2015-05-06 Thomas Koenig <tkoenig@gcc.gnu.org>

	PR fortran/37131
	* gfortran.h (gfc_isym_id):  Add GFC_ISYM_FE_RUNTIME_ERROR.
	(gfc_intrinsic_sym):  Add vararg.
	* intrinsic.h (gfc_check_fe_runtime_error):  Add prototype.
	(gfc_resolve_re_runtime_error):  Likewise.
	Add prototype for gfc_is_reallocatable_lhs.
	* trans-array.h (gfc_is_reallocatable_lhs):  Remove prototype.
	* check.c (gfc_check_fe_runtime_error):  New function.
	* intrinsic.c (add_sym_1p):  New function.
	(make_vararg):  New function.
	(add_subroutines):  Add fe_runtime_error.
	(gfc_intrinsic_sub_interface): Skip sorting for variable number
	of arguments.
	* iresolve.c (gfc_resolve_fe_runtime_error):  New function.
	* lang.opt (inline-matmul-limit):  New option.
	(gfc_post_options): If no inline matmul limit has been set and
	BLAS is called externally, use the BLAS limit.
	* frontend-passes.c:  Include intrinsic.h.
	(var_num):  New global counter for naming temporary variablbles.
	(matrix_case):  Enum for differentiating the different matmul
	cases.
	(realloc_string_callback):  Add "trim" to the variable name.
	(create_var): Add optional argument vname as part of the name.
	Use var_num. Set dimension of result correctly. Split off block
	creation into
	(insert_block): New function.
	(cfe_expr_0): Use "fcn" as part of temporary variable name.
	(optimize_namesapce): Also set gfc_current_ns. Call
	inline_matmul_assign.
	(combine_array_constructor):  Use "constr" as part of
	temporary name.
	(get_array_inq_function):  New function.
	(build_logical_expr):  New function.
	(get_operand):  new function.
	(inline_limit_check):  New function.
	(runtime_error_ne):  New function.
	(matmul_lhs_realloc):  New function.
	(is_functino_or_op):  New function.
	(has_function_or_op):  New function.
	(freeze_expr):  New function.
	(freeze_references):  New function.
	(convert_to_index_kind):  New function.
	(create_do_loop):  New function.
	(get_size_m1):  New function.
	(scalarized_expr):  New function.
	(inline_matmul_assign):  New function.
	* simplify.c (simplify_bound):  Simplify the case of the
	lower bound of an assumed-shape argument.

2015-05-06  Thomas Koenig  <tkoenig@gcc.gnu.org>

	PR fortran/37131
	* gfortran.dg/dependency_26.f90: Add option to suppress inlining
	matmul.
	* gfortran.dg/function_optimize_1.f90:  Likewise.
	* gfortran.dg/function_optimize_2.f90:  Likewise.
	* gfortran.dg/function_optimize_5.f90:  Likewise.
	* gfortran.dg/function_optimize_7.f90:  Likewise.
	* gfortran.dg/inline_matmul_1.f90:  New test.
	* gfortran.dg/inline_matmul_2.f90:  New test.
	* gfortran.dg/inline_matmul_3.f90:  New test.
	* gfortran.dg/inline_matmul_4.f90:  New test.
	* gfortran.dg/inline_matmul_5.f90:  New test.
	* gfortran.dg/inline_matmul_6.f90:  New test.



git-svn-id: https://gcc.gnu.org/svn/gcc/trunk@222864 138bc75d-0d04-0410-961f-82ee72b054a4

1 files changed, 23 insertions, 5 deletions

diff --git a/gcc/fortran/invoke.texi b/gcc/fortran/invoke.texi
index af123dbbe5a..fc9f1b0fa1d 100644
--- a/gcc/fortran/invoke.texi
+++ b/gcc/fortran/invoke.texi
@@ -178,6 +178,7 @@ and warnings}.
 -finit-character=@var{n} -finit-integer=@var{n} -finit-local-zero @gol
 -finit-logical=@var{<true|false>}
 -finit-real=@var{<zero|inf|-inf|nan|snan>} @gol
+-finline-matmul-limit=@var{n} @gol
 -fmax-array-constructor=@var{n} -fmax-stack-var-size=@var{n}
 -fno-align-commons @gol
 -fno-automatic -fno-protect-parens -fno-underscoring @gol
@@ -1537,6 +1538,22 @@ geometric mean of the dimensions of the argument and result matrices.
 
 The default value for @var{n} is 30.
 
+@item -finline-matmul-limit=@var{n}
+@opindex @code{finline-matmul-limit}
+When front-end optimiztion is active, some calls to the @code{MATMUL}
+intrinsic function will be inlined.  This may result in code size
+increase if the size of the matrix cannot be determined at compile
+time, as code for both cases is generated.  Setting
+@code{-finline-matmul-limit=0} will disable inlining in all cases.
+Setting this option with a value of @var{n} will produce inline code
+for matrices with size up to @var{n}. If the matrices involved are not
+square, the size comparison is performed using the geometric mean of
+the dimensions of the argument and result matrices.
+
+The default value for @var{n} is the value specified for
+@code{-fblas-matmul-limit} if this option is specified, or unlimitited
+otherwise.
+
 @item -frecursive
 @opindex @code{frecursive}
 Allow indirect recursion by forcing all local arrays to be allocated
@@ -1632,11 +1649,12 @@ if @option{-ffrontend-optimize} is in effect.
 @cindex Front-end optimization
 This option performs front-end optimization, based on manipulating
 parts the Fortran parse tree.  Enabled by default by any @option{-O}
-option.  Optimizations enabled by this option include elimination of
-identical function calls within expressions, removing unnecessary
-calls to @code{TRIM} in comparisons and assignments and replacing
-@code{TRIM(a)} with @code{a(1:LEN_TRIM(a))}. 
-It can be deselected by specifying @option{-fno-frontend-optimize}.
+option.  Optimizations enabled by this option include inlining calls
+to @code{MATMUL}, elimination of identical function calls within
+expressions, removing unnecessary calls to @code{TRIM} in comparisons
+and assignments and replacing @code{TRIM(a)} with
+@code{a(1:LEN_TRIM(a))}.  It can be deselected by specifying
+@option{-fno-frontend-optimize}.
 @end table
 
 @xref{Code Gen Options,,Options for Code Generation Conventions,