As of yesterday, Folding@home efforts are being powered up with CUDA support. Nvidia is stepping up to the front line to provide GPU acceleration, courtesy of its parallel computing platform, CUDA, in order to help evaluate and synthesise the sheer volume of molecules necessary in the battle against the ongoing global pandemic. So, projects like COVID Moonshot are going to get a serious boost in the fight against COVID-19. Where the fate of the world hangs in the balance, more processing power certainly doesn’t go amiss.
CUDA support means most projects using the open-source Folding@home distributed computing tech (based on the OpenMM toolkit) will be looking at a minimum of 15-30% performance increase from most standard GPUs, with some pushing a lot more. Nvidia Geforce GTX 660, 670 and 680 cards have even benchmarked upwards of 50-60% speed mark-ups.
COVID Moonshot Sprints is on another level. It uses OpenMM features to figure out how much of an impact potential drugs will have, and this can see speedups of 50-100% on many GPUs. On the top end of the spectrum, CUDA-enabled COVID Moonshot projects, ones that identify potential therapy, are benchmarking a whopping 50-400% performance increase with GPU acceleration.
COVID Moonshot is a crowdsourced, open science, drug discovery project that brings together an expanding network of volunteer scientists who run protein dynamics simulations on their personal PCs, in a race for the all important anti-viral. If projects like this are successful, we could be looking at low-cost, patent-free, two-a-day pill therapy for patients suffering from COVID-19, and in large part thanks to your humble gaming graphics card.
Over the next few days, CUDA support will be seamlessly integrated to many Folding@home projects, as the new release of version 0.0.13 of the core22 software rolls out.