This work presents a theoretical proposal for a base-10 (decimal) processing-in-memory (PIM) architecture built on a new class of magnetic materials called altermagnets. The proposal addresses two fundamental limitations of modern computing simultaneously: the von Neumann bottleneck — the energy and latency cost of moving data between separate memory and processing units — and the inability of binary hardware to represent decimal fractions exactly, which causes the well-known floating-point rounding problem (0.1 + 0.2 ≠ 0.3 in standard computing). The primary architecture proposes encoding decimal digits as the physical position of an antiferromagnetic domain wall at one of ten discrete sites along a nanoscale track in manganese telluride (MnTe) — a material recently confirmed as an altermagnet, a newly characterised class of magnetic material that enables electrical readout without an external magnetic field. Thermal stability above CPU operating temperatures is addressed through a MnTe/CrSb (chromium antimonide) superlattice, where CrSb acts as a magnetic anchor due to its much higher magnetic ordering temperature (~700 K). Arithmetic operations — addition, subtraction, comparison — are performed in-place by moving the domain wall, eliminating the need to transfer data to a separate processor. The document covers: the physical and materials science background; the reasoning for the primary material selection; a full device architecture with theoretical performance analysis; a step-by-step theoretical manufacturing pathway; six alternative material pathways including Mn₃Sn kagome antiferromagnets and pure CrSb variants; physical integration targets from co-processor to CPU+RAM replacement; application domains; and ten prioritised experimental hurdles with specific characterisation methods required to advance this concept toward a working prototype. This is an original theoretical proposal by an independent researcher, incorporating literature current to early 2026. It is intended to serve as both a foundation for discussion and an experimental roadmap, and the author invites contact from researchers with relevant experimental capabilities.
Kasey Azure (Sat,) studied this question.