Harnessing Human Computation Cycles for the FPGA Placement Problem

Abstract

Harnessing human computation is an approach to find problem solutions. In this paper, we investigate harnessing this human computation for a Field Programmable Gate Array (FPGA) placement problem. We create a game called Plummings. In this game, a player attempts to reduce the critical path of a digital design mapped to an FPGA by swapping clusters on the array, but the problem details are abstracted away, and instead, the game simply presents a challenging problem where paths must be minimised to save the game characters - the Plummings. Once players have played a level, the placement is can be evaluated in VPR. Our results show that 4 human players over a set of 5 benchmarks can create placement solutions with comparable critical paths compared to VPRs solu- tions. This is not always the case, and we suggest some reasons and further approaches to improving our results.