CEOInterviews.AI
Start App
Get an email when Steven Walton gives a new interview
One email per new interview. Unsubscribe in one click.
Steven Walton
Co-founder & Host, Hardware Unboxed, Hardware Unboxed

I Tested DLSS 5's Real Performance (No 6X Frame Gen)

📅 Sep 03, 2026 Hardware Unboxed 20 MIN 350616 VIEWS 36 SEGMENTS · 2 SPEAKERS
Support us on Patreon:   / hardwareunboxed   YT Membership:    / @hardwareunboxed   We tested DLSS 5 performance in NBA 2K27 using proper techniques (no 6X frame generation) at three resolutions (4K, 1440p and 1080p) across the seven main RTX 50 series GPUs. Buy relevant products from Amazon, Newegg and others below: GeForce RTX 5090 - https://geni.us/JyDFBvM GeForce RTX 5080 - https://geni.us/dRy7y07 GeForce RTX 5070 Ti - https://geni.us/AuxCu GeForce RTX 5070 - https://geni.us/Wg4vPDv GeForce RTX 5060 Ti 16GB - https://geni.us/5Kgs2I GeForce RTX 5060 - https://geni.us/xmaeB Radeon RX 9070...
Steven Walton 0:00 ↗
Earlier this week, Nvidia released official first-party benchmarks showing how DLSS 5 performed in NBA 2K27. But these benchmarks were pretty laughable because Nvidia used extremely high levels of frame generation to boost the output FPS numbers. 6x frame gen was used in every single chart for every single result, and we didn't really get proper before and after numbers showing the direct impact of enabling DLSS 5. I'm fixing all of that today by showing you the actual performance impact of DLSS 5 in NBA 2K27. No 6x frame gen, proper on versus off testing the official version of DLSS 5 across the full RTX 50 series lineup. Yes, that includes the RTX 5050.
I've only had a little over 24 hours of time with the official implementation of DLSS 5 so far through early access to NBA 2K27 and Nvidia's supported driver. That's why in this video I'm only focusing on DLSS 5 performance. I haven't gotten to any image quality analysis yet. You'll see that from me sometime next week if everything goes to plan. But given DLSS 5 will be available in NBA 2K27 at, I think it's the moment this video goes live, I thought it was at least worth exploring performance and why DLSS 5 performs the way it does, because there are some quirks here.
In-game performance using DLSS 5 is largely dictated by two significant factors. The first is that DLSS 5 is extremely performance intensive to run. The neural network pass which post-processes the final rendered frame takes a very long time to complete relative to other effects and shaders that you'd be familiar with. And this is a fixed cost based on the resolution. Lowering in-game settings does not make the DLSS 5 portion of the render time faster. What this means in practice is that when you enable DLSS 5, the performance you get is capped by having to run this intensive effect. And changing game settings doesn't have a significant impact.
On an RTX 5090 running NBA 2K27 using DAA and max settings, I achieved just 76 FPS with DLSS 5 enabled. When dropping the preset to high, performance only increased by 2 FPS, less than a 3% gain. This is despite a 5% gain to FPS when DLSS 5 is disabled. And when dropping all the way to the low preset, we run directly into what appears to be a 240fps engine cap in the game, at least in its single player mode, when DLSS 5 is disabled. But when DLSS 5 is enabled, we only improve performance by 9 FPS in total compared to max settings, just a 12% increase. There's not a lot of scalability here at all. It's pretty much a case where the performance you get with DLSS 5 enabled is what it is.
DLSS 5 performance is also tied directly to the output resolution, not the render resolution. This is because DLSS 5 processes the final rendered frame after upscaling, not before upscaling. There have been some modded DLSS 5 implementations that switch the DLSS 5 pass before upscaling. But in the official implementation, it runs after upscaling and therefore enabling upscaling has little to no impact on performance. Here we have the RTX 5060 Ti 16GB running at 1440p, which I chose to try and avoid the 240fps cap and potential CPU limitations. What we see in this title with DAA enabled is performance goes from 129 FPS without DLSS 5 to just 46 FPS with DLSS 5 enabled. That's a pretty huge performance loss.
