| SvennoJ said: Any benchmarks for handling transparancies? That always seems to be the biggest bottleneck. Also odd that the Titan X only doubles the ps4 in ALU numbers and instruction limit, whatever that means. So what makes the Titan X 4 to 12 times faster in this test? And does that mean that a Titan X can run games at 4K60, when ps4 runs them at 1080p30? 12 times should be more than enough for that? |
I'd be interested in how well consoles can handle transparency too but I suspect that the PS4 would most likely come on top since it has 5.86 GB/s of bandwidth per frame @ 30FPS compared to 2.23 GB/s of bandwidth @ 30FPS on the Xbox One to play with so it should definitely come out on top when doing multi-layer/pass rendering like transparency ...
Also the presentation is talking about the Fury X and the numbers in question are about the ratio ALU operations to rasterized triangles ...
What specifically makes the Fury X much faster in this test has to do with the fact that it's using GPU compute culling to accelerate the graphics pipeline ... (Fury X is known for it's infamous rasterization or other geometry related bottlenecks among developers so that is why it profits most in this research when it has a skewed ALU op to fixed function capability ratio.)
