Pasenau Andújar 2014a - Revision history

Cinmemj: Cinmemj moved page Draft Samper 107379809 to Pasenau Andújar 2014a

2018-03-26T13:05:26Z

Cinmemj moved page Draft Samper 107379809 to Pasenau Andújar 2014a

Cinmemj at 13:04, 26 March 2018

2018-03-26T13:04:37Z

Cinmemj at 12:49, 26 March 2018

2018-03-26T12:49:10Z

Cinmemj at 12:43, 26 March 2018

2018-03-26T12:43:57Z

Cinmemj at 12:42, 26 March 2018

2018-03-26T12:42:07Z

Cinmemj at 12:37, 26 March 2018

2018-03-26T12:37:05Z

Cinmemj at 12:35, 26 March 2018

2018-03-26T12:35:27Z

Cinmemj at 12:32, 26 March 2018

2018-03-26T12:32:56Z

Cinmemj at 12:32, 26 March 2018

2018-03-26T12:32:00Z

Cinmemj at 12:30, 26 March 2018

2018-03-26T12:30:42Z

@@ Line 2,291: / Line 2,291: @@
 {| style="width: 100%;border-collapse: collapse;"
 |-
-|  style="vertical-align: top;"|<span style="text-align: center; font-size: 75%;">int simplify_hash_uc( int grid_size,           // simplification grid = grid_size ^ 3</span>
+|  [[File:Draft_Samper_107379809_8776_codigo1.png]]
-−
-<span style="text-align: center; font-size: 75%;">t_simplification_type_t simp_type,</span>
-−
-<span style="text-align: center; font-size: 75%;">const float *lst_points_in,</span>
-−
-<span style="text-align: center; font-size: 75%;">const int num_points_in,</span>
-−
-<span style="text-align: center; font-size: 75%;">const int *lst_triangles_in,</span>
-−
-<span style="text-align: center; font-size: 75%;">const int num_triangles_in,</span>
-−
-<span style="text-align: center; font-size: 75%;">const float **lst_scalars, // may be NULL</span>
-−
-<span style="text-align: center; font-size: 75%;">const int num_scalars, // may be NULL</span>
-−
-<span style="text-align: center; font-size: 75%;">float **lst_points_out, int *num_points_out,</span>
-−
-<span style="text-align: center; font-size: 75%;">int **lst_lines_out, int *num_lines_out, // may be NULL</span>
-−
-<span style="text-align: center; font-size: 75%;">int **lst_triangles_out, int *num_triangles_out,</span>
-−
-<span style="text-align: center; font-size: 75%;">float ***lst_scalars_out, int *size_lst_scalars_out); // may be NULL</span>
 |}
@@ Line 2,323: / Line 2,301: @@
 For more complex meshes, lengthy sessions, or when the attributes should cell-averaged on-demand, the handler approach can be used with following procedures:
-// begin simplification session
+{| style="width: 100%;border-collapse: collapse;"
+|-
-void *shc_begin( int grid_size,           // simplification grid = grid_size ^ 3
+[[File:Draft_Samper_107379809_9352_codigo2.png]]
-t_simplification_type_t simp_type,
+|}
-−
-const float *lst_points_in,
-−
-const int num_points_in);
-−
-// simplify mesh using geometry or scalar or vector
-−
-int shc_simplify( void *handler,
-−
-const float *lst_points_in,
-−
-const int num_points_in,
-−
-const int *lst_triangles_in,
-−
-const int num_triangles_in,
-−
-const float **lst_scalars, // used if simp_type uses scalar/vector
-−
-const int num_scalars, // used if simp_type uses scalar/vector
-−
-float **lst_points_out, int *num_points_out,
-−
-int **lst_lines_out, int *num_lines_out, // may be NULL
-−
-int **lst_triangles_out, int *num_triangles_out,
-−
-float ***lst_scalars_out, int *size_lst_scalars_out); // optimal values
-−
-// clears cell contents, but not grid, hash table, so that it can be used to interpolate
-−
-void shc_clear_cells( void *handler);
-−
-// interpolate attributes, make sure to clear cell contents first
-−
-int shc_interpolate_scalar( void *handler,
-−
-const float *lst_points_in,
-−
-const int num_points_in,
-−
-const float *lst_scalars,
-−
-float **lst_scalar_out);
-−
-// interpolate a max of 6 scalars
-−
-int shc_interpolate_scalars( void *handler,
-−
-const float *lst_points_in,
-−
-const int num_points_in,
-−
-const float **lst_scalars,
-−
-const int num_scalars,
-−
-float ***lst_scalars_out, int *num_scalars_out);
-−
-size_t shc_get_memory_size( void *handler);
-−
-// end simplification session
-−
-void shc_end( void *handler);
 ==5.2. Usage in GiD==