And normally, if I was getting 46 FPS with DAA, I would just enable DLSS Super Resolution on something like the quality or balance settings to drastically improve FPS. But that's not what happens when DLSS 5 is enabled. Using DLSS quality upscaling, performance improved to just 50 FPS, a measly 9% increase. And I saw no further gains using any lower upscaling settings. In this title, 50 FPS is the hard limit for the RTX 5060 Ti 16GB at 1440p using DLSS 5.
I also saw a bit of a strange limit with DLSS 5 disabled. Performance never exceeded 165 FPS. So, there also wasn't any scaling below quality upscaling, but crucially without DLSS 5 enabled, we do see a 28% frame rate increase going from DAA to DLSS quality. That sort of gain is heavily reduced using DLSS 5 because the fixed processing cost of DLSS 5 dominates the total render time. This render time is not reduced using upscaling because DLSS 5 is processed after upscaling. So when using upscaling, you can only reduce the non-DLSS 5 render time, which becomes a smaller part of the total frame time when DLSS 5 is on.
Essentially, what this all means is that there is little you can do to improve DLSS 5 performance on your GPU when using a native output resolution. It's completely limited by the resolution of your display. And while you can improve performance outputting a non-native resolution, that is a terrible way to play as it will look very blurry. I'm sure many of you have tried outputting a 1080p game on a 1440p monitor and have seen how bad that looks. That sort of thing is really the only way to reduce DLSS 5's render time to improve performance. With all of that said, let's take a look at how DLSS 5 performs across a range of GPUs.
Starting with Nvidia's workstation model, the RTX 5090. If you happen to own one of these cards for 3D rendering or running local AI models, you can get a reasonably playable experience with DLSS 5 enabled. At 4K max settings, as you've seen previously, I was able to achieve 76 FPS without any frame gen enabled. At 1440p, this improved to 134 FPS and again to 182 FPS at 1080p. Now, of course, these results are playable at all resolutions, but it's still a huge performance impact compared to not using DLSS 5. The RTX 5090 ran NBA 2K27 60% slower at 4K and 40% slower at 1440p, which is especially substantial at 4K given the difference is 192 FPS versus 76 FPS. A massive change in latency and smoothness.
But running DLSS 5 on a $5,000 GPU is not that interesting because this really isn't a gaming GPU at those prices. It's well out of reach of most gamers unless you got lucky and purchased one near the MSRP well over a year ago. So instead, let's take a look at the RTX 5080. Now on this GPU, gaming at 4K using DLSS 5 is not really possible at acceptable frame rates. I achieved just 53 FPS, down from 185 FPS with the feature disabled, a 71% reduction in performance. This is due to DLSS 5 taking quite a while to process the 4K frame on an RTX 5080. It takes long enough that 60 FPS in any game will be difficult to achieve, and I'll explore the render time later in this video.
1440p though is playable, outputting 102 FPS, which is very sufficient in a game like NBA 2K. This is still a 54% reduction in performance, so a massive hit. But when starting from a baseline of 221 FPS, it's less painful. At 1080p, DLSS 5 ran at 142 FPS. Though I'm not sure how many people are pairing an RTX 5080 with a 1080p display. And remember, DLSS 5 is tied to the output resolution, so you can't just enable upscaling with a 1080p render resolution and expect to get that sort of frame rate.
The RTX 5070 Ti struggles even more with 4K DLSS 5 gaming, outputting just 44 FPS. Given the small gains we saw using lower settings, and even here I'm using DLSS quality upscaling to begin with, it's pretty much impossible to hit 60 FPS on this card. The RTX 5070 Ti only becomes usable with DLSS 5 enabled at a 1440p resolution where it renders 86 FPS. That's a 62% decrease compared to DLSS 5 disabled. It also ran at 126 FPS at 1080p, down 45% compared to the feature disabled.
