diff options
author | Jakub Jelinek <jakub@redhat.com> | 2021-10-12 09:28:10 +0200 |
---|---|---|
committer | Jakub Jelinek <jakub@redhat.com> | 2021-10-12 09:28:10 +0200 |
commit | fab2f61dc1b195fec9161be15635b87d312e4626 (patch) | |
tree | 6c12bfb3db92b9265cbb18c8d5d73d0c9c646637 /gcc | |
parent | 73c535a00bc4dfe9a939cd80facbe79a929cab3e (diff) | |
download | gcc-fab2f61dc1b195fec9161be15635b87d312e4626.zip gcc-fab2f61dc1b195fec9161be15635b87d312e4626.tar.gz gcc-fab2f61dc1b195fec9161be15635b87d312e4626.tar.bz2 |
vectorizer: Fix up -fsimd-cost-model= handling
> * testsuite/libgomp.c++/scan-10.C: Add option -fvect-cost-model=cheap.
I don't think this is the right thing to do.
This just means that at some point between 2013 when -fsimd-cost-model has
been introduced and now -fsimd-cost-model= option at least partially stopped
working properly.
As documented, -fsimd-cost-model= overrides the -fvect-cost-model= setting
for OpenMP simd loops (loop->force_vectorize is true) if specified differently
from default.
In tree-vectorizer.h we have:
static inline bool
unlimited_cost_model (loop_p loop)
{
if (loop != NULL && loop->force_vectorize
&& flag_simd_cost_model != VECT_COST_MODEL_DEFAULT)
return flag_simd_cost_model == VECT_COST_MODEL_UNLIMITED;
return (flag_vect_cost_model == VECT_COST_MODEL_UNLIMITED);
}
and use it in various places, but we also just use flag_vect_cost_model
in lots of places (and in one spot use flag_simd_cost_model, not sure if
we are sure it is a force_vectorize loop or what).
So, IMHO we should change the above inline function to
loop_cost_model and let it return the cost model and then just
reimplement unlimited_cost_model as
return loop_cost_model (loop) == VECT_COST_MODEL_UNLIMITED;
and then adjust the direct uses of the flag and revert these changes.
2021-10-12 Jakub Jelinek <jakub@redhat.com>
gcc/
* tree-vectorizer.h (loop_cost_model): New function.
(unlimited_cost_model): Use it.
* tree-vect-loop.c (vect_analyze_loop_costing): Use loop_cost_model
call instead of flag_vect_cost_model.
* tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
(vect_prune_runtime_alias_test_list): Likewise. Also use it instead
of flag_simd_cost_model.
gcc/testsuite/
* gcc.dg/gomp/simd-2.c: Remove option -fvect-cost-model=cheap.
* gcc.dg/gomp/simd-3.c: Likewise.
libgomp/
* testsuite/libgomp.c/scan-11.c: Remove option -fvect-cost-model=cheap.
* testsuite/libgomp.c/scan-12.c: Likewise.
* testsuite/libgomp.c/scan-13.c: Likewise.
* testsuite/libgomp.c/scan-14.c: Likewise.
* testsuite/libgomp.c/scan-15.c: Likewise.
* testsuite/libgomp.c/scan-16.c: Likewise.
* testsuite/libgomp.c/scan-17.c: Likewise.
* testsuite/libgomp.c/scan-18.c: Likewise.
* testsuite/libgomp.c/scan-19.c: Likewise.
* testsuite/libgomp.c/scan-20.c: Likewise.
* testsuite/libgomp.c/scan-21.c: Likewise.
* testsuite/libgomp.c/scan-22.c: Likewise.
* testsuite/libgomp.c++/scan-9.C: Likewise.
* testsuite/libgomp.c++/scan-10.C: Likewise.
* testsuite/libgomp.c++/scan-11.C: Likewise.
* testsuite/libgomp.c++/scan-12.C: Likewise.
* testsuite/libgomp.c++/scan-13.C: Likewise.
* testsuite/libgomp.c++/scan-14.C: Likewise.
* testsuite/libgomp.c++/scan-15.C: Likewise.
* testsuite/libgomp.c++/scan-16.C: Likewise.
