Epiphany-V: A 1024-core 64-bit RISC processor project (Distributed Task Farm)
Compared to leading HPC processors, the chip demonstrates an 80x advantage in processor density and a 3.6x advantage in memory density.
Epiphany-V Summary:
- 1024 64-bit RISC processors
- 64-bit memory architecture
- 64-bit and 32-bit IEEE floating point support
- 64 MB of distributed on-chip SRAM
- 1024 programmable I/O signals
- Three 136-bit wide 2D mesh NOCs
- 2052 separate power domains
- Support for up to One Billion shared memory processors
- Support for up to One Petabyte of shared memory
- Binary compatibility with Epiphany III/IV chips
- Custom ISA extensions for deep learning, communication, and cryptography
- TSMC 16FF process
- 4.56 Billion transistors, 117mm^2 silicon area
- DARPA funded
https://www.parallella.org/2016/10/05/epiphany-v-a-1024-core-64-bit-risc-processor/
https://www.parallella.org/2016/11/14/task-farms-on-the-epiphany/
So these chips can scale to 1 Billion cores in a single system.. we should seriously consider an Epiphany-V BOINC project. It would require its own project because it would likely be incompatible with existing projects.
There has been some skepticism as to whether or not these processors would be useful for any real-world gain, or if they were a "test-bench" so to speak, for somebody wanting to play with the framework of a supercomputer... to write code for one or whatever. I hope to open discussion further with some recent findings:
http://www.pcworld.com/article/3127176/hardware/adaptevas-1024-core-epiphany-v-mega-chip-packs-serious-wallop.html
According to this, the chips will be good for real-time image processing. Perhaps Asteroids@home could actually make some use of this... or possibly the beginning of a new project comparing images for transitory planets? Could even be useful for the burst communication project I was talking about... I have been scouring for some code related to finding JUST voice on an audio track... but it just doesn't really exist yet. It would apparently be quite lucrative as well, seeing as phone companies could use it to lower bandwidth usage.
Even some kind of distributed test project to see what kind of user-coded projects they are able to handle? Could be dangerous, but we could have people run their codes at home, and submit success stories to a forum or something.