Then we get to the RTX 5070. It should be no surprise given the 5070 Ti results that the 5070 is useless at 4K when DLSS 5 is enabled. The game is actually quite playable with DLSS 5 disabled at this resolution. I was getting 124 FPS with max settings and DLSS quality upscaling, which is a really enjoyable experience. But 33 FPS with DLSS 5 is a complete no-go. 1440p is also a bit of a stretch here, though it is playable, outputting 67 FPS. That's down from 217 FPS with DLSS 5 disabled. So, a near 70% reduction in performance. At 1080p, we scrape past the 100 FPS barrier, but we're still seeing a significant 57% drop in performance.
The RTX 5060 Ti 16GB is again actually very usable at 4K in this game when DLSS 5 is disabled, though you can completely rule out DLSS 5 at that resolution. It's a terrible experience. Even 1440p is a stretch. We go from 165 FPS down to 50 FPS with DLSS quality enabled, a 70% drop. It is technically possible to get 60 FPS here. The render time of DLSS 5 doesn't exceed the frame time interval of 60 FPS, but as we'll see shortly, it will be a big ask to hit that frame rate. The 5060 Ti 16GB is more suited to 1080p with DLSS 5, achieving 78 FPS here, though this is a 65% performance reduction compared to 1080p without DLSS 5.
Then we get to the 8GB models. Starting with the RTX 5060, I had to turn the game settings down to low to avoid a VRAM warning, but even then, DLSS 5 was largely unplayable on this GPU. 4K is completely out of the question at just 21 FPS, and at 1440p, I achieved just 46 FPS with no realistic path to 60 FPS gaming given the render time of DLSS 5. At best, you can use DLSS 5 at 1080p, achieving 73 FPS here, but that's a 70% performance reduction compared to not using DLSS 5, and the drop would have been even larger if the game wasn't just maxing out the 240 FPS cap in this configuration.
The RTX 5050 is completely useless at DLSS 5. This is a terrible GPU, and even on low settings, the game failed to achieve 60 FPS with DLSS 5 enabled at a 1080p resolution. This of course means it can't do it at 4K or 1440p either. In fact, it doesn't even get close, rendering just 33 FPS at 1440p despite outputting 152 FPS with DLSS 5 disabled. Even at 1080p, it's 75% slower when DLSS 5 is enabled. So, you're getting just a quarter of the frame rate.
With frame rate results before and after enabling DLSS 5, we can now figure out DLSS 5's render time on each GPU. This is calculated by converting frame rates to frame times and calculating the difference in frame time between DLSS on and off. This tells us how much additional time it took to render the frame due to DLSS 5. Note that for this testing, for some GPUs, I increased settings to DAA to avoid bottlenecks compared to the previous charts that you just saw. And the RTX 5090 is still bottlenecked at 1440p, so that render time is an extrapolated estimate.
When looking at just the isolated render times for DLSS 5, it's interesting to discover that this feature's cost is almost directly tied to output pixel count and the tensor performance of the GPU. On average, it took 2.2 times longer to render DLSS 5 at 4K than at 1440p across all GPUs. 4K has 2.25 times the pixels of 1440p, so we're getting near-perfect scaling there. Nvidia also claims that, for example, the RTX 5080 has 2.4 times the tensor core AI TOPS of the RTX 5060 Ti. At 1440p and 4K, it rendered DLSS 5 2.3 times faster based on my performance testing.
With these render time numbers, we can figure out what the sort of performance outcome may be in different situations. NBA 2K27 runs at a really high frame rate without DLSS 5 most of the time, so that is, I guess, a more favorable configuration, but this feature eventually will be available in games that will run at lower frame rates. So, what we've got here, the first number is the estimated DLSS 5 frame rate output when you were previously running the game at 120 FPS without DLSS 5 enabled. So, for example, on an RTX 5090 at 4K, DLSS 5's processing time is about 7.9 milliseconds. At 120 FPS before enabling DLSS 5, that means you were rendering the rest of the game at 8.3 milliseconds previously.