@@ Line 2,193: / Line 2,193: @@
 As the PLY format only supports scalar attributes, the following convention has been adopted to define vector attributes which will be used in the simplification algorithm that includes the vector information into the QEF matrix. The program uses as vector the first three attributes which begin with these prefixes:
-{| style="margin: 1em auto 0.1em auto;border-collapse: collapse;width:100%;text-aling:center;width:60%;"
+{| style="margin: 1em auto 0.1em auto;border-collapse: collapse;text-align:center;width:60%;"
 |-
-| style="border: 1pt solid black;vertical-align: top;"|vx, vy, vz
+| style="vertical-align: top;"|vx, vy, vz
 |-
-| style="vertical-align: top;text-aling:center;"|red, green, blue
+| style="text-align:center;"|red, green, blue
 |-
-| style="vertical-align: top;text-aling:center;"|nx, ny, nz
+| style="text-align:center;"|nx, ny, nz
 |-
-|style="vertical-align: top;text-aling:center;"|r, g, b
+|style="text-align:center;"|r, g, b
 |-
-|style="vertical-align: top;text-aling:center;"|x, y, z
+|style="text-align:center;"|x, y, z
 |-
 |  style="vertical-align: top;text-align:center;"|<span style="text-align: center; font-size: 75%;">'''Table 5.1: '''List of prefixes used to identify groups of three attributes as vector in the geometry + vector guided simplification process.</span>

@@ Line 2,193: / Line 2,193: @@
 As the PLY format only supports scalar attributes, the following convention has been adopted to define vector attributes which will be used in the simplification algorithm that includes the vector information into the QEF matrix. The program uses as vector the first three attributes which begin with these prefixes:
-{| style="margin: 1em auto 0.1em auto;border-collapse: collapse;width:100%;"
+{| style="margin: 1em auto 0.1em auto;border-collapse: collapse;width:100%;text-aling:center;width:60%;"
 |-
-| style="border: 1pt solid black;vertical-align: top;text-aling:center;"|vx, vy, vz
+| style="border: 1pt solid black;vertical-align: top;"|vx, vy, vz
 |-
-| style="border: 1pt solid black;vertical-align: top;text-aling:center;"|red, green, blue
+| style="vertical-align: top;text-aling:center;"|red, green, blue
 |-
-| style="border: 1pt solid black;vertical-align: top;text-aling:center;"|nx, ny, nz
+| style="vertical-align: top;text-aling:center;"|nx, ny, nz
 |-
-|style="border: 1pt solid black;vertical-align: top;text-aling:center;"|r, g, b
+|style="vertical-align: top;text-aling:center;"|r, g, b
 |-
-|style="border: 1pt solid black;vertical-align: top;text-aling:center;"|x, y, z
+|style="vertical-align: top;text-aling:center;"|x, y, z
 |-
-|  style="vertical-align: top;"|<span style="text-align: center; font-size: 75%;">'''Table 5.1: '''List of prefixes used to identify groups of three attributes as vector in the geometry + vector guided simplification process.</span>
+|  style="vertical-align: top;text-align:center;"|<span style="text-align: center; font-size: 75%;">'''Table 5.1: '''List of prefixes used to identify groups of three attributes as vector in the geometry + vector guided simplification process.</span>
 |}

