Re-opening the idea now that LabVIEW NXG has been discontinued.
The LabVIEW compiler currently appears to use one core of a multi-core processor. It would be nice if it fully utilized multiple cores to speed building of large projects.
Are you saying that .lvproj, Build Specifications, application, Build has several serial processes (in this and next gen) which cannot use multiple cores but one of those processes, the compiling [of hundreds] of VIs that used to take a very long time uses all processor cores in next gen to compile different VIs in parallel?
This is one of several NXG ideas being studied. Unfortunately, this is one of the ideas that lead to the creation of NXG in the first place. Retrofitting the old code is extremely hard because of the number of functions involved written 30+ years ago that just assumed a single thread of operation. There's a surprising amount of statefulness and thread-awareness even in a codebase that is mostly G code and whose C code early on adopted the "no global variables" policy for code -- the C code is still rife with shared references and static local variables.
A complete rewrite of AppBuilder code might be required (which would be a good thing in and of itself) or it might turn out to be just a sizeable refactor. Either way, it would be a major commitment of resources to achieve.
All I can tell you at this time is that it is being evaluated.
We have a project which takes 2.5 hours to build (14K+ VIs) on a dual CPU (Xeon) computer with 64 logical cores. My laptop builds faster because runs on higher frequency.
The attempt to move the build machine to a VMware VM on a similar dual Xeon computer didn't go well either: 4+ hours to build the project.
Debe ser un usuario registrado para añadir un comentario aquí. Si ya está registrado, inicie sesión. Si todavía no está registrado, hágalo e inicie sesión.
Re-opening the idea now that LabVIEW NXG has been discontinued.