Therefore, if you enable DLSS 5 in that state and add 7.9 milliseconds to the total render time, you have a total frame time of 16.2 milliseconds, which equates to 62 FPS. Across all the GPUs, you can see how much of a frame rate hit DLSS 5 will have in situations where you were previously getting a pretty decent frame rate. The RTX 5090 holds up okay at 4K, but the rest of the GPUs are pretty useless. At 1440p, you get more than a 50% hit on the RTX 5070 and an even higher hit on the lower tier cards as it takes longer to process DLSS 5 at this resolution on those models.
I've run the same calculations for a game frame rate of 80 FPS prior to enabling DLSS 5. At 4K, this won't be a high enough starting frame rate on all GPUs, while at 1440p, you'll also be on struggle street for the most part. The RTX 5090 scrapes over the line here, but the RTX 5080 won't hit 60 FPS in that configuration, though the performance drop will only be 33%.
The other numbers I have in this table are the required game frame rate to hit 60 and 80 FPS with DLSS 5 enabled. This is the performance state the game will need to be in via setting tweaks or upscaling prior to adding the processing time of DLSS 5. At 4K, you can see that pretty much all of the GPUs will never hit 60 FPS with DLSS 5 enabled. This is because the DLSS 5 render time exceeds 16.7 milliseconds. The RTX 5080 just scrapes through and 60 FPS is possible, but only if the rest of the game renders in just 3 milliseconds or so, which means a starting frame rate of 330 FPS.
At 1440p, playable frame rates are possible on a wider range of graphics cards. The RTX 5080 requires a starting 95 FPS to achieve 60 FPS after DLSS 5 and 157 FPS to achieve 80 FPS. The RTX 5070 Ti is also reasonable here, provided you are okay with 60 FPS and a performance hit around 50%. 80 FPS is also possible on that card at a 62% hit. Below that, we're in the realms of implausibility. The 5070 can sort of do 1440p 60 FPS at a larger to the frame rate, but anything more than that will be difficult. And the RTX 5060 Ti is pretty much out of the question for DLSS 5 at 1440p.
Nvidia claims they are continuing to work on DLSS 5's performance, so possibly this render time will be reduced with further optimizations. We'll have to wait and see on that one. The launch version though, as things stand, there are just some GPUs where no matter what in-game configuration of settings you use, you will never achieve a playable frame rate using this technology at your display's native resolution.
Now, I was told a couple of interesting things since the announcement of DLSS 5. One very contentious issue at launch is that DLSS 5 is restricted to the RTX 50 series, even though modders already got this feature working on all RTX GPUs, right down to the 20 series. The RTX 40 series is an especially painful restriction given it seems those GPUs are powerful enough to run DLSS 5. The RTX 4090, for example, is faster than the RTX 5080, which can run DLSS 5 in some configurations at playable frame rates.
Nvidia tell me they are planning to work on expanding official support to the GeForce RTX 40 series once RTX 50 series performance is more fully tuned, whatever that means. But it seems like they are going to bring this feature to at least the 40 series at some point in the future, which is great news for owners of those cards disappointed in the current restrictions. I'm not sure whether Nvidia intended to support the 40 series from the start based on how they typically operate. I suspect this was designed to be a 50 series exclusive to sell people on those GPUs, but that's just speculation on my part.
In any case, plans for that absolutely had to be scrapped once modders got DLSS 5 working fine on the 40 series because there is no genuine technical limitation preventing support. Of course, Nvidia have not confirmed official support for the RTX 30 and 20 series. This makes sense to me from a couple of perspectives. Firstly, DLSS 5 is designed to use FP8 instructions which are hardware accelerated on Blackwell and Ada architectures. RTX 30 and 20 series lack hardware FP8 support. So modders had to switch DLSS 5 over to FP16 to get it working at more than just a few FPS.
