Reassessing Processor Design Assumptions in Cryptography
Lead Research Organisation:
University of Bristol
Department Name: Computer Science
Abstract
The design of computer processors is based on many years of research andunderpins the performance of most other areas of Computer Science. One canroughly split processor designs into two groups: general purpose, which arecapable of being programmed for any task, and special purpose, which arelimited to one task only. The trade-off between the two approaches is thatspecial purpose processors generally deliver higher performance or smallersize and power budgets because they can focus on one particular task and notworry about needing to execute the others.Special purpose processors are attractive in cryptography. Cryptographicalgorithms are representative of a domain which uses specialist data typesand operations and demands high performance. However, cryptography is a fastmoving field and if a cryptographic algorithm is found to be weak, it needsreplacing quickly to avoid breaches in security. This need for algorithmagility negates the benefit of special purpose processors. Therefore, if ageneral purpose processor were able to deliver the required performance itwould be a better choice since it could simply be reprogrammed to run a newalgorithm rather than simply discarded.Like most aspects of computer architecture, successful general purposeprocessor designs rely heavily on selecting an effective trade-off betweenmany competing factors. Such decisions are typically resolved by performinga workload characterisation of the sorts of algorithm that the processor willexecute. By considering an average program, the architecture is designed sothat the average case is optimised while non-typical cases are marginalisedor omitted. A quarter century after many design decisions and assumptionswere made by the pioneers of the field, we are still using largely similardesigns. One expects that such decisions were initially made using a mix ofresearch and common sense based on prevailing technologies of the time.Despite the success of these assumptions, the technology landscape has nowchanged radically: the types of algorithm we execute today are differentand many of the constraints which guided initial thinking have disappeared.We claim that current general purpose processor design is based on entrenchedassumptions that are limiting their flexibility and long term viability forexecuting a wide range of algorithms. The crux of our proposed research isthe reassessment of such assumptions and investigation of alternativeapproaches using cryptography algorithms as an example domain. We aim, throughrelatively minor and non-disruptive alterations in design which lie outsidecurrent thinking, to deliver a high performance platform for cryptographicalgorithms while allowing algorithm agility: essentially delivering the bestof both world from general and special purpose processor design.
Organisations
Publications
Day G
(2022)
The supply-side climate policy of decreasing fossil fuel tax profiles: can subsidized reserves induce a green paradox?
in Climatic change
Grabher P
(2008)
Cryptographic Hardware and Embedded Systems - CHES 2008
Ju T
(2023)
A new prediction method of industrial atmospheric pollutant emission intensity based on pollutant emission standard quantification.
in Frontiers of environmental science & engineering
Koschuch M
(2008)
Cryptology and Network Security
Regazzoni F
(2009)
Transactions on Computational Science IV
Slucock T
(2022)
A Systematic Review of Low-Cost Actuator Implementations for Lower-Limb Exoskeletons: a Technical and Financial Perspective.
in Journal of intelligent & robotic systems