Cryio
Member
1 TF from one uArch isn't equal to 1 TF from a different uArch.Found this benchmark of all PC GPUs at PS4 settings.
Those numbers look great until you realize the 6800xt is a 20 tflops card with a 50% IPC gain over the original PS4 offering essentially 15x more power over the original PS4 GCN gpu.
The boost you are getting is less than 6x. The 2070 Super and the 5700xt shows just how poorly this game performs on AMD cards, but the 2070 Super should be able to do far better than just 115 fps using PS4 settings at 1080p.
NX Gamer's 2070 sits at 70% utilization at higher framerates because the game is CPU bound and his ryzen 2700 is a shit CPU, but that is a very interesting comparison because that CPU is roughly equivalent to the PS5 CPU in terms of performance.
Let's say a PS4 is roughly equivalent to a HD 7850. An RX 580 is a bit over twice faster. An RX 5700 TX is about twice faster than a 580. A 6900 XT is about twice faster than a 5700 XT at 4K.
So the 6900XT is a bit over 8X as fast as a base PS4. Maybe 9-10x, but that's stretching it. Nowhere near 15-16x, lol.
Besides raw horsepower, due to various uArch improvements, the final performance number can vary. Is the game more geometry heavy? More tessellation heavy? More compute heavy? More bandwidth heavy? Not all things scale linearly the higher up you go, so that's why you don't always see a 1:1 scale or BETTER than 1:1. It's usually much, much worse than 1:1
Last edited: