Reduceron, an efficient processor for functional programs

WHAT IS REDUCERON?

Reduceron is a high performance FPGA softcore for running lazy functional programs, complete with hardware garbage collection. Reduceron has been implemented on various FPGAs with clock frequency ranging from 60 to 150 MHz depending on the FPGA. A high degree of parallelism allows Reduceron to implement graph evaluation very efficiently.

Reduceron is the work of Matthew Naylor, Colin Runciman and Jason Reich, who have kindly made their work available for others to use. Please see https://mn416.github.io/reduceron-project/ (the original http://www.cs.york.ac.uk/fp/reduceron no longer works) for supporting articles, memos, and original distribution.

OK, WHAT'S THIS THEN?

The present is a fork of the original distribution with the aim of keeping it running. Originally there were more lofty goals, but that's all abandonned.

While Reduceron technically refers to the FPGA implementation, it is supported by

• Flite: the F-lite to Red translator.
• A Red emulator in C
• Red Lava: Reduceron is a Red Lava program, which generate Verilog
• Support for Verilog simulation and synthesis for various FPGA boards

As much of the history as was available has been gathered and Reduceron, Lava, and the Flite distribution have been merged into one repository.

HOW DO I USE IT?

The was last tested with Glasgow Haskell Compiler, Version 8.4.4 on macOS 10.14.3 and Linux, 64-bit.

Optionally: just run make in the toplevel directory and a large regression run will start. The Verilog simulation part will take weeks to finish.

To build:

make

Or run a specific test suite:

make -C programs $X

where $X is one of regress-emu, regress-flite-sim, regress-flite-comp, or regress-red-verilog-sim.

Note: the code generated by the C backend for Flite (used in the regress-flite-comp) depends on GCC features, such as nested functions. To build on macOS, install real gcc (say via Mac Homebrew) and invoke make as make CC=gcc-7 (assuming you installed version 7 of gcc).

To build a hardware version of a given test

cd fpga; make && flite -r ../programs/$P | ./Red -v

where $P is one of the programs (.hs). Next, build a Reduceron system for an FPGA board, fx the BeMicroCV A9:

make -C Reduceron/BeMicroCV-A9

Unfortunately programs can't currently be loaded dynamically but are baked into the FPGA image. It's a high priority goal to change that.

OPEN QUESTIONS, with answers from Matthew:

Q1: Currently there doesn't seem an efficient way to handle toplevel variable bindings (CAFs). What did the York team have in mind there or does it require an extension? (Obviously one can treat them all other functional arguments, but that would mean a lot of parameters to pass around).

A1: "Some mechanism would be needed to construct graphs at a specified location on the heap at the beginning of program execution. The initial (unevaluated) graphs have constant size so can be linked to at compile time."

Q2: Why does Flite default to 0 for the MAXREGS parameter? Eg, why is

  redDefaults = CompileToRed 6 4 2 1 0

A2: (Historical reasons it would appear).

Q3: What happend to Memo 24?

A3: "I'd like to say it was our best kept secret, but in reality it probably got trashed :)"