"next generation RDNA architecture" with hardware accelerated ray tracing

[chris1515]

I began writing, then realized I'd seen an MS blog post explaining this exact thing in far more detail than a forum post should even be: https://devblogs.microsoft.com/dire...on-shaders-reinventing-the-geometry-pipeline/

No digging through videos and tweets if anyone wants, less sync and more polys are always cool.

http://www.reedbeta.com/blog/mesh-shader-possibilities/

Excellent blog post about Mesh shader

 

[bbot]

What no website has answered is what kind of ray tracing performance can we expect from RDNA2 gpus. In other words, how many giga rays per seconds. They need to start digging.

 

[BRiT]

Impossible to dig without hardware.

 

[trinibwoy]

Basically, massively parallel processing with compute shader-like scheduling that is not limited by issue rates of fixed-function geometry hardware.
https://forum.beyond3d.com/threads/direct3d-mesh-shaders.61286/

My layman's understanding is that mesh shaders simplify the current hodge podge of fixed function geometry stages with the added benefit of compute shader like flexibility in that arbitrary data can be passed into the shader.

It doesn't seem to be an obvious performance win though. The rasterizer is still going to be a bottleneck. Also mesh and task shaders are likely slower at tessellation than dedicated hardware.

 

[BRiT]

Until there are actual specifications released from AMD, please continue discussion in the existing speculation thread here: https://forum.beyond3d.com/threads/amd-navi-speculation-rumours-and-discussion-2019.61042/