Oriel — AI & technology worth noticing, every day.

A Specialised Haskell Processor on FPGA

Web page · https://bathtub-01.github.io/technical/ouros-demo

First seen by Oriel on .

AI summary

Ouros is a specialised graph-reduction processor that runs Haskell directly on an FPGA, providing hardware-managed IO, automatic threading, and garbage collection without a software runtime or kernel.

  • Haskell programs compile into fixed-size combinator graphs whose Apps flow through a Dereference Heap, Reducer, and ALU instead of executing instruction sequences.
  • State-passing orders lazy IO operations, while dedicated hardware translates readAXIReg and writeAXIReg primitives into AXI4-Lite transactions; the current demo polls UART registers.
  • On an AUP-ZU3, Ouros runs at 100 MHz and uses approximately 27% of LUTs, 7% of registers, and 40% of BRAM.
  • Current limitations include no Float, Double, or Int division, no memory hierarchy, dynamic program loading, or hardware interrupts, and only polling-based IO.
  • The FPGA implementation is likely several times slower than a modern laptop, while fair comparisons are complicated by differing implementation technology, circuit size, power, and memory organization.

#haskell #fpga #graph reduction #processor design

Generated by gpt-5.6-sol. Check the original for details.

Metric trends

Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained

Points

Score reported by the source.

  • Lobsters: 19, latest observation ; +18 over observed period since

Comments

Comment count reported by the source.

  • Lobsters: 4, latest observation ; +4 over observed period since

Aliases

SourceExternal IDURL
Lobsters8o9wushttps://bathtub-01.github.io/technical/ouros-demo/ Read discussion