@@ Line 2,195: / Line 2,195: @@
 {| style="margin: 1em auto 0.1em auto;border-collapse: collapse;width:100%;"
 |-
-| style="vertical-align: top;"|vx, vy, vz
+| style="border: 1pt solid black;vertical-align: top;text-aling:center;"|vx, vy, vz
-|style="vertical-align: top;"|red, green, blue
+|-
-|style="vertical-align: top;"|nx, ny, nz
+| style="border: 1pt solid black;vertical-align: top;text-aling:center;"|red, green, blue
-|style="vertical-align: top;"|r, g, b
+|-
-|style="vertical-align: top;"|x, y, z
+| style="border: 1pt solid black;vertical-align: top;text-aling:center;"|nx, ny, nz
 |-
 |  style="vertical-align: top;"|<span style="text-align: center; font-size: 75%;">'''Table 5.1: '''List of prefixes used to identify groups of three attributes as vector in the geometry + vector guided simplification process.</span>

@@ Line 2,193: / Line 2,193: @@
 As the PLY format only supports scalar attributes, the following convention has been adopted to define vector attributes which will be used in the simplification algorithm that includes the vector information into the QEF matrix. The program uses as vector the first three attributes which begin with these prefixes:
-{| style="margin: 1em auto 0.1em auto;border-collapse: collapse;font-size:85%;width:100%;"
+{| style="margin: 1em auto 0.1em auto;border-collapse: collapse;width:100%;"
 |-
 | style="vertical-align: top;"|vx, vy, vz
-|red, green, blue
+|style="vertical-align: top;"|red, green, blue
-|nx, ny, nz
+|style="vertical-align: top;"|nx, ny, nz
-|r, g, b
+|style="vertical-align: top;"|r, g, b
-|x, y, z
+|style="vertical-align: top;"|x, y, z
 |-
 |  style="vertical-align: top;"|<span style="text-align: center; font-size: 75%;">'''Table 5.1: '''List of prefixes used to identify groups of three attributes as vector in the geometry + vector guided simplification process.</span>

← Older revision	Revision as of 13:05, 26 March 2018
(No difference)

← Older revision		Revision as of 12:32, 26 March 2018
Line 1,836:		Line 1,836:
	</nowiki>		</nowiki>

−	~~==endif==~~
−
−	~~'''omp_unset_lock'''( &lst_locks[ idx_to_insert]); // release lock for location idx_to_insert~~

	With this modification, this implementation of the hash table needs extra memory to store the locks. To avoid using the memory that this algorithm saves to hold the locks table, one lock is used to block the access to several consecutive locations.		With this modification, this implementation of the hash table needs extra memory to store the locks. To avoid using the memory that this algorithm saves to hold the locks table, one lock is used to block the access to several consecutive locations.

@@ Line 1,827: / Line 1,827: @@
 As ''atomicMax() ''operation is not present in the OpenMP standard''  [OpenMP11, OpenMP13], this operation will be emulated using the lock mechanism''. The line
-old_key = '''atomicMax'''( &hash_table[ idx_to_insert].key(), key_to_insert);
+ <nowiki>
+ old_key = '''atomicMax'''( &hash_table[ idx_to_insert].key(), key_to_insert);
 is transformed to:
+ '''omp_set_lock'''( &lst_locks[ idx_to_insert]); // get lock for location idx_to_insert
-'''omp_set_lock'''( &lst_locks[ idx_to_insert]); // get lock for location idx_to_insert
+ old_key = hash_table[ idx_to_insert].key();
+ '''if''' ( key_to_insert > old_key) '''then'''
-old_key = hash_table[ idx_to_insert].key();
+ hash_table[ idx_to_insert].key() = key_to_insert;
+</nowiki>
 '''if''' ( key_to_insert > old_key) '''then'''
-−
 hash_table[ idx_to_insert].key() = key_to_insert;
 ==endif==

@@ Line 1,579: / Line 1,579: @@
 ==4.2. Single-level cuckoo hash==
 '''Concept'''
@@ Line 1,610: / Line 1,611: @@
 We target only CPU implementations and the shared memory parallelization scheme we used is OpenMP, which has no ''atomicExchange() operation [OpenMP11, OpenMP13]'':
-<nowiki>
+ <nowiki>
   #pragma omp atomic capture
   {v = x; x = expr; }    // not supported in OpenMP 3.1, 4.0