MULTITHREADED MONSTER

It’s easy to build a multicore PC, but what can it do?

It’s now easier than ever to build a multicore PC, but what can you do with it?

Multicore CPUs are so commonplace, you’d be hard-pressed to build a PC today that isn’t packed with cores. You certainly wouldn’t want to, as the performance benefits of additional processing cores are noticeable, and keep today’s PCs on the cutting edge. The Pentium Dual Core brand dates back to 2006’s Yonah core, while AMD was producing Athlon 64 X2s with Hammer cores in 2005. That’s not to mention the work outside the X86 space, with ARM and the graphics card manufacturers also using multiple cores.

Recently, however, there’s been a core explosion. Quad-core held the top spot, at least outside of servers, for a long time, as Intel worked through several iterations of Core chips, but the resurgence of AMD with its Zen and Zen+ architecture means you can now buy a six-core Ryzen 5 2600 for $200. Simultaneous multithreading (the ability for each core to work on two software threads at the same time, known over at Intel as HyperThreading, and used since the Pentium 4) means this chip can process 12 threads simultaneously. Intel will sell you an 18-core, 36-thread chip, while AMD’s line goes all the way to 32 cores and 64 threads. There’s never been a better time to chew through large amounts of data on your desktop.

These are processors aimed at home PCs, not destined for supercomputers (which have moved over to GPUs with tens of thousands of cores for their specialized workloads), so the datasets they deal with are more mundane. GPU-based supercomputers are starting to run the world, from universities carrying out scientific research to Amazon Web Services and the tech behind the Google Assistant and Siri, offloading tasks such as AI and speech recognition to what’s euphemistically called “the cloud.” That doesn’t mean you can’t have a little fun at home, however. From the future of gaming to creative endeavors to making new scientific discoveries, there are many more uses for your new multicore monster than just manipulating gigantic databases.