By using this site, you agree to our Privacy Policy and our Terms of Use. Close
JRPGfan said:

watching video... intrest.
The slightly weird:

Hardware accelerated ray-tracing (amd version).

*** can run parallel to normal operation workloads, without effecting the other stuff.
This is a "extra" 13 Teraflops of Raytraceing workloads via parallel processing.

Might not have as hard a "hit" from useing Raytracing, as Nvidia has on PC side.

Uh.. sorry to rain into your parade, but this one has me worried the most. AMD/MS/DF say 380 GIntersections/s. How did they get to this number?

Solution: 52 CUs * 4 TMUs * 1.825 GHz = 380 GIntersections/s

This tells us that AMD's RDNA2 accelerates RT using the TMUs (or the TMUs simply limit RT). This is by far inferior to NVidias solution. and it is not an extra TFlops, it would actually mean taking away from the 12TFlops...