Reading this thread I don't instantly jump to the conclusion that BF4 / Frostbite is broken in regards with mantle, here is my reasoning.
I think that the mantle API is probably slightly more VRAM hungry than DX11, and depending on the implementation of this API in different engines some will be see a greater jump in VRAM when compared to DX (in this case frostbite could just be very unlucky and need very VRAM hungry operations when comparing to DX)
Also depending on how "back to the metal" mantle is and how much control it gives over the VRAM memory management small differences in the game engines implementation if the mantle API could see major inconsistencies between the DX to Mantel VRAM usage differences per game.
Im going to put it down to the following :-
Ill just go play some BF4 with the fps boost of mantle
- Other mantle games increase in memory usage when comparing DX to Mantle
- Normally memory management relates issues at this level would result in an ever expanding use of memory (memory leak). This is not what we are seeing
I think that the mantle API is probably slightly more VRAM hungry than DX11, and depending on the implementation of this API in different engines some will be see a greater jump in VRAM when compared to DX (in this case frostbite could just be very unlucky and need very VRAM hungry operations when comparing to DX)
Also depending on how "back to the metal" mantle is and how much control it gives over the VRAM memory management small differences in the game engines implementation if the mantle API could see major inconsistencies between the DX to Mantel VRAM usage differences per game.
Im going to put it down to the following :-
- Mantle is still in closed beta so we wont know if the cause is within mantle itself, and if it is ..... its in beta so expect some bugs
- Frostbite (BF4) was the very first game to implement mantle, its implementation could well be "rough" in comparison to others
Ill just go play some BF4 with the fps boost of mantle
