Recently we had a look at the famously popular RTX 3060 Ti and how it compared to its latest successor, the RTX 5060 Ti. The Ampere based GPU that when launched in 2020 for the price of $399, was quickly snapped up by gamers for being pretty potent for the time, but still showed that it can keep up in 2025 and made for an interesting comparison to the latest Blackwell based GPU, as while the 3060 Ti can still hit some impressive performance numbers in the latest games, it also painted a clearer picture as to what the upgrade path could be for those on the ageing hardware, or based on our summary, what the upgrade path potentially shouldn’t be.
Sticking with the same theme, I wanted to move up a tier to the RTX 3070 and to see how that stacks up against the RTX 5070 and again to look at if there’s an argument for upgrading 5 years later on, especially given that the RTX 3070 packs just 8GB of GDDR6 memory on a 256-bit bus, while the newer Blackwell GPU does increase that to 12GB of GDDR7, but on a slightly nerfed memory interface of just 192-bit, though of course, due to the faster iteration of GDDR, we do see an increased bandwidth of 672GB/s compared to just 448GB/s on the RTX 3070.
Beyond that, we do see increases in some of the core specs, and while architecture will make a big difference, it’s all about the uplift of what that translates to percentage wise to see if it earns your hard earned cash, especially considering the RTX 5070 launched for an extra $50 when we’re comparing MSRP to MSRP, though you can get base models for a little under the $549 launch price, now sitting at $529 and up.
Now obviously we’ve reviewed the RTX 5070, so we’re not wanting to go over the same things we said in our launch day review, though for the most part, I will say, much like the rest of the RTX 50 series, we thought while the new tech is good in terms of upscaling and raytracing, for those wanting a pure rasterisation bump in performance, it was definitely on the lower side of what I believe the average consumer would expect, and that memory configuration for $549, just wasn’t giving us the wow factor.
To see how things stacked up, we put both the RTX 3070 and RTX 5070 onto our GPU test system with an AMD Ryzen 7 9800X3D and 32GB of Corsair Vengeance RGB 6000MHz CL30 memory on a Gigabyte B650E AORUS Master motherboard. All testing was done on the latest version of Windows with the latest updates and for our testing, we used the latest NVIDIA 581.80 GameReady driver. We’ll be looking at 1080p, 1440p and 4K results today, and looking at the performance increase that you’d get going from an RTX 3070 8GB card, and moving up to an RTX 5070 12GB GPU. We’ll also be putting all of our test results over on our Patreon, so if you want to check them out while also supporting everything that we do, check it out here. So with that out of the way, let’s get into those glorious benchmarks.
