The Story of the Amulet: A Brief History of Asynchronous Events in Manchester

This presents an overview of the significant achievements in Asynchronous Logic over the last quarter of a century at the University of Manchester. Including the Amulet (and subsequent) asynchronous Microprocessors, the synthesis tools Balsa and Teak, the MARBLE and Chain on-chip Interconnects, together with research arising from external collaborations with the University of Newcastle. 1 Pre-history an Overview by Steve Furber The Amulet story begins in 1989. Through the 1980s I (SBF) worked at Acorn Computers Ltd in Cambridge, UK, where I was a designer of the BBC Microcomputer and the ARM (then “Acorn RISC Machine”) 32-bit RISC microprocessor. We didn’t publish much, but one paper we published was on the ARM3 the first ARM with an on-chip cache at VLSI’89 in Munich [7]. At that conference I was struck by a paper [14] presented by Craig Mudge, then CEO of Austek Microsystems Ltd based in Adelaide. This was my first exposure to asynchronous design, and of course Mudge’s paper cited Ivan Sutherland’s Turing Award paper [19] which really sold the concept to me. Over the following year I doodled some ideas on the possible design of an asynchronous version of the ARM based on Sutherland’s micropipelines approach. This was around the time the EU launched the OMI Open Microprocessor systems Initiative which was in part stimulated by input from Acorn. When I accepted the ICL Chair at Manchester, to which I moved on 1st August 1990, Acorn was generous in allowing me to take with me to Manchester a part of an early OMI project that I had been involved in developing at Acorn the OMI-MAP project. OMI-MAP provided the early funding that got the Amulet (asynchronous ARM) research off the ground at Manchester. The University allocated a lectureship appointment linked to my chair, to which Jim Garside was appointed, and OMI-MAP funded the appointment of Nigel Paver and Paul Day as post-docs. In addition, existing academic staff joined in Doug Edwards, Linda Brackenbury and Viv Woods, bringing in additional perspectives and new funding, all of which enabled the Amulet group to grow to critical mass. Why “Amulet”? Well, in searching for a group or project name you have to start somewhere, and writing down keywords on a whiteboard is one way to start. “Asynchronous”, “Manchester University”,“Low-Energy Technology”? 155 x 238 mm 120 Doug Edwards et al.