Joint offspring of Rothamsted and the New South Wales Department of Primary Industries, ASReml is a Residual Maximum Likelihood package for fitting and analysis of linear mixed models. Focussing on fast and efficient treatment of big and complex data sets, it has always thus far been a forbidding command line interface (CLI) batch environment, in which speed and power was everything. Shifting to it from the padded comfort of a graphical user interface was like moving from a limousine to a bobsleigh - temptingly luxurious ease of use exchanged for the undoubted benefits of stripped down streamlined bare bones fitness for function. The starkness of that choice will be softened for Microsoft Windows users of release 2 by the arrival of WinASReml, developed by distributor VSN International.

WinASReml is a welcome new, graphically-interfaced environment within which ASReml2 projects can be developed in greater comfort (including graphics review) before release into the batch processing wild. It is not present in the install available as I write this (slightly confusing, since references are made to it in the help system), but should be present and correct very soon. The pre-release development version I’ve seen is very encouraging - even as a tyro, I was able to work up sample code quickly and confidently, with a high level of first time success.

Until the final release version of that environment emerges, ConTEXT (, historically the most popular ASReml program editor, remains a good choice. Quite apart from the new front end, ASReml2 offers plenty of reason for existing users to consider the advantages of upgrading and prospective adopters to think about taking the plunge for the first time.

A range of new transformations has been added: two for genetic marker handling (assignment of Haldane mappings, including missing values, and formation of dominance covariables), while the rest offer various options for normally or uniformly distributed randomisation replacements. ANOVA and autoregression have been enriched, inversion of sparse matrices for G-structures can be done inline, and numerous other enhancements added to both power and handling. Data reading and assembly, linear model specification and runs programming are more comprehensive and more flexible. The improvements go on, at some length and in detail

The command line has gained additional sophistication, with nine new options (one administrative, eight operational) and they can be conveniently specified on a ‘top job control line’. This is a line within the program that specifies as qualifiers the options that would otherwise be given as program arguments. A file template can be generated from the command line, too, taking variable, case and factor definitions from an Excel or Genstat file in the process of conversion to CSV text. The template isn’t perfect, but much of the heavy lifting is certainly taken from the programmer’s shoulders.

A number of changes in behaviour are documented, and need to be looked at for possible code compatibility implications. In some cases, ignoring them will negate upgrade benefits; others may be less tolerant of sloppiness on the user’s part, resulting in interruption. Having said that, the number of such revisions is small and, in every case, a minor price paid for a much larger benefit. And, thanks to that new WinASReml front end, it will all be easier to use.


Robert Roe reports on developments in AI that are helping to shape the future of high performance computing technology at the International Supercomputing Conference


James Reinders is a parallel programming and HPC expert with more than 27 years’ experience working for Intel until his retirement in 2017. In this article Reinders gives his take on the use of roofline estimation as a tool for code optimisation in HPC


Sophia Ktori concludes her two-part series exploring the use of laboratory informatics software in regulated industries.


As storage technology adapts to changing HPC workloads, Robert Roe looks at the technologies that could help to enhance performance and accessibility of
storage in HPC


By using simulation software, road bike manufacturers can deliver higher performance products in less time and at a lower cost than previously achievable, as Keely Portway discovers