Using the massive speedup offered by using GPU processors, WideLM performs linear regression on multiple similarly-shaped models, in parallel. Advanced GPUs speed WideLM calculations by as much as 150x, although the degree of performance improvement depends on the GPU hardware available. However, because of WideLM's blocking algorithm, there is no predetermined limit on the number of models it can handle.

WideLM is envisioned as a fast, streamlined version of R's venerable linear modelling command, "lm", tailored specifically for the case of numerous modestly-sized submodels taken from a large under-determined (p >> n) system.

WideLM is available for free download on CRAN.