GPU technology is backed by the years of research demanded by the gaming industry, and has seen market penetration in nearly all desktop computers and many consoles today.
GPUs are here to stay; given their enormous existing demand, contined product development is guaranteed, and development of General Purpose libraries is expanding unabated. See nVidia's CUDA Zone for a showcase of applications.
So many complex computing challenges can by expressed in matrix form - and that's where GPUs shine. From solving simultaneous linear systems to running Monte Carlo simulations, GPUs have demonstrated 10-100x performance based on the application space and solution algorithm when compared to their multi-core CPU equivalents.
By leveraging the work of our toolkit partners, algorithm users and application developers don't need to become experts in parallel thread optimization; rely on the expertise of domain experts, and leave infrastructure hassles in our court.
yCumulus' architecture is completed by our high-level Python bindings layer, so application developers can use a powerful yet comfortable scripting language to orchestrate their apps.
But there's no need for performance concerns! Under the hood, all our bindings talk directly to the low-level C code crafted by our toolkit providers.
Leverage the technology you've already built by making it available on the yCumulus platform.
Broaden your client-base to include those without dedicated hardware. Your financial, pharmaceutical and similar industrial clients won't always be excited as you at the prospect of installing a giant GPU cluster! Offer them the same great results, without unnecessary hardware burdens.