 Super. While we reviewed RTX 4070 Super two weeks ago, and RTX 4070 Ti Super last week. Today, the final review embargo has lifted, for the GeForce RTX 4080 Super. Unlike previous launches, both MSRP and non-MSRP reviews go live at the same time and the first cards should be on-shelf now.

Architecture
RTX 4080 Super is built using the same AD103 GPU as the RTX 4080 non-Super. This means that the RTX 4090 is still the only gaming GPU that's using the more powerful AD102 graphics chip. While the RTX 4080 came with 9,728 cores on the chip, the RTX 4080 Super runs the full shader count available in the GPU—10,240, or +5.3%. NVIDIA hasn't increased card's power limit of 320 W, but the rated boost frequency has received a 1.8% clock bump, to 2550 MHz, up 45 MHz from 2505 MHz on the RTX 4080 non-Super. Last but not least, the memory is running at 23 Gbps now, vs 22.4 Gbps (+2.7%). The VRAM size is still 16 GB, connected using a 256-bit memory bus, and the L2 cache is identical, too, at 64 MB.

MSI GeForce RTX 4080 Super Expert is a new addition to the company's lineup. The new card uses a design philosophy similar to the NVIDIA Founders Edition—and it looks gorgeous, too. MSI is used an all-metal coolers design paired with smoothly curved, matte surfaces that look fantastic. You also get a factory overclock to a rated boost of 2610 MHz, which is a 60 MHz increase over the NVIDIA baseline, or +2.4%.

Performance
Averaged over the 25 games in our test suite, at 4K, we find the NV Founders Edition only 1.5% faster than 4080 non-Super Founders Edition, which is MUCH less than expected. While NVIDIA never said "+5%," I definitely expected more. It's not a power limit issue, running at max power yields another +1% only. These numbers are pretty constant across resolutions, and even with RT enabled, too. Looking at the individual games, the differences are 1 or 2 FPS, nothing you'd ever notice subjectively. While that's certainly a bit disappointing, the fact remains that RTX 4080 Super, just like the RTX 4080 non-Super, is a fantastic card for gaming. In a pure raster scenario, the AMD Radeon RX 7900 XTX is still a tiny bit faster, but NVIDIA's card makes up for that with much better RT performance. Still, I feel like beating the XTX across the board was one of the goals of RTX 4080 Super, and NVIDIA failed here. In NVIDIA's defense, the RTX 4080 Super leverages the full GPU, no units were held back, even though small additional gains could have been achieved with higher clock speeds and power limit. Despite its factory overclock, the MSI Expert ends up slightly slower than the NVIDIA Founders Edition, but the differences are minimal. There is always some random variances in GPU silicon which can affect the card's performance in small ways—nothing you'd ever notice subjectively though, probably not even with an FPS counter unless you run two cards side-by-side on a controlled setup like we do. Compared to the RTX 4070 Ti Super, the performance uplift is 18%, the card is also 28% faster than RTX 4070 Ti non-Super. RTX 3080 Ti performance is roughly comparable to RTX 3090, which makes the gen-over-gen uplift a very solid 28%. NVIDIA's flagship, the RTX 4090 is still the undisputed king of the hill, offering almost 30% better

Taking a closer look at our test results, most of the games are in the 1.5% range—it's not that one game is 7% and the others are 0%, the gains are really small across the board. There is definitely some variations in GPU chip silicon quality, which affects performance due to the way NVIDIA's Boost algorithm works, but with my sample size of 9 cards tested, I think that I can confidently say that there's no way you're getting +5% out of the 4080 Super, unless you manually overclock it. Other reviewers have similar numbers, so it's not just me. Surprisingly, the RTX 4070 Super, initially considered the least exciting among the new releases on forums, showed the highest performance increase (+15%). In contrast, the RTX 4070 Ti Super, despite receiving a GPU upgrade and a 33% memory bandwidth increase, only saw a modest +5% in extra performance. Today's release could almost be labeled a "rebrand," but the improved pricing certainly adds to its appeal.

Optimum Resolution, Upscaling
No doubt, if you desire a great 4K experience, then RTX 4080 Super is what you want. In our testing the card reached well over 60 FPS at 4K with maximized settings and no upscaling. While it's a bit overkill for 1440p, I could still see people with high-refresh monitors opting for it, even though I'd probably buy a better monitor first, before spending $1000 on a graphics card. On the other hand, this could be seen as some sort of future proofing. Just like the other GeForce 40 cards, RTX 4080 Super has support for all of NVIDIA's DLSS technologies: NVIDIA DLSS 2 upscaling, DLSS 3 frame generation and DLSS 3.5 ray reconstruction. On top of that you can enable AMD FSR 2 and FSR 3 in games, because those technologies work on all GPUs from all vendors. Basically this means that you'll be covered in terms of upscaling and frame generation. While DLSS 3 is definitely the leading solution right now, with the best game support, AMD is pushing hard and their frame generation solution will come to several major titles in 2024. From a technology perspective, DLSS 3 is superior, because it uses the optical flow hardware unit in Ada GPUs, and NVIDIA Reflex will help bring down the input latency.

VRAM & Memory Bus
GeForce RTX 4080 Super comes with 16 GB VRAM, which makes a lot of sense in this segment, even though I feel a bit more could be useful in the future. While AMD offers 20 GB on the RX 7900 XT and 24 GB on the XTX, the extra memory cannot make a difference, not even in our most memory intensive test (Alan Wake 2 RT at 4K). If VRAM was something that you could scale up and down like clock frequency then I would have loved to see 20 GB, but that's easier said than done. VRAM size is tied to the number of memory chips installed on the card. These come in 2 GB typically, which means for 20 GB you'd have to install 10 chips, which requires a 320-bit wide memory interface in the GPU (10x 32 bit). The memory interface on AD103 supports only up to 256-bit, so NVIDIA would have had to use the much bigger (= more expensive) AD102. Using 4 GB chips, which aren't available in any significant volume would mean five chips, using a narrow 160-bit bus—I'd prefer 256-bit 16 GB over that any day. Technically that does leave room for a RTX 4080 Ti Super, with AD102, 20 GB and 320-bit, maybe something for the $1200-$1400 price range that just opened up with 4080 Super.

