Present: Alastair, Richard, Abouzied, George, David, Tobias, Edo, Peter, Jonathan, Benedict. - Actions Action: Richard to speak to Katie just before next meeting. Action: email around working group meetings that are known about. Action: Richard will look at setting up a meeting in January for a cross-excalibur meeting. Action: Richard Set up coordination between working groups Action: Richard Swift training day in January Action: Alastair - website/minutes   - General updates SP... call which should be bid for. Tobias has submitted this. task based parallelism as a cross cutting thing. But doing it by interviewing PIs. Gen-X meeting a week ago. Tried to do interactive whiteboard. This worked surprisingly well. Virtual postit notes. Anonymous. Content takeaway: particular community, much not relevant to everybody. Can you use 1 big exascale simulation? Mileage varies with this. Different communities will look at this at different levels. Swift does a lot of parameter variation and fitting. Parallel-in-time could be a solution for strong scaling. Is exascale a training exercise for petascale, rather like knl was? Possibility of using excalibur money to support sph-sig, workshops. What do we pick as a particular topic? Parallel in time stuff, task based parallelism. Need something badged as part of the project, but wants to be something people want to do. Makes sense to leave some of the dissemination until later. Signal to some of the right groups that we will do something. Need to coordinate between the working groups. Compile what people are planning. Invite working group leaders, to give 5 minute presentation at a 1 hour session. About how to run cross cutting themes, and what workshops are you putting on. Or look at a website to host this info? Richard to do this for next time. Do we want to look at an astrophysics related theme? Swift training day. January. Richard to look at doing this. Alastair to look at a website for hosting minutes. - Contracts (Richard) 2 responses from commercial partners, feeding in comments into the document. Stuff is progressing.   - Hydro weak-scaling (Edo) Weak scaling update improved. Now allocating 28 ranks in slurm, but using 27 of them. Need to crank up number of nodes, rather than ranks. Exact weak scaling test should be run without metis. Could go up to 7x7x7 on cosma7.   - RDMA updates (Peter) Some progress implementing one sided mpi in swift. Single exchanges. Runs a bit slowly. Trying to use windowed mpi. Works for 2 nodes, but not more. About as fast for 2 nodes as normal mpi. Should then be easy to plug in rdma rather than mpi. Talking between nodes, tearing down and up could become a bottleneck. - Abouzied update: Getting to grips with the code, compiling etc, getting ideas of how to use dualsphysics with swift. Initially focussing on getting gpu kernels to work with swift. Then move on to async comms with cuda. Start with the hydro part? Compile in minimal sph setup. Run a cubic box with nothing in it and see where the different parts are being called. - Individual workpackage update [all] - Training (Tobias?) - Outreach (?)   - Others/AOB?