In forum topic, "Community Components" #136, regarding porting Community Components to WL7, Eugene stated,

QUOTE:
"We have not decided yet but are taking a different approach with regard to most frequently used methods. Solutions in demand will become part of Wealth-Lab one way or another:...

In forum topic, "The Future of WL7", there are repeated requests for parallelism and CUDA/GPU support. I would suggest that testing on advanced hardware can dramatically increase testing time, if the hardware is even available to the development team. Some failures may not be reproduceable based on what else was running. This persists when increased support time is needed as new hardware is introduced.

I would suggest strongly, at this point in the WL7 life cycle, that speed is secondary. If your strategy won't port for lack of functionality currently provided by Community Components, Community Indicators, Community Visualizers and others, you'll never know how fast it might have run. Distracting the very busy team with these requests for parallelism is not particularly useful.

Just my opinion - Concentrate on building functionality. Fully exploiting advanced hardware can wait until after WL7 is launched.

I totally agree. I don't think CUDA/GPU support is important at this time. But employing a design pattern to both vectorize and parallelize calculations in the future is a good plan. For now, I would just get things running employing appropriate code design, which can be sped up later. And well written code (e.g. employing tight loops) will speed up the FPU in the general processor as well since it employs pipelining architecture.

Most / many users won't have access to advanced machines to test with - most have simple laptops. The software has to be usable by the masses, including me:)

True. No worries, we got you covered.