50
50
one of 'tensor' (default) or 'quadrature' (experimental).
53
Specify code generation options. The list of options available
54
depends on the specified language (format). Current options
55
include -fprecision=n, -fquadrature_points=n,
56
-fsplit, -fblas and -fno-foo, described in detail
53
Specify code generation options. The list of options available
54
depends on the specified language (format). Current options
55
include -fblas, -fno-foo, -fprecision=n,
56
-fprecompute_basis_const, -fprecompute_ip_const,
57
-fquadrature_degree=n and, -fsplit, described in detail below.
60
Generate code that uses BLAS to compute tensor products.
61
This option is currently ignored, but can be used to reduce the
62
code size when the BLAS option is (re-)implemented in future
66
Don't generate code for UFC function with name 'foo'. Typical
67
options include -fno-evaluate_basis and
68
-fno-evaluate_basis_derivatives to reduce the size of the
69
generated code when these functions are not needed.
60
72
Set the number of significant digits to n in the generated code.
61
73
The default value of n is 15.
64
Will generate a quadrature rule accurate up to order n regardless
65
of the polynomial order of the form. This option is only valid
66
for UFL forms and the specified order will apply to ALL terms of
67
the given form for which no order has been specified through
68
metadata! As default FFC will determine the order automatically
71
-f quadrature_points=n
72
Will generate n quadrature points in each spatial direction
73
regardless of the polynomial order of the form. This option is
74
only valid for the standard FFC forms i.e., *.form files, not
75
UFL forms. The specified number of points will apply to ALL terms
76
of the given form! As default FFC will determine the number of
77
quadrature points needed for exact representation of the form.
75
-f precompute_basis_const
76
Additional optimisation option for quadrature representation.
77
This option is ignored if optimisation is not used
78
(see -O option), and it also implies the precompute_ip_const
79
option. This option will generate code that precompute terms
80
which are constant in the loops involving basis indices. This can
81
result in a reduction of the operation count and thereby
82
improve the runtime efficiency of the generated code. However,
83
the improvements depends on the GCC compiler options as well as
84
the characteristics of the variational form.
86
-f precompute_ip_const
87
Like the precompute_basis_const option with the only difference
88
that code will be generated to compute terms which are constant
89
in the loops involving the integration points only.
91
-f quadrature_degree=n
92
Will generate a quadrature rule accurate up to degree n
93
regardless of the polynomial degree of the form. This option is
94
only valid for UFL forms and the specified degree will apply to
95
ALL terms of the given form for which no degree has been
96
specified through metadata! As default FFC will determine the
97
degree automatically from the form.
80
100
Generate separate files for declarations and the implementation.
83
Generate code that uses BLAS to compute tensor products. This
84
option is currently ignored, but can be used to reduce the code
85
size when the BLAS option is (re-)implemented in future versions.
88
Don't generate code for UFC function with name 'foo'. Typical
89
options include -fno-evaluate_basis and -fno-evaluate_ba-
90
sis_derivatives to reduce the size of the generated code when
91
these functions are not needed.
94
103
Generate optimized code with a lower operation count compared to
95
104
non-optimized code for the assembly of the local element tensor.
added
removed
doc/manual/chapters/manpagecopy.tex