RT
As expected, ray tracing works very well on the GeForce RTX 4080 Super, clearly offering a superior experience than what any AMD card can achieve. For example, RX 7900 XTX is over 20% slower with RT enabled. That's because the dedicated RT units in AMD's cards are weaker, so more tasks are offloaded to the GPU shaders. While I have to admit that RT isn't as crucial in the lower performance segments, where pricing is much more relevant, I feel that RT is very important in this price range. You're spending a lot of money, which means you want to be able to dial the settings up as high as possible, for maximized eye candy.

GPU Compute
With our new 2024 test suite, we are adding a section focused on testing GPU Compute, which is becoming more and more important every day. Emerging AI Technologies like ChatGPT, Stable Diffusion and others are transforming the world. While the use of compute on the consumer desktop is limited today, it will grow considerably. For our first round of testing we've picked three real-life workloads that allow us to get a feel what to expect. Here, NVIDIA is the clear leader with a substantial advantage over both AMD and Intel. It's not only about performance, but also about the software ecosystem, which is much more advanced on the NV side, but the other players are working hard to catch up, I'm sure.

Physical Design, Heat & Noise
While MSI's Expert design looks awesome, the cooling performance is surprisingly low. I'm not sure what the exact issue is, but I expected more from the design. Our noise-normalized cooler comparison test shows that at the same noise level and heat load as the NVIDIA FE, the cooler runs 5°C warmer. That's not a lot, but it's still surprising, given the design. On the other hand, we've seen cards today with seven, eight and nine heatpipes, which makes the six on the Expert seem a bit low. MSI made the right choice to not try to match the FE temperatures, but rather match its noise as that provides more of a tangible benefit for the gamers. Temperatures are slightly increased vs the FE, 69°C vs 63°C—insignificant for daily operation and still very cool. I did try reapplying the thermal paste and changing the orientation of the card while running—no significant difference. As expected for a modern graphics card in 2024, the fans will stop spinning when not gaming, for the perfect noise-free experience.

Power
Power efficiency of the RTX 4080 Super is just as amazing as the RTX 4080 non-Super. The card offers fantastic performance, yet consumes only around 300 W during gaming, which means PSU requirements are minimal, and this helps with heat and noise, too, because the cooler has to work less hard. AMD's Radeon RX 7900 XTX is a bit more power-hungry with 350 W, but it's not a night-and-day difference, neither for your power bill, nor for PSU capacity planning. While there is still some controversy around the 12-pin power connector, all RTX 4080 Super cards come with it, and I'm a big fan, even though I'm not sure if the location in the middle of the card is the best possible choice.

Overclocking
There is not much to report on overclocking, other than it works, is easy to do and will yield you around 6 to 7%, which is the typical range that we've been seeing from most cards in recent years. With this review we've switched our OC testing to use maximized power limits instead of the default, but it really doesn't make much of a difference, RTX 4080 Super is not power starved. While there's some MSRP cards that don't allow any manual power limit adjustments, MSI offers a decent range, up to 370 W.

Pricing & Alternatives
NVIDIA has set an MSRP of $1000 for the RTX 4080 Super and that's the real innovation here. Compared to the current $1200 for the RTX 4080 non-Super, this introduces a significant 20% discount. I have to applaud NVIDIA for that, especially, considering that there's not much competition in this segment. AMD is happy with their $970 price point for the 7900 XTX, but that changes today. RTX 4080 Super at $1000 means that RX 7900 XTX becomes unsellable unless its price is lowered considerably. The RTX 4080 Super offers superior RT performance, similar raster perf and support for DLSS—exactly what people in the premium segment are looking for. Even at $900 I'm not sure if I'd prefer 7900 XTX over a $1000 4080 Super, it's just a 10% delta. Still, $1000 is definitely not a steal for the RTX 4080 Super and what it offers—the 2024 GPU market is still expensive.

MSI's price point for the RTX 4080 Super Expert is $1150, which is a pretty steep $150 increase over the Founders Edition. That's simply too expensive—you can get the ASUS TUF OC for $1100, the ASUS TUF non-OC for $1000. While thermal performance could be better, the looks of the Expert are outstanding and rival the NVIDIA Founders Edition. Still, unless the card is sold much closer to $1000 there's too many alternatives that offer better value. If you can, however, find the MSI RTX 4080 Super Expert at competitive pricing, then it's definitely a good choice that will offer excellent gaming performance.

If you want to save a bit of money, probably the most interesting alternative is RX 7900 XT, which currently sells for $710, but is considerably slower, which means lower detail settings or upscaling, but there's no DLSS on the card to help with that. NVIDIA has confirmed that the RTX 4080 non-Super is now end-of-life. You could potentially get a card at a good price; anything $950 and below is what I'd call "interesting." If it's higher, go for the Super model, also for its better resale value. If you really must have the best, then the RTX 4090 is what you want—that hasn't changed with the release of the RTX 4080 Super, but be prepared to pay for it: +80% for an almost 30% increase in performance is tough. At the end of the day, RTX 4080 Super is disappointing in terms of the changes it brings, but it redeems itself thanks to its greatly improved pricing. While I'm sure there will be a lot of drama about the minimal gains, what the GPU market really needs is lower prices, not marginally better performance for the same price. In this regard, the RTX 4080 Super can be considered a success.