TECHNOLOGY PROFILE · TECH-0044
CuPy
NumPy-like GPU array library.
Reference link only — not reviewed this edition. This profile preserves the prepared catalog’s original descriptions. Advantages depend on the workload; comparisons are editorial, not benchmark results.
Where it has an edge
Ports compatible array-heavy calculations to GPUs with less rewriting than hand-written CUDA.
Best-fit problems
GPU-resident numerical preprocessing and custom array algorithms.
Tradeoffs & cautions
Transfers can erase gains; CUDA/ROCm and operation compatibility must be checked.
What to test before choosing
Measure wall time, peak memory and IO on a representative dataset; verify numerical and schema equivalence.
Runtime & hardware
Local CPU; GPU where stated
Access & licensing
Open-source core; verify component licenses
Execution
Local / self-hosted
Lifecycle
Not independently audited
Comparison basis
Editorial problem-fit judgement; no head-to-head benchmark performed
Connected concepts
Alternatives to investigate
NumPy · JAX · PyTorch