The NVIDIA RTX 5080 has been a highly misused graphics card since its launch. It’s smaller than the ADA’s predecessor, it’s a power-hungry GPU, and offers far more silicon than the RTX 4080 Super. Yes, it’s the same price on the surface, but offers about 15% higher frame rates in 4K and less at low resolution.
Ah, I wasn’t expecting this. I’ve never seen this level of overclocking headroom on a GPU released in years.
That doesn’t stop them all from selling out the moment they’re sold out. After all, the impressive performance of the multi-frame generation and its $999 sticker price are much cheaper than the $2,000+RTX 5090, and can offer a healthy generation uplift with DLSS 4 support games .
If the truth was said, I felt that the two-tier RTX Blackwell GPU was pretty uninspired. This is a great basis for future high-end gaming PCs, but in raw silicon terms the needles have hardly moved from the ADA to the Blackwell.
Well, I wasn’t quite inspired until I started overclocking this. And, oh, I wasn’t expecting this. I’ve never seen this level of overclocking headroom on a GPU released in years. There are plenty of raw performances left on the engineering room floor, making for a rather mid-way performance hike around the super-sized RTX 4080.
I’ve seen playing additional performances above 500 MHz. And everything doesn’t really put a strain on the power or thermal performance of the Founders Edition committee.
Entering the boost clock area, the smart GPU essentially changed its clock speed in situ, depending on the thermal and power headroom available, so there isn’t much left in the overclocker.
Certainly, you can offset the frequency with a touch, but not just a meaningful amount, not a level registered as something like the variance of game frame rates, especially in standard testing. Even the factory overclocking cards offered little particularly exciting. Look at the overclocked ASUS RTX 4080 Super. We would be lucky if you could deliver an additional 90 MHz.
The difference here with the RTX 5080 is that even those factory overclocking cards offer something different. The ASUS ROG Astral RTX 5080 offers +173 MHz overclocking and Gigabyte Aorus RTX 5080 A +188 MHz overclocking. For a long time, we have not seen a level of confidence from the manufacturer when it comes to OC cards.
Now people may be suggesting I won the silicon lottery – and users shouldn’t expect Nvidia to be able to hit the same 0.5 GHz boost clock bump I see As far as I want to point out – it is clear here that the GB 203 GPU has a lot of headroom.
I’m not a very special overclocker. I’ve been professionally messing around with GPUs in the best parts of 20 years, but I’m a very basic over-krutter. MSI Afterburner is my jam and I don’t go too deep. I was playing with an OC scanner that tried to adjust the voltage curve to suit the functionality of the GPU.
But you just need to go backwild with the MHZ slider. With the black mythical Woocon benchmark running, I was able to quickly get through the gear and lock the offset at +500 MHz. It didn’t give me any 3.1 GHz clock I wanted, so I slashed the GPU a little more and hit +525 MHz.
However, it doesn’t go to +550 MHz. That way you put in a pretty tentative GPU about catastrophic failure, a black screen, and what gets really high again when you reboot. I went back to +525 MHz and have remained stable ever since.
I was pretty Cavalier on the memory clock and went all the way to +1,000 MHz. So I pulled a line out of fear.
As you can see, there is an increase in performance. There’s almost 10% bump in 4K and not too far at 1440p (this is important, the resolution you’re actually running, as if you were running DLSS in quality mode in 4K).
When you combine that with the RTX 5080 Gen-on-gen bumps at the RTX 4080 Super, you’re talking about a 25% increase rather than a 15% bump. This is a much more attractive number, making the RTX Blackwell card a more attractive option. Especially because it comes with Multi-Frame Gen as an additional performance panacea, with its priced at the same price.
Of course, it’s far below the 50% or more framerate increase from the RTX 3080 to the RTX 4080, but the second RTX Blackwell card looks better than ever presented, with the RTX 5090 being the RTX Increases by 30% above 4090. It also allows the RTX 5080 to match the raw rendering power of its top ADA GPU several times.
It shifted little to the thermal or performance requirements of the GPU.
Another thing that impressed me and made me feel like this is more than just a game of numbers, overclocking, where I can legally live full-time, almost changes the thermal or performance requirements of a GPU. That’s something that wasn’t there. I was hoping for an exponential increase to get that extra 10%. But no, that hasn’t changed much.
I don’t know why so many performance remains as a standard. I only got one RTX 5080 Founders Edition, so it would be interesting to see what other manufacturers’ cards deliver on that front, but still, I can see the performances I play I’m looking forward to it.
What’s interesting here is that the upcoming RTX 5070 TI also uses the same GB 203 GPU. Certainly, it would be a rather cut-down version, but there is also a default boost clock that is far below that of the RTX 5080. If that card has the same headroom, it could be a midrange overclocking king on the top.
Still, for now, I am going to be immersed in the glory of winning the Silicon lottery. And just keep running Football Manager 2024 anyway. That means the 2025 version has been cancelled and we have to keep doing it for a while.
