You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am looking to see how to building template reference lapack and blas for packages within, and called from, GALAHAD. The intention, just as elsewhere in GALAHAD, is to template both real and integer types via CPP comands.
In doing so, I find that there are a number of lapack procedures (and their blas dependencies) that do not appear to be needed by either GALAHAD or the optional subset of HSL packages that GALAHAD uses. In particular, neither GALAHAD or the HSL subset requires complex LAPACK procedures. These seem to have been part of GALAHAD since at least V 2.1, and possibly before, and it looks like they were added in error (presumably by me!).
I thus intend to remove them unless there is any strong objection. The main advantage is that by sticking to real and integer procedures, there are no ambiguities between real and double precision procedures (the complex ones require variants that do not have real equivalents).
Any objections?
The text was updated successfully, but these errors were encountered:
I am looking to see how to building template reference lapack and blas for packages within, and called from, GALAHAD. The intention, just as elsewhere in GALAHAD, is to template both real and integer types via CPP comands.
In doing so, I find that there are a number of lapack procedures (and their blas dependencies) that do not appear to be needed by either GALAHAD or the optional subset of HSL packages that GALAHAD uses. In particular, neither GALAHAD or the HSL subset requires complex LAPACK procedures. These seem to have been part of GALAHAD since at least V 2.1, and possibly before, and it looks like they were added in error (presumably by me!).
I thus intend to remove them unless there is any strong objection. The main advantage is that by sticking to real and integer procedures, there are no ambiguities between real and double precision procedures (the complex ones require variants that do not have real equivalents).
Any objections?
The text was updated successfully, but these errors were encountered: