Since its inception in 2007, the particle engine that later became the middleware known as PopcornFX has been using a highly parallel stream processing architecture, which is at the core of its high performance and flexibility.
It is extremely similar to the way modern GPUs work. We’ve been improving the system since then, and keep it evolving as new hardware comes out.
Looking for distributing work across worker threads on multiple cores with minimum contention?
Efficient, hand-optimized SIMD code?
Leveraging the power of platform-specific hardware?
Careful cache management, write-combined-buffers and LHS aware code?
Yes, we do all of that. And more.
In addition, our build servers run 24/7 and check feature and performance regressions to bring you a top-quality runtime.