Revision 7faae9b10b4475ce4566f9cbbe45f77aac366452 authored by edwards on 12 March 2005, 18:43:57 UTC, committed by edwards on 12 March 2005, 18:43:57 UTC
1 parent 8922869
TODO
Laundry List - 23 Feb 2005
\item (*) H_T ContFrac cost-vs-mres tests
\item (*) Add force monitoring for HMC. Currently, we have energy
contributions for each monomial. More useful is contributions to force
for each monomial. This monitor is simply norm2(ds_u)
\item (*) Bagel/QMP-2 tests for IBM Bluegene/L and NERSC SP. {\em B\'alint: BAGEL QMP2 Runs on BlueGene/L. QMP-2 is a big but fixed hit. NERSC Still to be Done}
\item (*) PARTFILE I/O in QIO, for efficient I/O on QCDOC.
Based on force monitoring tests:
\item Fractional power monomials???
\item Sexton-Weingarten integration for gauge piece ala Degrand???
\item SAP for force term???
\item Stout-link deriv. for HMC. How do we add stout smearing for the
actions??? Is this a new "state" ?? We can't overload on the arg types
to the factory calls.
\item Aniso precond. and even-odd aniso precond. for linops
\item Aniso precond. and even-odd aniso precond. for HMC
\item Aniso support in GaugeActs - its nearly there, only IO.
\item Make generic update "harness" that also works for pure-gauge
heatbath.
\item Make heat-bath production ready. Need over-relaxation. Need I/O !
\item Need some kind of reader/writer (??) for ILDG meta-data conventions???
BJ points out management stuff not workable. Could do physics portion.
\item Make xcheck that uses XMLDiff for complete regression tests.
\item A. Hart says the subset-ted reunit is slow - should use
QDP's profile and isolate.
DONE:
\item Change check_PROGRAMS to bin_PROGRAMS and install binaries
in chroma/mainprogs/main
\item Simple aniso for DWF - fix weirdness
\item 4D Chrono Predictor Done. (23/Feb/2005) - 5D Chrono Predictor Done (24/Feb 2004)
\item BAGEL Running on Bluegene/L using QCDOC target and QMP_v2. IBM needs checking/porting still (potential XLC/isms). Detailed report to write.
Computing file changes ...