@Whyborn I removed some that were directly interfering. The ones I removed were made by Vanessa Haverd. I hope a test shows this is bitwise comparable and I can confidently avoid having to review every instance over 26 files :)

It will definitely be bitwise compatible in CABLE offline, as we don't set qopenmp in the compilation, and I don't think we do for ESM1.6 either? And given the code isn't integrated directly with AM3, it won't make any difference there either. I'll do a benchcab run just to check, but I can't see how it would possibly fail.

I asked for your review mainly because you may know some of the history as to why this was done in the past. There are some bits that look like they're true comments, some bits that looked like they have been used as debugging by misusing -qopenmp, and some that might have been intended to be included in coupled applications.

@Whyborn - You're right - there is no legitimate reason for it. In between CMIPs (so it had no meaning offline as you say) Vanessa added these comments which appeared simply to be her style. There was one instance where some of the !$ lines were a problem (I think it was in canopy), but then this seemed to go away again in another. Anyway, we're better off without them - oh I remember - I think - it was in trying to reproduce ESM1.5 behavior with ESM1.6 (all the since removed cable_runtime%esm15 switches and #IF def ESM15) but I don't think these lines were causing a difference in the output, I think it wasn't even getting that far, maybe the build even. It interpreted it as an OMP directive but then there was nothing valid following it.

The benchcab run shows bitwise equivalence. The lines were causing issues when I tried a naive copy of the CABLE science files into JULES, as -qopenmp is specified for the AM3 build.