On Saturday 19 March 2011 19:59:56 Thomas Koenig wrote:
> Am 19.03.2011 00:23, schrieb Tobias Burnus:
> > I have not followed the discussion nor have I fully read the patch, but
> > what's the reason for allowing ELEMENTAL functions?
>
> Here's an updated version of the patch, which removes the elemental
> functions as well. I have also added an option which allows full access
> to all function call eliminations, so if any user wants it, it is there.
> (I know I will use it :-) This option is not enabled by any
> optimization option.
>
> Regression-tested. Before committing, I'll check on the status of the
> gfc_free removal patch, and re-test. Also tested with "make dvi" and
> "make info".
>
> OK for trunk?
Not yet, comment here, nits below.
I'm a bit worried about the patch possibly conflicting with other
optimizations at code generation time. I'm thinking especially about cases
where we manage not to create a temporary without the patch:
- Inline intrinsics have only been for scalar cases so far, so it should not
matter for now.
- Array elementals: here it's hard to tell which one is faster; save function
calls to a temporary and use the temporary or do multiple function calls every
time (but without temporary).
- Transpose optimization: here the transpose call is changed into a direct
array access, so your patch will definitely make things worse. Even if
transpose calls are used multiple times as actual argument for example, better
create multiple descriptors than copy the whole lot to a temp. The common
function elimination should be disabled (IMO) in this case.
Mikael.
> Index: frontend-passes.c
> ===================================================================
> --- frontend-passes.c (Revision 170960)
> +++ frontend-passes.c (Arbeitskopie)
> @@ -106,11 +128,237 @@ optimize_expr (gfc_expr **e, int *walk_subtrees AT
> return 0;
> }
>
> +/* Compare two functions for equality. We could use gfc_dep_compare_expr
> + except that we also consider impure functions equal, because anybody
> + changing the return value of the function within an expression would
> + violate the Fortran standard. */
Given how much the codes match, it looks like one can move the relevant
gfc_dep_compare_expr code to a function, use that function, driving the small
differences with a flag, no?
> +
> +static bool
> +compare_functions (gfc_expr **ep1, gfc_expr **ep2)
> +{
> + gfc_expr *e1, *e2;
> +
> + e1 = *ep1;
> + e2 = *ep2;
> +
> + if (e1->expr_type != EXPR_FUNCTION || e2->expr_type != EXPR_FUNCTION)
> + return false;
> +
> + if ((e1->value.function.esym && e2->value.function.esym
> + && e1->value.function.esym == e2->value.function.esym)
> + || (e1->value.function.isym && e2->value.function.isym
> + && e1->value.function.isym == e2->value.function.isym))
> + {
> + gfc_actual_arglist *args1, *args2;
> +
> + args1 = e1->value.function.actual;
> + args2 = e2->value.function.actual;
> +
> + /* Compare the argument lists for equality. */
> + while (args1 && args2)
> + {
> + /* Bitwise xor, since C has no non-bitwise xor operator. */
> + if ((args1->expr == NULL) ^ (args2->expr == NULL))
> + return false;
> +
> + if (args1->expr != NULL && args2->expr != NULL
> + && gfc_dep_compare_expr (args1->expr, args2->expr) != 0)
> + return false;
> +
> + args1 = args1->next;
> + args2 = args2->next;
> + }
> + return args1 == NULL && args2 == NULL;
> + }
> + else
> + return false;
> +
> +}
> +
> +/* Callback function for gfc_expr_walker, called from cfe_expr_0. Put all
> + eligible function expressions into expr_array. We can't do allocatable
> + functions. */
> +
> +static int
> +cfe_expr (gfc_expr **e, int *walk_subtrees ATTRIBUTE_UNUSED,
> + void *data ATTRIBUTE_UNUSED)
Please use more descriptive names.
There are 3 cfe* functions and only one has a comment explaining what cfe
means.
Something like "register_function_expr" seems to match what this function
does, but you may propose something better ;-)
> +{
> + if ((*e)->expr_type != EXPR_FUNCTION)
> + return 0;
> +
> + /* We don't do character functions (yet). */
> + if ((*e)->ts.type == BT_CHARACTER)
> + return 0;
> +
> + /* If we don't know the shape at compile time, we do not create a
temporary
> + variable to hold the intermediate result. FIXME: Change this later
when
> + allocation on assignment works for intrinsics. */
> +
> + if ((*e)->rank > 0 && (*e)->shape == NULL)
> + return 0;
> +
> + /* Skip the test for pure functions if -faggressive-function-elimination
> + is specified. */
> + if (!gfc_option.flag_aggressive_function_elimination)
> + {
> + if ((*e)->value.function.esym)
> + {
> + if ((*e)->value.function.esym->attr.allocatable)
> + return 0;
Is it expected that you allow allocatables with -faggressive-function-
elimination?
> +
> + if (!(*e)->value.function.esym->attr.pure
> + && !(*e)->value.function.esym->attr.implicit_pure)
> + return 0;
> + }
> + }
> +
> + if ((*e)->value.function.isym)
> + {
> + if ((*e)->value.function.isym->id == GFC_ISYM_CONVERSION)
> + return 0;
> +
> + if (! (*e)->value.function.isym->pure
> + && !(*e)->value.function.isym->elemental)
Tobias' comment also applies here, even if there is no intrinsic impure
elemental: the following code in intrinsic.c makes the elemental check
redundant.
