Compute Units: 84
Shaders: 10,752
Ray processors: 84
AI/Tensor processors: 336
Boost clock: 2,617MHz
Memory type: GDDR7
Memory pool: 16 GB
Memory speed (effective): 30 Gbps
Memory bandwidth: 960 GB/s
Bus interface: 256-bit
TGP: 360W
Power connector: 1 x 16-pin
Slot width: Dual (Depends on AIB)
For
- Great performance
- GDDR7 and PCIe 5.0
- Slimmer form-factor
- Better cooling
- Same price as RTX 4080 Super
Against
- Still expensive
- Not a huge gen-on-gen boost
- Best performance tied to DLSS 4
- Nvidia RTX 4080 Super might be a better value
Nvidia GeForce RTX 4080 Super
Compute Units: 80
Shaders: 10,240
Ray processors: 80
AI/Tensor processors: 320
Boost clock: 2,550MHz
Memory type: GDDR6X
Memory pool: 16 GB
Memory speed (effective): 23 Gbps
Memory bandwidth: 736 GB/s
Bus interface: 256-bit
TGP: 320W
Power connector: 1 x 16-pin
Slot width: Dual (Depends on AIB)
For
- Better performance than RTX 4080
- Much cheaper than RTX 4080
- Gorgeous finish
Against
- Still very expensive
- Still power hungry
It’s been an eventful time for Nvidia, with the company recently rolling out its RTX 50 series at the beginning of 2025, just over a year after it launched the RTX 40 Series Super line. That means there are more options than ever for 80-class cards and stiff competition vying for your money.
It’s not necessarily cut and dry when it comes to the RTX 5080 vs RTX 4080 Super. While there are generational improvements and sweeping architectural differences, you may not be automatically suited to the newer of the two.
That’s why we’re comparing the RTX 5080 against the RTX 4080 Super based on their price, specs, and performance to help you choose between the two or, more crucially, see whether it’s worth upgrading from 80-class Ada to Blackwell in 2025. They’re both among the best graphics cards on the market, but which is right for your rig?
Nvidia RTX 5080 vs RTX 4080 Super: Price
One of the biggest deciding factors in choosing between the RTX 5080 and RTX 4080 Super is the price, and it’s a nuanced issue. Starting with the Blackwell card, the RTX 5080 MSRP is $999 / £979 / AU$2,019 for the Founders Edition (Nvidia-made) model. That’s in an ideal world, however, because RTX 5080 stock has been near impossible to find for its sticker price, especially for those trying to track down a coveted FE model.
The RTX 4080 Super price is identical in the US and Australia at $999 and $AU2,019. However, it’s slightly cheaper than its successor in the UK at £949, likely as a result of Team Green looking to move stock. Regardless of whether you’re after an Ada or Blackwell card, the MSRP is only half the story, as it’s highly likely you’re going to pay above Nvidia’s recommended retail price and turn to one of its partners for the GPU.
If the RTX 5080 was readily available to prospective buyers right now, we would easily hand it the win in this category. But it isn’t, and tracking down even an expensive AIB model is incredibly challenging, nearly three months after launch with no signs of improvement on the horizon. As such, considering you can actually go out and buy the RTX 4080 Super, we’re giving the Ada refresh the edge here, at least for now.
- Winner: RTX 4080 Super (for now)
Nvidia RTX 5080 vs RTX 4080 Super: Specs
| Header Cell – Column 0 |
Nvidia RTX 5080 |
Nvidia RTX 4080 Super |
% Difference |
|---|---|---|---|
|
Process Technology |
TSMC N4P |
TSMC N4 |
N/A |
|
Transistors (Billion) |
45,600 |
45,900 |
0.6 |
|
Compute units |
84 |
80 |
5 |
|
Shaders |
10,752 |
10,240 |
5 |
|
AI/Matrix cores |
336 |
320 |
5 |
|
Ray Tracing Cores |
84 |
80 |
5 |
|
Render Output Units |
128 |
112 |
14.2 |
|
Texture Mapping Units |
336 |
320 |
5 |
|
Boost Clock (MHz) |
2,617 |
2,550 |
2.6 |
|
Memory type |
GDDR67 |
GDDR6X |
N/A |
|
VRAM (GB) |
16 |
16 |
0 |
|
VRAM Bus Width |
256 |
256 |
0 |
|
VRAM Speed (Gbps) |
30 |
23 |
30.4 |
|
Bandwidth (GB/s) |
960 |
736 |
30.4 |
|
TDP (watts) |
360 |
320 |
12.5 |
|
PCIe Interface |
PCIe 5.0 x16 |
PCIe 4.0 x16 |
N/A |
In many ways, the RTX 5080 and RTX 4080 Super are remarkably similar cards when analyzing their hardware specifications, including near-identical amounts of Compute Units, Shaders, Tensor Cores, RT Cores, and Texture Mapping Units. However, it’s the more sweeping architectural differences in utilization between Blackwell and Ada that fully separate the two graphics cards.
