12 Commits

Author	SHA1	Message	Date
azusayamaguchi	fad96cf250	StencilBufs	2016-10-21 13:36:00 +01:00
paboyle	a762b1fb71	MPI3 working with a bounce through shared memory on my laptop. ... Longer term plan: make the "u_comm_buf" in Stencil point to the shared region and avoid the send between ranks on same node.	2016-10-21 09:03:26 +01:00
paboyle	5fe2b85cbd	MPI3 and shared memory support	2016-10-20 16:58:01 +01:00
paboyle	32bc7a6ab8	MPI back out of change that hangs ... AVX2 for clang, gcc needs the -mfma flag.	2016-08-05 10:36:00 +01:00
paboyle	ef97e32152	Adding persistent communicators	2016-07-08 17:16:08 +01:00
paboyle	d6b64f47d9	Uint64 sum for IO rates	2016-03-16 02:27:22 -07:00
Peter Boyle	6aeaf6f568	Parallel IO worked on. I'm puzzled because I already thought I shook this out on MacOS + OpenMPI and then ... turned up problems on the BlueWaters Cray. Gets 75MB/s from home filesystem on parallel configuration read. Need to make the RNG IO parallel, and also to look at aggregating bigger writes for the parallel write. Not sure what the home filesystem is.	2016-02-21 08:03:21 -06:00
Peter Boyle	41c2b09184	Shmem comms [NO MPI] target added. The dwf test runs and passes. ... Not really shaken out to my satisfaction though as I want more tests done, so don't declare as working. But committing my current while I try a few experimentals.	2016-02-14 14:24:38 -06:00
paboyle	e2f73e3ead	Updates for shmem	2016-02-10 16:50:32 -08:00
paboyle	aae8bf31a7	Global edit adding copyright and license info to every source file.	2016-01-02 14:51:32 +00:00
Peter Boyle	dc814f30da	Binary IO file for generic Grid array parallel I/O. ... Number of IO MPI tasks can be varied by selecting which dimensions use parallel IO and which dimensions use Serial send to boss I/O. Thus can neck down from, say 1024 nodes = 4x4x8x8 to {1,8,32,64,128,256,1024} nodes doing the I/O. Interpolates nicely between ALL nodes write their data, a single boss per time-plane in processor space [old UKQCD fortran code did this], and a single node doing all I/O. Not sure I have the transfer sizes big enough and am not overly convinced fstream is guaranteed to not give buffer inconsistencies unless I set streambuf size to zero. Practically it has worked on 8 tasks, 2x1x2x2 writing /cloning NERSC configurations on my MacOS + OpenMPI and Clang environment. It is VERY easy to switch to pwrite at a later date, and also easy to send x-strips around from each node in order to gather bigger chunks at the syscall level. That would push us up to the circa 8x 18*4*8 == 4KB size write chunk, and by taking, say, x/y non parallel we get to 16MB contiguous chunks written in multi 4KB transactions per IOnode in 64^3 lattices for configuration I/O. I suspect this is fine for system performance.	2015-08-26 13:40:29 +01:00
Peter Boyle	1d0df449e8	Reorganise of file naming	2015-06-03 12:47:05 +01:00