But even then, even if Nvidia did that, the RTX 3090 is only around the performance of an RTX 5070, which can barely run DLSS 5 at acceptable frame rates as is with the current model using native FP8. Not providing support for DLSS 5 on older cards makes sense if it won't run well no matter what.
Overall though, the performance we're seeing here, it's pretty much in tech demo sort of areas. Sure, if you have a $5,000 GPU, you can get a playable experience at 4K and 1440p with DLSS 5 enabled. Outside of that, it's increasingly difficult. The RTX 5080 struggles at 4K, while the RTX 5070 is only borderline usable at 1440p and sees performance reduced considerably compared to not using the feature. The performance I'm seeing here on most models suggests to me that most gamers won't use this feature while actually playing games. Maybe for screenshots and to check it out, but not for actual gameplay.
And I say that not just because of low frame rates, but also because improving the frame rate using DLSS 5 is really difficult. DLSS 5's processing time dominates overall frame times. So lowering in-game settings and using upscaling has little impact. This is where DLSS 5 differs quite substantially from a technology like path tracing, which is also very heavy on GPU hardware. Upscaling works wonders to improve path tracing performance because the render resolution of path tracing is reduced when upscaling is enabled, giving a hefty performance boost.
You don't always get playable frame rates depending on the hardware you have, but there's some flexibility there to take, say, a 40 FPS path tracing experience and push it over 60 FPS. With DLSS 5, there is little to no flexibility. With the current implementation, if you only get 40 FPS with DLSS 5 enabled on max settings, it will be practically impossible to achieve 60 FPS no matter what settings you adjust.
The performance I'm seeing here, you know, it doesn't make me overly excited to use DLSS 5. I personally love practical features that everyday gamers can use to improve the gaming experience. DLSS Super Resolution is one such example that works great on high-end and low-end hardware. DLSS 5 though, it just doesn't fall into that category right now, and I find it hard to become enthusiastic about a feature you basically need an RTX 5090 to use.
With all of that said, I'm yet to dive into the image quality benefits that DLSS 5 brings or really do any visual comparisons whatsoever. As I said at the start, I've only had enough time with DLSS 5 to do some performance benchmarking. My full analysis of DLSS 5 will be coming later on once I've had a lot more time with it in NBA 2K27. So anyway, that's it for this one. As I said, just a look at the performance of DLSS 5. If you do want to use this feature and you happen to own a copy of NBA 2K27 and an RTX 50 series GPU, hey, you can go use it right now if you want to.
But yeah, the performance hit is pretty significant and there's not a lot you can do to improve that. So yeah, have fun with that. Anyway, that's it for this one. If you want to support Hardware Unboxed, you can do so via our Patreon. Links to that is in the description as usual. But anyway, thanks for watching and I'll catch you in the next one.

1 more exchange in this transcript

Sign in free to read the rest of this interview. No card required.

Sign in to read the full transcript

Cite this transcript

APA, MLA, BibTeX
APA

Walton, S. (2026, September 3). I Tested DLSS 5's Real Performance (No 6X Frame Gen) [Interview transcript]. Hardware Unboxed. CEOInterviews.AI. https://ceointerviews.ai/interview/3229060/

MLA

Steven Walton. "I Tested DLSS 5's Real Performance (No 6X Frame Gen)." Hardware Unboxed, 3 Sep. 2026. Transcript, CEOInterviews.AI, https://ceointerviews.ai/interview/3229060/.

BibTeX
@misc{walton2026_3229060,
  author       = {Steven Walton},
  title        = {I Tested DLSS 5's Real Performance (No 6X Frame Gen)},
  howpublished = {Interview transcript, Hardware Unboxed. CEOInterviews.AI},
  year         = {2026},
  month        = {sep},
  url          = {https://ceointerviews.ai/interview/3229060/},
  note         = {Speaker-attributed transcript with timestamps}
}