Introduction
The Swallow system is a many-core processing system, based on the XMOS XS1-L architecture and build in 'slices' of 16 processors each.
Swallow processors are interconnected using the XMOS-based network links, which offer very low latency processor-to-processor communication. These links are either on-board in the form of PCB traces, or board-to-board in the form of flexible ribbon cables.
The aim is to build a system capable of offering hundreds of cores, all interconnected into a massively many-core system.
Tutorial material
In terms of the architecture, here's the XCORE the overview available from XMOS:
and the programming guide
The architecture definition is available here
Current status
Currently, we have sufficient slices to build a 680 processor system, but the largest we have interconnected so far is 288 cores. You can see a video of this system being booted on youtube .
Recent updates
Recently Updated
Navigate space
Attachments:
Swallow_V3-Annotated-small.png (image/png)
Swallow_V3-Annotated-smallish.png (image/png)
Swallow_V3-Annotated-small.png (image/png)