forked from milc-qcd/milc_qcd
-
Notifications
You must be signed in to change notification settings - Fork 0
/
README_RELEASE_NOTES
306 lines (228 loc) · 9.99 KB
/
README_RELEASE_NOTES
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
This is MILC Version 7 code
Please note that a few of the application directories contain code
that has not yet been ported to version 7. The following are
supported application directories:
arb_overlap
clover_dynamical
clover_invert2
ext_src
file_utilities
gauge_utilities
gluon_prop
hvy_qpot
ks_eigen
ks_imp_dyn
ks_imp_rhmc
ks_imp_utilities
ks_measure
ks_spectrum
pure_gauge
schroed_cl_inv
schroed_ks_dyn
smooth_inst
symanzik_sl32
u1_gauge
The following are provided, but not supported:
arb_dirac_eigen
arb_dirac_invert
clover_hybrids
h_dibaryon
schroed_pg
Build the code with GNU make. See the README file in this directory
for general instructions.
Here are highlights of the changes
For version 7.8.1
Added support for QPhiX single and multimass staggered inversions
Upgraded regression tests for ks_spectrum to allow QPhiX testing
Added path product optimization by Rama Maladi and Steve Gottlieb
Added improved QPhiX remapping
More threading
Restructured the top-level Makefile to provide variations for different architectures
For version 7.8.0
Added Hiroshi Ono's staggered CG deflation and eigcg support to ks_measure
Added Doug's updated random number generator for thread-safe operation and large lattices
For version 7.7.13
More quark source/sink operators
Added u1_gauge
For version 7.7.11
Wilson flow code
Clover and KS spectrum code supports linear combinations, extended sources,
and extended propagation.
Some Open MP threading.
Support for QUDA-0.5.0
(Note: no multi-GPU support for the ks_imp_dyn/su3_rmd target)
Quenched U(1) spectrum
Quenched improved Fermilab (OK) action
HYP smearing for hvy_qpot
Variable stride for covariant Gaussian smearing
KS and clover spectrum code supports time slice projection and sequential inversion
Some bug fixes
For version 7.7.5
Add staggered/clover hopping matrix source/sink operator for
spectrum calculations.
Support for general interpolating operators for staggered mesons.
GPU/QUDA support for single-GPU HISQ molecular dynamics. multi-GPU
support for asqtad (and HISQ) inverters.
Consolidated support for clover and KS sources, allowing still more
flexibility.
Support for naive propagators arising from extended Dirac sources.
New ks_measure for computing staggered scalar operators, such
as psi-bar-psi, quark number susceptibilities, etc.
Support for HISQ EOS, including Naik epsilon derivatives.
Momentum twist and boundary twist capability.
HISQ fermion force support in QOP plus some performance counters.
Overhaul of the fermion links treatment.
Added arb_overlap from DeGrand for computing eigenvalues and
chirality of the overlap operator.
For version 7.6.3
Standardization of byte counts for the header in the open-meson file
generated by clover_invert2/su3_clov.
Optimizations for memory and speed in clover_invert2 and ext_src
Support multifile sources in ext_src.
Latest updates of HISQ code from Alexei Bazavov.
Further changes in metadata for FNAL correlator files
Support open meson correlators for B_K project
Add several new clover sources
Add FFTW support
Add gauge_utilities for translating the gauge field and
standalone gauge fixing
Add support for internal translation of sources for clover_invert*
Updated MILC HISQ treatment
For version 7.6.2
Revise metadata for FNAL correlator files in clover_invert2
Add QOP and QDP support for clover inversions.
Add heavy-light-light baryon code from Heechang Na.
Add ks_eigen code from version 6 and fix QDP support.
Add ext_src for generating extended source from propagators
Add some spectrum calculations to HISQ targets in ks_imp_rhmc
Add HISQ fermion code from Doug and Alexei
Add link and action coeffs to all fermion_force calls.
Remove global accessor calls for these coeffs.
Change normalization of point sink props in ks_hl_spectrum
fix seg fault, and allow suppression of correlators in log file.
Fix nrestart treatment in QOP inverter calls.
For version 7.6.0
Adds pw_nr_meson. Does P-wave charmonium.
Support for FNAL correlator format for ks_hl_spectrum.
Support for double-precision NERSC archive lattices.
clover_invert2 enlarges the capabilities of clover_invert.
Support for changed QOPQDP Asqtad inverter conventions
in qopqdp-0.10.1 with a backward compatibility option in Makefile.
Partial preparation for HISQ: move calls for fat link creation out of
all staggered dslash and congrad. Changes API for all ks_congrad
and dslash calls - hence a new minor version number.
Fix bugs in io_source_w_fm.c and in io_helpers_w.c.
Add capability of reading a sink smearing function.
For version 7.5.0
Precision selection for QOP and QDP Asqtad fermion force
(single & multisource)
Change to true residual stopping condition in single mass staggered
and clover inverters. Prepare for supporting FNAL relative residual.
Support for fixed mesh layouts (e.g. BG/L and QCDOC)
Support a variety of FNAL Wilson sources and propagator formats.
Support for USQCD/SciDAC gauge file format in some applications.
Dcache I/O support (FNAL)
Drop site structure temps from gaugefix args.
Drop site structure args from restrict_fourier args
New interface for clover inverters.
Z2 random number support in f_meas
Support qopqdp-0.7.7 and later for fermion force.
EVENANDODD KS matrix inversion now does odd-site polish
Support for QCDOC multi Asqtad force and multi Asqtad inverter
A few more less-used version 6 apps upgraded to version 7.
For version 7.4.0
Mixed precision for QOP and QDP inverters, dslash, and supporting
fermion links.
Introduced precision selection "prec_pbp" for f_meas.c inversions.
Made multicg option choice available to applications other than RHMC.
Support qopqdp-0.7.7 including FNMAT version of multi fermion force
"initialize_machine" now takes pointers to argc and **argv.
Support USQCD metadata in SciDAC file output
Print list of selected options during setup.
Segregate inclusion of SciDAC headers for faster compilation times
Libraries "make" builds both precisions independently.
Make reporting of remapping time optional with macro REMAP.
For version 7.3.1
Polished the parameter input format for the RHMC.
Now allow comment lines beginning with # in parameter input.
Added NERSC checksum to link trace.
Added more precise timing to the QOP and QDP modules
including remapping times.
For version 7.3.0
Nearly complete QOP and QDP support for all major components of
Asqtad molecular dynamics, including Doug's new multi fermion force.
Reworked RHMC parameter input and Remez parameter output.
Added support for more than two pseudofermion fields in RHMC
Support different CG residuals for RHMC MD, FA, GR.
Changed inline global make macros to C_GLOBAL_INLINE and
SSE_GLOBAL_INLINE.
Added file pointer to get_i, get_f, etc.
Added nrestart to KS CG routines throughout.
(These last changes affect most application subdirectories.)
For version 7.2.2
Upgrade RHMC code to do concurrent EOS calculations
Fix p4 capability in ks_imp_dyn.
Distinguish inverter residuals for mol dyn and action.
For version 7.2.1
Start supporting the QOP/QDP package
Split out fermion_links_fn from quark_stuff but no QOP support
for fat and long link generation, yet.
Make sample files precision-specific
For version 7.2.0.1
Correct the weights for QOP_asqtad_force_multi in fermion_force_asqtad_qop.c
Restore clobbered sse directory.
For version 7.2.0
Port clover_dynamical to Version 7
Add RHMC code and rearrange multicg and multi fermion force procedures
Add Level 3 wrappers for Level 3 multicg and multi fermion force
Begin adding plaquette and link trace to MILC file metadata.
Add Urs's gluon_prop
For version 7.1.12
Fix Level 3 fermion force coef usage
Support new level 3 interface for asqtad_invert and asqtad_force
Add clover_invert to Version 7.
For version 7.1.11
Support new Level 3 API for Asqtad fermion force and inverter
Support for QOP/QDP awaits upgrade.
Change write_ksprop_info_item to use fwrite instead of fprintf
Add MILC imitations of Level 3 force and inverter for testing
Make dummy QCDML be legitimate XML
For previous versions 7
Unified Makefile replaces Make_vanilla Make_linux_mpi, etc.
See various macros in Makefile, including MPP.
Requires GNU make.
Unified support for single and double precision
See the PRECISION macro in Makefile.
Support for SciDAC QIO
See the WANTQIO macro in Makefile
Support for SciDAC QDP
See the WANTQDP macro in Makefile
Support for single and double precision SSE
Uses the same SSE and SSE_INLINE macros as in version 6,
except that P3 or P4 compiler macros are not used.
Standalone MILC/FNAL to SciDAC file conversion in file_utilities.
Uses only temp vectors for dslash_fn and d_congrad5_fn.
New applications:
ks_hl_spectrum: Ludmila's MILC version for the clover heavy
staggered light spectrum.
ks_imp_invert_multi: multimass inversion
ks_imp_utilities: for checking fermion force and inverter
New file_utilities
Conversion among binary file types for KS and Wilson propagators.
Differencing for propagator files.
Unified I/O for KS propagators supporting Wingate, Fermilab
and SciDAC formats.
Unified I/O for Wilson propagators supporitng MILC, Fermilab,
and SciDAC formats.
Time-slice checksums for Fermilab KS and Wilson propagators.
QCDOC support - including the SciDAC Level 3 inverter
C-coded optimized fermion force term for QCDOC
An assortment of C-inline routines
Support for SciDAC Level 3 QOP for QCDOC
Support for ILDG lattice formats and QIO parallel I/O.
Few-gather and few-sum versions of the MILC inverter.
Support for parallel I/O and ILDG LFN
Modified multimass inverter to allow arbitrary source points.
Add C versions of MILC QOP wrappers. Make wrappers support QOP/QDP.
C. DeTar, Utah
15 February 2016