Toward a Universal GPU Instruction Set Architecture: A Cross-Vendor Analysis of Hardware-Invariant Computational Primitives in Parallel Processors

We present the first systematic cross-vendor analysis of GPU instruction set architectures spanning all four major GPU vendors: NVIDIA (PTX ISA v1.0 through v9.2), AMD (RDNA 1-4 and CDNA 1-4), Intel (Gen11 through Xe-HPC), and Apple (G13, reverse-engineered). Drawing on over 5,000 pages of ISA reference manuals, architecture whitepapers, patent filings, and community reverse-engineering efforts across 16 distinct microarchitectures, we identify ten hardware-invariant computational primitives, six parameterizable dialects, and six true architectural divergences. We propose an abstract execution model for a vendor-neutral GPU ISA following the thin abstraction principle, validated with benchmark results on NVIDIA T4 and Apple M1 demonstrating that universal primitives match or exceed native performance on five of six configurations.