Gives O(32 x 8 x 18*8*8) chunk size on configuration I/O.
At 150KB should be getting close to packet sizes and 4MB filesystem
block sizes that are reasonably (!?) performant. We shall see once I move
this off my laptop and over to BNL and time it.
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.
class, changing the nature of covariant Cshifts used in
plaquettes, rectangles and staples.
As a result same code is used for the plaq and rect action independent of the BC type.
Should probably isolate the BC in a separate class that Gimpl takes as a template param.
Do the same with smearing policies.
This would then allow composition of BC with smearing etc....
Can save and restore RNG state via new (serial) I/O routines in a NERSC header style file.
Store a Parallel (one per site) and a single serial RNG file.
preparing to pass integrator, smearing, bc's as policy classes to hmc.
Propose to unify "integrator" and integrator algorithm in a base/derived
way to override step. Want to read through ForceGradient to ensure
that abstraction covers the force gradient case.
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.
