public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug tree-optimization/52112] New: Vectorizer fails when using CRTP
@ 2012-02-03 17:07 ddesics at gmail dot com
2012-02-03 17:40 ` [Bug tree-optimization/52112] " pinskia at gcc dot gnu.org
` (4 more replies)
0 siblings, 5 replies; 6+ messages in thread
From: ddesics at gmail dot com @ 2012-02-03 17:07 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52112
Bug #: 52112
Summary: Vectorizer fails when using CRTP
Classification: Unclassified
Product: gcc
Version: 4.6.1
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: tree-optimization
AssignedTo: unassigned@gcc.gnu.org
ReportedBy: ddesics@gmail.com
Created attachment 26565
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=26565
The test case code.
Using the CRTP along with static_cast pointers prevents auto-vectorization.
If the code below is compiled with -ftree-vectorizer-verbose=7, the CRTP method
fails to vectorize with "not vectorized: control flow in loop." However, if
compiled with -O3 -S -fno-tree-vectorize, both methods produce identical
assembly.
I don't know how difficult this would be to change, but it could certainly
speed up a lot of c++ code. For instance, this currently prevents boost ublas
from vectorizing.
#include<iostream>
template<typename E, typename Tp> class CRTP_base {
public:
typedef E& reference;
typedef Tp value_type;
reference operator()() { return *static_cast<E*>(this); }
value_type square() { return (*this)().x() * (*this)().x(); }
protected:
CRTP_base() {}
~CRTP_base() {}
};
template<typename Tp> class CRTP_child : public CRTP_base<CRTP_child<Tp>,Tp> {
Tp xval;
typedef CRTP_base<CRTP_child<Tp>,Tp> parent;
public:
CRTP_child(Tp xv = Tp()) : xval(xv) {}
Tp x() { return xval; }
using parent::square;
};
int main() {
const int N = 100;
double A[N] __attribute__((aligned(16)));
double B[N] __attribute__((aligned(16)));
double sum1=0.0;
for(int i = 0; i < N; ++i) { A[i] = i; }
for(int i = 0; i < N; ++i) { B[i] = A[i]*A[i]; }
for(int i = 0; i < N; ++i) { sum1 += B[i]; }
std::cout << "Sum of method 1: " << sum1;
for(int i = 0; i < N; ++i) { B[i] = CRTP_child<double>(A[i]).square(); }
for(int i = 0; i < N; ++i) { sum1 += B[i]; }
std::cout << "\nSum of method 2: " << sum2 << std::endl;
return 0;
}
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug tree-optimization/52112] Vectorizer fails when using CRTP
2012-02-03 17:07 [Bug tree-optimization/52112] New: Vectorizer fails when using CRTP ddesics at gmail dot com
@ 2012-02-03 17:40 ` pinskia at gcc dot gnu.org
2012-02-03 17:43 ` pinskia at gcc dot gnu.org
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: pinskia at gcc dot gnu.org @ 2012-02-03 17:40 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52112
--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> 2012-02-03 17:40:23 UTC ---
This is vectorized for me on x86_64 on the trunk.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug tree-optimization/52112] Vectorizer fails when using CRTP
2012-02-03 17:07 [Bug tree-optimization/52112] New: Vectorizer fails when using CRTP ddesics at gmail dot com
2012-02-03 17:40 ` [Bug tree-optimization/52112] " pinskia at gcc dot gnu.org
@ 2012-02-03 17:43 ` pinskia at gcc dot gnu.org
2012-02-03 17:45 ` pinskia at gcc dot gnu.org
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: pinskia at gcc dot gnu.org @ 2012-02-03 17:43 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52112
--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> 2012-02-03 17:43:41 UTC ---
Even in 4.4, both multiplication loops are vectorized
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug tree-optimization/52112] Vectorizer fails when using CRTP
2012-02-03 17:07 [Bug tree-optimization/52112] New: Vectorizer fails when using CRTP ddesics at gmail dot com
2012-02-03 17:40 ` [Bug tree-optimization/52112] " pinskia at gcc dot gnu.org
2012-02-03 17:43 ` pinskia at gcc dot gnu.org
@ 2012-02-03 17:45 ` pinskia at gcc dot gnu.org
2012-02-03 18:25 ` ddesics at gmail dot com
2012-02-03 19:01 ` jakub at gcc dot gnu.org
4 siblings, 0 replies; 6+ messages in thread
From: pinskia at gcc dot gnu.org @ 2012-02-03 17:45 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52112
--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> 2012-02-03 17:44:36 UTC ---
<bb 10>:
vect_var_.146 = MEM[base: &A, index: ivtmp.196];
MEM[base: &B, index: ivtmp.196] = [mult_expr] vect_var_.146 * vect_var_.146;
ivtmp.196 = ivtmp.196 + 16;
if (ivtmp.196 != 800)
goto <bb 10>;
else
goto <bb 11>;
Is what we get for 4.4 which is obviously vectorized.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug tree-optimization/52112] Vectorizer fails when using CRTP
2012-02-03 17:07 [Bug tree-optimization/52112] New: Vectorizer fails when using CRTP ddesics at gmail dot com
` (2 preceding siblings ...)
2012-02-03 17:45 ` pinskia at gcc dot gnu.org
@ 2012-02-03 18:25 ` ddesics at gmail dot com
2012-02-03 19:01 ` jakub at gcc dot gnu.org
4 siblings, 0 replies; 6+ messages in thread
From: ddesics at gmail dot com @ 2012-02-03 18:25 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52112
--- Comment #4 from Daniel Davis <ddesics at gmail dot com> 2012-02-03 18:24:49 UTC ---
Any thoughts on why it won't vectorize for me on x86_64 4.6.1?
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug tree-optimization/52112] Vectorizer fails when using CRTP
2012-02-03 17:07 [Bug tree-optimization/52112] New: Vectorizer fails when using CRTP ddesics at gmail dot com
` (3 preceding siblings ...)
2012-02-03 18:25 ` ddesics at gmail dot com
@ 2012-02-03 19:01 ` jakub at gcc dot gnu.org
4 siblings, 0 replies; 6+ messages in thread
From: jakub at gcc dot gnu.org @ 2012-02-03 19:01 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52112
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |RESOLVED
CC| |jakub at gcc dot gnu.org
Resolution| |FIXED
--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-02-03 19:00:44 UTC ---
Before http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=181014
only 2 loops are vectorized, after that 3. With -Ofast before that change 4
loops are vectorized and after that 5 (the double reduction needs -ffast-math
to be vectorized).
So fixed for 4.7.
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2012-02-03 19:01 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-02-03 17:07 [Bug tree-optimization/52112] New: Vectorizer fails when using CRTP ddesics at gmail dot com
2012-02-03 17:40 ` [Bug tree-optimization/52112] " pinskia at gcc dot gnu.org
2012-02-03 17:43 ` pinskia at gcc dot gnu.org
2012-02-03 17:45 ` pinskia at gcc dot gnu.org
2012-02-03 18:25 ` ddesics at gmail dot com
2012-02-03 19:01 ` jakub at gcc dot gnu.org
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).