While both the RTX 5080 and RTX 4080 Super have a 16GB memory pool on a 256-bit memory bus, the former uses faster GDDR7 memory compared to the latter’s slower GDDR6X memory.
As a result, there’s a substantial difference in bandwidth, as the Blackwell card achieves 960 GB/s compared to the Ada’s card’s 736GB/s (making the RTX 5080 30.4% faster overall). The advancement in bandwidth is possible through the new PCIe 5.0 x16 port (allowing for double the total bandwidth of the PCIe 4.0 x16 port of the older Ada GPU).
We also see this in the speed of VRAM, as the RTX 5080 achieves staggering rates of 30 Gbps, leading numbers, against the 23 Gbps effective memory speed of the RTX 4080 Super.
Make no mistake, the Ada refresh is by no means slow, but the Blackwell card runs rings around it when directly compared (at least on paper). With everything considered, the RTX 5080 is the better GPU across the board when looking at its technical specifications.
Nvidia RTX 5080 vs RTX 4080 Super: Performance
While the RTX 5080 and RTX 4080 are similar in terms of price and technical specifications, we can judge which is the better GPU based on their respective performance capabilities. Make no mistake: Both models are among the best 4K graphics cards (being primarily marketed on their gaming and productivity prowess), but there’s a clear lead here in all facets of testing.
Starting with synthetic tests, the RTX 5080 impresses with scores of 238,042 and 63,850 in 3DMark’s Night Raid and Fire Strike. In contrast, the RTX 4080 Super achieved 208,592 and 54,779 in both tests, making the Blackwell GPU faster by 14.1% and 16.5%, respectively.
It’s a similar story with PassMark 3D Graphics (Total) and GeekBench 6 Compute (OpenGL); the RTX 5080 leads with 41,625 and 285,993 compared to the RTX 4080 Super’s scores of 34,749 and 270,683 (differences of 19.7% and 5.6%).
It’s a similar story when analyzing the 4K gaming performance. The RTX 5080 leads the charge with a 75fps average in Black Myth: Wukong (Cinematic with 50% upscaling), but the results are more impressive in well-optimized games.
Cyberpunk 2077 (Ultra) was no sweat for the RTX 5080 with a 68ps average, and we also see impressive figures from Dying Light 2 (High Quality) at 185fps. The card also chews through Returnal (Epic) at 150fps average.
Interestingly, the RTX 4080 Super performs slightly better in Black Myth: Wukong, with an average of 77fps, but that’s about as far as the lead goes. It can’t quite hit 4K60 natively in Cyberpunk 2077, falling just short with a 58fps average.
However, things are comparable with Dying Light 2 at 174fps and 93fps average in Returnal. That makes the RTX 5080 faster by between 6.3% and 17.2%, barring the exception of Black Myth: Wukong, where it leads by 2.6%.
Lastly, we can look at how the RTX 5080 and RTX 4080 handle creative workloads, which can be incredibly similar. This is true in the case of PugetBench for Adobe Premiere, with respective scores of 15,206 and 15,715 in favor of the Ada refresh.
The Handbrake 1.6 4K to 1080p H.264 encode was near-identical with only a frame separating the two GPUs at 215fps and 216fps as the Blackwell card comes out on top. The RTX 5080 leads in Blender 4.30’s benchmarks, Monster, Junkshop, and Classroom, by between 1 and 8%, showing a slight improvement.
Depending on the task and the game in question, the RTX 5080 can be a much faster card, but the lead seems situational; the jump from GDDR6X to GDDR7 (and PCIe 5.0 from PCIe 4.0) can make a difference, it’s not always guaranteed to be an improvement. We see this in some games as well, being a minor bump sometimes, but a bigger jump in others.
Nvidia RTX 5080 vs RTX 4080 Super: Verdict
Overall, the RTX 5080 is a superior graphics card to the RTX 4080 Super for what (should be) the same price. If you can find the RTX 5080 for MSRP then it’s worth considering for your new machine, but that value proposition drops significantly if you’re paying over the odds.
Similarly, our recommendation comes with an asterisk if you’re thinking of upgrading from an RTX 4080 Super to an RTX 5080. The Blackwell card can be faster and more adept for gaming and productivity, however, it seems situational as to when it can pull ahead; whether a slight lead is worth an extra $999 / £949 / AU$2,019 in a year will come down to how important GDDR7 and PCIe 5.0 bandwidths are.