next_sym->pure = (cl != CLASS_IMPURE);
next_sym->elemental = (cl == CLASS_ELEMENTAL);
> + return 0;
> + }
> +
> + if (expr_count >= expr_size)
> + {
> + expr_size += expr_size;
> + expr_array = XRESIZEVEC(gfc_expr **, expr_array, expr_size);
> + }
> + expr_array[expr_count] = e;
> + expr_count ++;
> + return 0;
> +}
> +
> +/* Returns a new expression (a variable) to be used in place of the old
one,
> + with an an assignment statement before the current statement to set
> + the value of the variable. */
> +
> +static gfc_expr*
> +create_var (gfc_expr * e)
> +{
> + char name[GFC_MAX_SYMBOL_LEN +1];
> + static int num = 1;
> + gfc_symtree *symtree;
> + gfc_symbol *symbol;
> + gfc_expr *result;
> + gfc_code *n;
> + int i;
> +
> + sprintf(name, "__var_%d",num++);
> + if (gfc_get_sym_tree (name, current_ns, &symtree, false) != 0)
> + gcc_unreachable ();
> +
> + symbol = symtree->n.sym;
> + symbol->ts = e->ts;
> + symbol->as = gfc_get_array_spec ();
> + symbol->as->rank = e->rank;
> + symbol->as->type = AS_EXPLICIT;
> + for (i=0; i<e->rank; i++)
> + {
> + gfc_expr *p, *q;
> +
> + p = gfc_get_constant_expr (BT_INTEGER, gfc_default_integer_kind,
> + &(e->where));
> + mpz_set_si (p->value.integer, 1);
> + symbol->as->lower[i] = p;
> +
> + q = gfc_get_constant_expr (BT_INTEGER, gfc_index_integer_kind,
> + &(e->where));
> + mpz_set (q->value.integer, e->shape[i]);
> + symbol->as->upper[i] = q;
> + }
> +
> + symbol->attr.flavor = FL_VARIABLE;
> + symbol->attr.referenced = 1;
> + symbol->attr.dimension = e->rank > 0;
> + gfc_commit_symbol (symbol);
> +
> + result = gfc_get_expr ();
> + result->expr_type = EXPR_VARIABLE;
> + result->ts = e->ts;
> + result->rank = e->rank;
> + result->shape = gfc_copy_shape (e->shape, e->rank);
> + result->symtree = symtree;
> + result->where = e->where;
> + if (e->rank > 0)
> + {
> + result->ref = gfc_get_ref ();
> + result->ref->type = REF_ARRAY;
> + result->ref->u.ar.type = AR_FULL;
> + result->ref->u.ar.where = e->where;
> + result->ref->u.ar.as = symbol->as;
> + }
> +
> + /* Generate the new assignment. */
> + n = XCNEW (gfc_code);
> + n->op = EXEC_ASSIGN;
> + n->loc = (*current_code)->loc;
> + n->next = *current_code;
> + n->expr1 = gfc_copy_expr (result);
> + n->expr2 = e;
> + *current_code = n;
> +
> + return result;
> +}
> +
> +/* Callback function for the code walker for doing common function
> + elimination. This builds up the list of functions in the expression
> + and goes through them to detect duplicates, which it then replaces
> + by variables. */
> +
> +static int
> +cfe_expr_0 (gfc_expr **e, int *walk_subtrees,
> + void *data ATTRIBUTE_UNUSED)
> +{
> + int i,j;
> + gfc_expr *newvar;
> +
> + expr_count = 0;
> + gfc_expr_walker (e, cfe_expr, NULL);
> + /* Walk backwards through all the functions to make sure we
> + catch the leaf functions first. */
> + for (i=expr_count-1; i>=1; i--)
> + {
Tiny optimization here ;-):
/* Don't bother if the expression has been factored already. */
if ((*(expr_array[i]))->expr_type == EXPR_VARIABLE)
continue;
> + newvar = NULL;
> + for (j=i-1; j>=0; j--)
> + {
> + if (compare_functions(expr_array[i], expr_array[j]))
> + {
> + if (newvar == NULL)
> + newvar = create_var (*(expr_array[i]));
> + gfc_free (*(expr_array[j]));
> + *(expr_array[j]) = gfc_copy_expr (newvar);
> + }
> + }
> + if (newvar)
> + *(expr_array[i]) = newvar;
> + }
> +
> + /* We did all the necessary walking in this function. */
> + *walk_subtrees = 0;
> + return 0;
> +}
> +
Small comment here, or better, more descriptive name
> +static int
> +cfe_code (gfc_code **c, int *walk_subtrees ATTRIBUTE_UNUSED,
> + void *data ATTRIBUTE_UNUSED)
> +{
> + current_code = c;
> + return 0;
> +}
> +
> /* Optimize a namespace, including all contained namespaces. */
>
> static void
> optimize_namespace (gfc_namespace *ns)
> {
> +
> + current_ns = ns;
> +
> + gfc_code_walker (&ns->code, cfe_code, cfe_expr_0, NULL);
> gfc_code_walker (&ns->code, optimize_code, optimize_expr, NULL);
>
> for (ns = ns->contained; ns; ns = ns->sibling)
> Index: options.c
> ===================================================================
> --- options.c (Revision 170960)
> +++ options.c (Arbeitskopie)
> @@ -150,6 +150,7 @@ gfc_init_options (unsigned int decoded_options_cou
> gfc_option.flag_align_commons = 1;
> gfc_option.flag_protect_parens = 1;
> gfc_option.flag_realloc_lhs = -1;
> + gfc_option.flag_aggressive_function_elimination = 0;
>
> gfc_option.fpe = 0;
> gfc_option.rtcheck = 0;
> @@ -972,6 +973,10 @@ gfc_handle_option (size_t scode, const char *arg,
> gfc_option.flag_align_commons = value;
> break;
>
> + case OPT_faggressive_function_elimination:
> + gfc_option.flag_aggressive_function_elimination = value;
> + break;
> +
> case OPT_fprotect_parens:
> gfc_option.flag_protect_parens = value;
> break;
