I already wrote about this issue in other threads but I'm gonna make my own in the hope that something gets done about it. Unlikely, seeing as not a lot of people own Vega 56 but I have to at least try to have clean conscience.
Basically when patch 1.6 arrived, my fps took a nosedive. Since then I changed my CPU from Ryzen 3600 to 5600 but it didn't do anything. I'm aware graphical presets have changed slightly. Below I'm pasting an image with my testing results from CapFrameX. Custom identical settings in both builds. Data from the benchmark but also from a ride around the city (64 seconds long).
In-game average CPU usage in 1.6 is lower (41% vs 56%) but most importantly GPU power draw is way lower (145W vs 196W). Very odd (Checked at the wall too).
Basically, on these settings, older 1.52 build has 50% more average FPS and 66% more 1-percentle FPS. Sorry, but that's not normal.
Some more recent examples at different settings in current build (1.61 DLSS-Reflex). FPS reading took right after loading a save from downtown (most demanding):
low settings: 76 fps (1.52) -> 42 fps (1.61)
med settings: 60 fps (1.52) -> 35 fps (1.61)
max settings: 33 fps (1.52) -> 23 fps (1.61) - power draw from ~240W -> ~190W
I probably tried everything you can think of (older drivers, DDU, AMD cleanup utility, fresh windows 10 and 11, restore default game settings, some tool called MPOGPUFIX, I think I even rolled back to older BIOS). And probably many other things I can't be bothered to remember but if you have any non-time-consuming ideas, go right ahead.
I might send these numbers to AMD but then again all other games work fine, 1.52 works fine. AMD didn't change anything. It's the game that got "updated". But hey, maybe AMD will know what's happening and contact you. One can hope.
I did the first playthrough on 1.52, heavily modded. But all the readings here are done without any mods installed.
My build: Ryzen 5600, Sapphire Nitro+ Vega 56, 16GB 3733 MHz CL16 DDR4 with tight sub-timings. Up-to-date Windows 10, 2560x1080, game installed on NVMe SSD.
EDIT: Corrected CPU load and GPU power draw numbers to match the numbers from attached image (must have been numbers from older measurements).
Basically when patch 1.6 arrived, my fps took a nosedive. Since then I changed my CPU from Ryzen 3600 to 5600 but it didn't do anything. I'm aware graphical presets have changed slightly. Below I'm pasting an image with my testing results from CapFrameX. Custom identical settings in both builds. Data from the benchmark but also from a ride around the city (64 seconds long).
In-game average CPU usage in 1.6 is lower (41% vs 56%) but most importantly GPU power draw is way lower (145W vs 196W). Very odd (Checked at the wall too).
Basically, on these settings, older 1.52 build has 50% more average FPS and 66% more 1-percentle FPS. Sorry, but that's not normal.
Some more recent examples at different settings in current build (1.61 DLSS-Reflex). FPS reading took right after loading a save from downtown (most demanding):
low settings: 76 fps (1.52) -> 42 fps (1.61)
med settings: 60 fps (1.52) -> 35 fps (1.61)
max settings: 33 fps (1.52) -> 23 fps (1.61) - power draw from ~240W -> ~190W
I probably tried everything you can think of (older drivers, DDU, AMD cleanup utility, fresh windows 10 and 11, restore default game settings, some tool called MPOGPUFIX, I think I even rolled back to older BIOS). And probably many other things I can't be bothered to remember but if you have any non-time-consuming ideas, go right ahead.
I might send these numbers to AMD but then again all other games work fine, 1.52 works fine. AMD didn't change anything. It's the game that got "updated". But hey, maybe AMD will know what's happening and contact you. One can hope.
I did the first playthrough on 1.52, heavily modded. But all the readings here are done without any mods installed.
My build: Ryzen 5600, Sapphire Nitro+ Vega 56, 16GB 3733 MHz CL16 DDR4 with tight sub-timings. Up-to-date Windows 10, 2560x1080, game installed on NVMe SSD.
EDIT: Corrected CPU load and GPU power draw numbers to match the numbers from attached image (must have been numbers from older measurements).
Last edited: