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
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
|
README_1.0_to_2.0
=================
This document exists to help those who already have a working user defined
reader (using the 1.0 API) to change it into the 2.0 API format - if desired.
Note that you do not have to update your (1.0 API) user defined reader if it
is already working fine for you.
You should consider it if:
- efficieny gains are needed or
- you need access to complex variables or
- you need access to tensor variables or
- you need multiple timeset capability or
- you want to provide your own "border" elements (as opposed to EnSight's
computation of them)
As an indication of the differences that might be realized in efficency,
consider the following comparison on an unstructured model consisting of:
1,639,058 nodes
7,079,211 elements 240530 tria3
3984 quad4
5927663 tetra4
653 pyramid5
906381 penta6
12 parts
The same model was represented in EnSight6 and EnSight Gold format.
EnSight6 format into: EnSight Gold format into:
------------------------------------ -------------------------
EnSight7.1 |Ensight7.2 |Ensight 7.1 |EnSight7.2 |Ensight7.2
internal |internal |userd reader |internal |userd reader
reader |reader |(API 1.0) |reader |(API 2.0)
| | | |
Time | Mem |Time | Mem |Time | Mem |Time | Mem |Time | Mem
(sec)| (Mb) |(sec)| (Mb) |(sec)| (Mb) |(sec)| (Mb) |(sec)| (Mb)
----------- |----------- |----------- |----------- |-----------
@ part 4.3 27.6 | 3.5 28.4 | 4.0 27.6 | 3.3 8.8 | 3.3 8.9
loader | | | |
| | | |
after 14.0 243.4 |12.8 244.3 |49.8 475.8 | 6.0 211.5 | 6.2 211.6
loading all | | | |
12 parts | | | |
(non-visual) | | | |
| | | |
after 16.8 263.2 |16.0 264.2 |52.8 490.7 | 9.1 236.2 | 9.5 236.2
activate of | | | |
a vector. | | | |
^ ^
/|\ /|\
| |
| |
| |
Compare these two!
Significant is the inefficiency of the 1.0 API, and the fact that the
2.0 API has the same improved efficiency (both in speed and memory) as
the gold internal reader!
Note: Structured data will not show much difference between the two API's,
but it was more efficient initially.
=========================================================
A note on philosophical differences between the two API's:
=========================================================
API 1.0 deals with:
-------------------
-> global coordinate array & corresponding
-> global node id array
-> global nodal variables
-> for each part:
-> local element connectivities (grouped by type) & corresponding
-> local element ids
-> local elemental variables
The element connectivities, within parts, reference the global coordinate
array. If node ids are provided, the element connectivities have to be in
terms of the node ids. If node ids are not provided, the connectivities are in
terms of the (one-based) index number of each node in the global coordinate
array. Thus, node ids are more than labels - they are a part of the
connectivity referencing scheme. Element ids are purely labels.
This API was originally setup to try to make the interface to other codes as
straightforward as possible. Efficiency was not the major consideration.
EnSight must do a fair amount of work to get data provided in the manner
described above into the form that it uses internally. There is mapping that
has to be setup and maintained between the global arrays and the local part
arrays so that updating over time can be accomplished efficiently. There is
hashing that is required in order to deal efficently with node ids.
All of this leads to a considerable amount of temporary memory and processing,
in order to get a model read into EnSight.
API 2.0 deals with:
-------------------
-> for each part:
-> part coordinates & corresponding
-> part node ids
-> part nodal variables
-> part element connectivities (grouped by type) & corresponding
-> part element ids
-> part elemental variables
API 2.0 requires that the coordinates and corresponding nodal variables be
provided per part. This eliminates the global to local mapping with all its
associated temporary memory and processing time. The connectivity of the
elements in each part reference the node indicies of its own (one-based) part
coordinate array. The connectivity of the elements do not reference the nodes
according to node ids. Node ids (and element ids for that matter) are purely
labels for screen display and for query operations within EnSight. This
eliminates the need for node id hashing as a model is read.
The 2.0 API has been created for those needing more efficiency - both in terms
of memory use and speed. The increased efficiency is possible because data is
requested in a manner which more closely represents the way that EnSight
stores and manipulates information internally. The new API requests size
information and allocates the actual internal structures and arrays
accordingly. Pointers to these arrays are passed directly to you in the
routines which gather data, thus eliminating a considerable amount of
temporary memory (and allocation time) that is needed in the old
API. Depending on what you must do to get your data into the form required,
the memory savings and the speed improvement when loading models can be quite
significant!!
Additionally, the ability to handle tensor and complex variables has been
added to the new API, and support for multiple timesets is provided.
------------------------------------------------
So, with that said, if you determine that you want to convert your existing
reader to the new API format, The following may be helpful.
====================
First the Good News! The following routines are identical in both API's!!
==================== ----------------------------------------------------
USERD_bkup
USERD_get_block_coords_by_component
USERD_get_block_iblanking
USERD_get_changing_geometry_status
USERD_get_dataset_query_file_info
USERD_get_element_label_status
USERD_get_name_of_reader
USERD_get_node_label_status
USERD_get_number_of_files_in_dataset
USERD_get_number_of_model_parts
USERD_get_number_of_variables
USERD_set_filenames
USERD_stop_part_building
========================
Second, pretty Good News! The following routines have minor changes,
======================== namely a slight name change and the addition
of arguments related to complex data, constant
(Note, the name changes type, or self contained parts vs global coords.
are needed so both
API's can exist together) The arguments must be added, but depending on
your situation, many might simply be place
holders.
-------------------------------------------------------------------------------
-----------------------------------------------------
A) Changes related to imaginary flag for complex data
=====================================================
If you don't deal with complex variables, simply add
this flag to your argument list and ignore its value.
-----------------------------------------------------
API 1.0 API 2.0
------- -------
USERD_get_constant_value USERD_get_constant_val
( (
int which var int which_var,
int imag_data
) )
USERD_get_description_lines USERD_get_descrip_lines
( (
int which_type, int which_type,
int which_var, int which_var,
int imag_data,
char line1[Z_BUFL], char line1[Z_BUFL],
char line2[Z_BUFL] char line2[Z_BUFL]
) )
USERD_get_variable_value_at_specific USERD_get_var_value_at_specific
( (
int which_var, int which_var,
int which_node_or_elem, int which_node_or_elem,
int which_part, int which_part,
int which_elem_type, int which_elem_type,
int time_step, int time_step,
float values[3] float values[3],
int imag_data
) )
---------------------------------------------------------
B) Changes related to complex data info, and constant type
(and some of the multiple timeset support)
=========================================================
If you don't deal with complex variables, simply add the
arguments for var_complex, var_ifilename, and var_freq
and assign var_complex to be FALSE.
The argument var_contran needs to be added, and set
appropriately if you have constant variables, to indicate
if the constant variable is fixed for all time or varies
over time.
The argument var_timeset needs to be added, and set
appropriately.
---------------------------------------------------------
API 1.0 API 2.0
------- -------
USERD_get_variable_info USERD_get_gold_variable_info
( (
char **var_description, char **var_description,
char **var_filename, char **var_filename,
int *var_type, int *var_type,
int *var_classify int *var_classify,
int *var_complex,
char **var_ifilename,
float *var_freq,
int *var_contran,
int *var_timeset
) )
------------------------------------------------------
C) Changes related to self contained part coordinates
======================================================
The number_of_nodes argument needs to be added and
set for each part. This one is critical for you to do.
------------------------------------------------------
API 1.0 API 2.0
------- -------
USERD_get_part_build_info USERD_get_gold_part_build_info
( (
int *part_numbers, int *part_types,
int *part_types, int *part_types,
char *part_description[Z_BUFL], char *part_description[Z_BUFL],
int *number_of_nodes,
int *number_of_elements[Z_MAXTYPE], int *number_of_elements[Z_MAXTYPE],
int *ijk_dimensions[3], int *ijk_dimensions[3],
int *iblanking_options[6] int *iblanking_options[6]
) )
------------------------------------------------------
D) Changes related to multiple timeset support
======================================================
The timeset_number argument needs to be added for the
following three routines.
The multiple timeset support also includes the change
in B) above for USERD_get_gold_variable_info and the
last three new routines in the third section of this
readme file.
------------------------------------------------------
API 1.0 API 2.0
------- -------
USERD_get_number_of_time_steps USERD_get_num_of_time_steps
( (
void int timeset_number
) )
USERD_get_solution_times USERD_get_sol_times
( (
int timeset_number,
float *solution_times float *solution_times
) )
USERD_set_time_step USERD_set_time_set_and_step
( (
int timeset_number,
int time_step int time_step
) )
------------------------------------------------------
E) Changes related to global_extern.h
======================================================
Be sure to include the updated global_extern.h file that comes
with the EnSight 7.2 release (not the one from previous releases).
=================================================================
Third, deleted and new routines. (Here is where the work lies)
Several old routines are gone. You will have to create the new
routines that replace them. I think you will find in most cases
that your old routines will form the basis of the new routines,
and that it isn't too difficult to provide the information in
the new way.
The detailed specifications for these new routines can be found
in README_USERD_2.0 (or the headers in libuserd.c of the
dummy_gold or ensight_gold readers).
=================================================================
API 1.0 API 2.0
------- -------
These routines: replaced by the single routine:
--------------------------- -------------------------------
USERD_get_block_scalar_values USERD_get_var_by_component
USERD_get_block_vector_values_by_component
USERD_get_scalar_values
USERD_get_vector_values
These global coordinate routines: replaced by part coord routines:
--------------------------------- --------------------------------
USERD_get_global_coords USERD_get_part_coords
USERD_get_global_node_ids USERD_get_part_node_ids
USERD_get_number_of_global_nodes
These part connectivity routines: replaced by part by type routines:
--------------------------------- ----------------------------------
USERD_get_element_connectivities_for_part USERD_get_part_elements_by_type
USERD_get_element_ids_for_part USERD_get_part_element_ids_by_type
These are New Routines
----------------------
(Can be a dummy) -> USERD_exit_routine
(Can be a dummy) -> USERD_get_model_extents
(Required) -> USERD_get_reader_version
multiple timeset releated:
(Required) -> USERD_get_number_timesets
(Required) -> USERD_get_timeset_description
(Required) -> USERD_get_geom_timeset_number
border provided by the reader option:
(Required) -> USERD_get_border_availability
(Can be a dummy) -> USERD_get_border_elements_by_type
transient model allocation efficency:
(Can be a dummy) -> USERD_get_maxsize_info
Possible use with Server-of-Servers:
(Can be a dummy) -> USERD_set_server_number
|
