ARCA

Abstract Research of Computer Architecture

hello! We are ARCA

We are the Abstract Research of Computer Architecture group. We generally explore novel computational systems with low-level applications and more. Our work spans fundamental microarchitecture, memory hierarchies, interconnects, and the co-design of hardware with compilers and runtimes to unlock efficiency and reliability at scale.

Research icon

Applied research bridging theory and real systems—validated on simulators, FPGA prototypes, and silicon when feasible.

CPU icon

Focus on low-level applications: graph analytics, ML inference, secure boot, and telemetry-driven optimization.

Publication icon

Open science ethos: publishing specs, datasets, and code where possible to accelerate community progress.

Areas of Inquiry

Arrow icon
  • Microarchitecturepipeline, out-of-order
  • Memory & CoherenceCache coherency protocols
  • Hardware/Software Co-Designhardware compilation

Current Research Projects

Chip layout visualization

Neon - a hardware definition language

A hardware definition language with a minimalist, functional style syntax in order to lower the barrier of high performance hardware design

HDLCompilingFPGA
Alpha
Memory modules

Self reprogramming hardware

combining the best of both from FPGAs and classic fetch-execute style computation to optimise performance across dynamic workloads

CPUCompilingFPGA
Active
RISC-V board

Janitor cache - a hybrid cache coherency protocol

looking at using free bus resources to update cache lines in order to get the most from update and invalidate based coherency protocols.

CPUCache
Active

FPGA-Based Analysis of Memory-Hard Password Hashing

argon2i hash security tested with fpga attacks. measuring practical adversary cost scaling for password protection.

FPGA
Active
Compiler code editor

Decode-staged based instruction reordering

researching novel instruction reordering concepts for use in lightweight, out of order, superscalar architectures

CPUISACache
Active
Secure boot schematic

using hardware queues for optimised SIMD processing

adapting features from functional programming and other mediums to design an efficient SIMD processor

CPUISA
Active
looking for something? contact us to enquire about our closed source projects
Collaboration type
Thanks! Your message was sent. We’ll get back to you at the email provided.