There's three related things: Parallella, Adapteva, and Epiphany.
Parallella is the board project. Originally with 16-core and 64-core boards, I think only 16-core boards are still available.
Epiphany is the CPU architecture. It has a bunch of limitations, but it's also small, fast, and low-power, so if your application works within those limits, it might be a good option.
Originally specced to include 16, 64, and 1024-core chips, I think the 64-core and 1024-core variants are no longer available.
And finally, Adapteva was/is the company. It ran a successful crowd-funding campaign for the Parallella project, but wasn't able to convert that momentum into an ongoing CPU-manufacturing business.
Andreas Olofsson, the core of the whole thing, moved on to work at DARPA, I think, and I fear that's the end of the road. There's still some academic work done with the Parallella boards, and there's no reason you couldn't use one today, but I wouldn't expect there to be any new products in the family.
I don't know anything about the GPIOs, sorry.
For your algo, there's a bunch of cheap(-ish) FPGA boards out there. Unless you're planning to use the Epiphany chip as well as the Zync, you might be better off using something with a future roadmap?