New to Overclock.net
Join Date: Apr 2015
Location: South United States Of Europe
Instead hairworks is exactly the perfect example.
Tessellation is common to DX11, right?
Hairworks was nvidia implementation of a dx11 standard (exactly like RTX is own implementation of DXR software layer) that is optimized for their own architecture,of course..... it relays on dedicated hardware or, just like amd, generic use shaders dedicated to particular tasks.
The nvidia trick was to force the haiworks thing to work on a fixed tessellation value very high, too much high even for kepler architecture which had weak tessellation perf compared to the new maxwell arch (and people with 700€ 780tis buyed six months before that complained all over the nvidia forums), then they let free the developer to unlock the tessellation values as per user demand.........these "tricks" can be done even on RT affect both paths: nvidia and amd.....because they'll do what the engine ask them to do (ie. tessellation 64 instead of 32,16,8....).
Am I doing it wrong?
Well, at the end seems that even amd will have "rt dedicated hw" so it may be different case but for sure both arch will act differently on a hw level....dunno, we will see were we will end up......rt right now it is a joke, so why bother....
Asrock Taichi X470
sapphire fury nitro oc
GSkill Trident Z RGB 16GB 3200 Cl14
SSD Crucial MX500 500GB
Antec HPC 850 Platinum
noctua dual fan 120
Phanteks Enthoo Pro M Acrylic
microsoft windows 10 pro
AMD Radeon R9 200 Series
Crucial M4 256GB
Bitfenix Prodigy mitx
▲ hide details ▲