Subwarp Interleaving Could Help Future NVIDIA GPUs Be 20% Better at Ray Tracing

Peter_Brosdahl

Moderator
Staff member
Joined
May 28, 2019
Messages
8,893
Points
113
Image: NVIDIA



A research paper from NVIDIA illustrates how it could use new technology to improve ray tracing performance by up to 20% in future graphics cards. The authors begin by explaining that ray tracing “applications have naturally high thread divergence, low warp occupancy and are limited by memory latency,” and an “architectural enhancement called Subwarp Interleaving” could be one solution for performance gains.



Subwarp Interleaving allows for fine-grained interleaved execution of diverged paths within a warp with the goal of increasing hardware utilization and reducing warp latency. However, notwithstanding the promise shown by early microbenchmark studies and an average performance upside of 6.3% (up to 20%) on a simulator across a suite of raytracing application...
Click to expand...

Continue reading...


 
From the article it doesn't sound like it will ever impact current GPU's, but rather future GPU designs.
 
and are limited by memory latency
Click to expand...

Hmm surprised we haven't seen an HBM version then. It's not like in today's market that cost is even a consideration.
 
I had a vague idea, but decided to look it up

Introduction to GPUs: CUDA

nyu-cds.github.io nyu-cds.github.io

The CUDA architecture is built around a scalable array of multithreaded Streaming Multiprocessors (SMs) as shown below. Each SM has a set of execution units, a set of registers and a chunk of shared memory.

Streaming Multiprocessors


In an NVIDIA GPU, the basic unit of execution is the warp. A warp is a collection of threads, 32 in current implementations, that are executed simultaneously by an SM. Multiple warps can be executed on an SM at once.
 
You must log in or register to reply here.
Become a Patron!
Back
Top