Diffstat (limited to 'gcc')
-rw-r--r-- | gcc/testsuite/gcc.dg/gomp/simd-2.c | 2 | ||||
-rw-r--r-- | gcc/testsuite/gcc.dg/gomp/simd-3.c | 2 | ||||
-rw-r--r-- | gcc/tree-vect-data-refs.c | 10 | ||||
-rw-r--r-- | gcc/tree-vect-loop.c | 4 | ||||
-rw-r--r-- | gcc/tree-vectorizer.h | 16 |
5 files changed, 22 insertions, 12 deletions
diff --git a/gcc/testsuite/gcc.dg/gomp/simd-2.c b/gcc/testsuite/gcc.dg/gomp/simd-2.c index f491212..85acb98 100644 --- a/gcc/testsuite/gcc.dg/gomp/simd-2.c +++ b/gcc/testsuite/gcc.dg/gomp/simd-2.c @@ -1,5 +1,5 @@ /* { dg-do compile } */ -/* { dg-options "-O2 -fopenmp -fvect-cost-model=cheap -fdump-tree-vect-details" } */ +/* { dg-options "-O2 -fopenmp -fdump-tree-vect-details" } */ /* { dg-additional-options "-msse2" { target { i?86-*-* x86_64-*-* } } } */ /* { dg-additional-options "-mavx" { target avx } } */ /* { dg-final { scan-tree-dump-times "vectorized \[1-9]\[0-9]* loops in function" 5 "vect" { target i?86-*-* x86_64-*-* aarch64-*-* } } } */ diff --git a/gcc/testsuite/gcc.dg/gomp/simd-3.c b/gcc/testsuite/gcc.dg/gomp/simd-3.c index c75060c..86fee85 100644 --- a/gcc/testsuite/gcc.dg/gomp/simd-3.c +++ b/gcc/testsuite/gcc.dg/gomp/simd-3.c @@ -1,5 +1,5 @@ /* { dg-do compile } */ -/* { dg-options "-O2 -fopenmp -fvect-cost-model=cheap -fdump-tree-vect-details" } */ +/* { dg-options "-O2 -fopenmp -fdump-tree-vect-details" } */ /* { dg-additional-options "-msse2" { target { i?86-*-* x86_64-*-* } } } */ /* { dg-additional-options "-mavx" { target avx } } */ /* { dg-final { scan-tree-dump-times "vectorized \[1-9]\[0-9]* loops in function" 5 "vect" { target i?86-*-* x86_64-*-* aarch64-*-* } } } */ diff --git a/gcc/tree-vect-data-refs.c b/gcc/tree-vect-data-refs.c index 07b5ba1..1e13148 100644 --- a/gcc/tree-vect-data-refs.c +++ b/gcc/tree-vect-data-refs.c @@ -2236,7 +2236,7 @@ vect_enhance_data_refs_alignment (loop_vec_info loop_vinfo) { unsigned max_allowed_peel = param_vect_max_peeling_for_alignment; - if (flag_vect_cost_model <= VECT_COST_MODEL_CHEAP) + if (loop_cost_model (loop) <= VECT_COST_MODEL_CHEAP) max_allowed_peel = 0; if (max_allowed_peel != (unsigned)-1) { @@ -2334,7 +2334,7 @@ vect_enhance_data_refs_alignment (loop_vec_info loop_vinfo) do_versioning = (optimize_loop_nest_for_speed_p (loop) && !loop->inner /* FORNOW */ - && flag_vect_cost_model > VECT_COST_MODEL_CHEAP); + && loop_cost_model (loop) > VECT_COST_MODEL_CHEAP); if (do_versioning) { @@ -3751,7 +3751,9 @@ vect_prune_runtime_alias_test_list (loop_vec_info loop_vinfo) unsigned int count = (comp_alias_ddrs.length () + check_unequal_addrs.length ()); - if (count && flag_vect_cost_model == VECT_COST_MODEL_VERY_CHEAP) + if (count + && (loop_cost_model (LOOP_VINFO_LOOP (loop_vinfo)) + == VECT_COST_MODEL_VERY_CHEAP)) return opt_result::failure_at (vect_location, "would need a runtime alias check\n"); @@ -3760,7 +3762,7 @@ vect_prune_runtime_alias_test_list (loop_vec_info loop_vinfo) "improved number of alias checks from %d to %d\n", may_alias_ddrs.length (), count); unsigned limit = param_vect_max_version_for_alias_checks; - if (flag_simd_cost_model == VECT_COST_MODEL_CHEAP) + if (loop_cost_model (LOOP_VINFO_LOOP (loop_vinfo)) == VECT_COST_MODEL_CHEAP) limit = param_vect_max_version_for_alias_checks * 6 / 10; if (count > limit) return opt_result::failure_at diff --git a/gcc/tree-vect-loop.c b/gcc/tree-vect-loop.c index e94356d..961c162 100644 --- a/gcc/tree-vect-loop.c +++ b/gcc/tree-vect-loop.c @@ -1850,7 +1850,7 @@ vect_analyze_loop_costing (loop_vec_info loop_vinfo) /* If using the "very cheap" model. reject cases in which we'd keep a copy of the scalar code (even if we might be able to vectorize it). */ - if (flag_vect_cost_model == VECT_COST_MODEL_VERY_CHEAP + if (loop_cost_model (loop) == VECT_COST_MODEL_VERY_CHEAP && (LOOP_VINFO_PEELING_FOR_ALIGNMENT (loop_vinfo) || LOOP_VINFO_PEELING_FOR_GAPS (loop_vinfo) || LOOP_VINFO_PEELING_FOR_NITER (loop_vinfo))) @@ -1922,7 +1922,7 @@ vect_analyze_loop_costing (loop_vec_info loop_vinfo) /* If the vector loop needs multiple iterations to be beneficial then things are probably too close to call, and the conservative thing would be to stick with the scalar code. */ - if (flag_vect_cost_model == VECT_COST_MODEL_VERY_CHEAP + if (loop_cost_model (loop) == VECT_COST_MODEL_VERY_CHEAP && min_profitable_estimate > (int) vect_vf_for_cost (loop_vinfo)) { if (dump_enabled_p ()) diff --git a/gcc/tree-vectorizer.h b/gcc/tree-vectorizer.h index ed4a7ff..4aa84ac 100644 --- a/gcc/tree-vectorizer.h +++ b/gcc/tree-vectorizer.h @@ -1701,14 +1701,22 @@ get_dr_vinfo_offset (vec_info *vinfo, } +/* Return the vect cost model for LOOP. */ +static inline enum vect_cost_model +loop_cost_model (loop_p loop) +{ + if (loop != NULL + && loop->force_vectorize + && flag_simd_cost_model != VECT_COST_MODEL_DEFAULT) + return flag_simd_cost_model; + return flag_vect_cost_model; +} + /* Return true if the vect cost model is unlimited. */ static inline bool unlimited_cost_model (loop_p loop) { - if (loop != NULL && loop->force_vectorize - && flag_simd_cost_model != VECT_COST_MODEL_DEFAULT) - return flag_simd_cost_model == VECT_COST_MODEL_UNLIMITED; - return (flag_vect_cost_model == VECT_COST_MODEL_UNLIMITED); + return loop_cost_model (loop) == VECT_COST_MODEL_UNLIMITED; } /* Return true if the loop described by LOOP_VINFO is fully-masked and |