The world's first fully functional 32-bit bendy CPU can run while wrapped around a straw, but boy is it slow
While bendy, flexible screens have been demonstrated for a while now (with a fair few drawbacks, I might add), most of the components inside our devices have so far remained stiff as a board. This RISC-V processor, however, bucks that trend by being as bendy as a professional gymnast at a yoga class, even while it's running.
Pragmatic Semiconductor has published a blog announcing the hyper-flexible chip, which it says is the «world's first 32-bit microprocessor in a flexible technology that is fully functional while flexed» (via Tom's Hardware). A bit of a mouthful that—may I suggest something more catchy, like «The world's first 32-bit bendy boi»? Anyways, moving on.
The Flex-RV isn't your traditional silicon-based chip. According to the accompanying paper, it's been fabricated with indium gallium zinc oxide thin-film transistors on a flexible polyimide substrate in order to make such a flexible computational device, and I wouldn't blame you if your eyes crossed while reading that sentence. Luckily, however, Pragmatic Semiconductor has provided a demonstration video that makes the benefits of this tech abundantly clear.
The chip is shown being rolled up and then unfurled on what looks like a clear plastic straw, while running without issue. It's a bit like watching a CPU on a medieval rack, and while the flexibility is obviously impressive, our demonstrator does seem to be treating it with an overabundance of care for the most part. I'm willing to bet the prototype is still very delicate, but it's an impressive result nonetheless.
This is a RISC-V processor, which means it's based on an open standard instruction set architecture—and if you're expecting great performance, you're about to be disappointed. The test sample has 12,600 logic gates and a maximum clock speed of a stonking, err, 60 kHz.
For reference, that's 0.06 MHz, or 0.0006 GHz, although it does have an onboard ML accelerator—meaning that it's technically machine-learning capable. It's not going to be