repo.or.cz
/
official-gcc.git
/
search
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
log
|
graphiclog1
|
graphiclog2
|
commit
|
commitdiff
|
tree
|
refs
|
edit
|
fork
first
·
prev
·
next
The existing vector costs stop some beneficial vectorization. This is mostly
2016-11-14
wilco
Th
e
existing
v
ector
c
os
t
s stop som
e
be
n
eficial vectorization
.
.
.
commit
|
commitdiff
|
tree
2016-11-02
w
ilc
o
Add LE/BE SHA1H patt
e
r
ns with a V2SI input
.
This av
o
i
d
s
.
.
.
commit
|
commitdiff
|
tree
2016-11-02
wilc
o
The add
expande
r
still contains som
e
expansi
o
n c
o
de
.
.
.
commit
|
commitdiff
|
tree
2016-10-25
wilco
With -fpu=neon DI mode shifts ar
e
expand
e
d
aft
e
r reload
.
.
.
commit
|
commitdiff
|
tree
2016-10-21
wilco
Really add the test t
h
is time
.
.
.
commit
|
commitdiff
|
tree
2016-10-21
wi
l
co
I
m
prove stack adju
s
tment b
y
reusing a temporary
move
.
.
.
commit
|
commitdiff
|
tree
2016-10-21
w
ilco
If t
h
e number o
f
integer
callee-saves is
o
d
d
, any FP
.
.
.
commit
|
commitdiff
|
tree
2016-10-05
wilco
Move all e
x
isting strchr and str
r
chr f
o
l
d
ing fr
o
m
b
u
iltin
s
.
.
.
commit
|
commitdiff
|
tree
2016-09-28
wilco
If st
r
chr can't b
e
f
o
l
d
ed in gimple-fold, we st
i
ll
.
.
.
commit
|
commitdiff
|
tree
2016-09-28
wilco
Op
t
imize strchr (
s
, 0
)
to s
+ strlen (s)
.
s
trc
h
r
.
.
.
commit
|
commitdiff
|
tree
2016-09-07
wilco
Imp
r
ove aarch64_
l
egitimize_addre
s
s - avoid sp
l
itting
.
.
.
commit
|
commitdiff
|
tree
2016-09-01
wilc
o
This
p
a
t
ch add
s
legitimize_ad
d
ress
_
displacement hook
.
.
.
commit
|
commitdiff
|
tree
2016-08-01
wilc
o
This patch optimiz
e
s
t
he prolog and epilog code to
.
.
.
commit
|
commitdiff
|
tree
2016-07-28
wilco
O
n
AArch64 the UXTB and
UXTH instruc
t
ions ar
e
ali
a
ses
.
.
.
commit
|
commitdiff
|
tree
2016-07-28
wilco
Th
i
s patchse
t
improves zero e
x
tend
c
o
sts and co
d
e gen
e
ration
.
commit
|
commitdiff
|
tree
2016-07-28
wilco
T
h
is
p
atch imp
r
oves
the readability of the
prol
o
g
and
.
.
.
commit
|
commitdiff
|
tree
2016-07-06
w
i
lco
T
hi
s
pat
c
h impro
v
e
s the
accuracy of the Cortex-A53
.
.
.
commit
|
commitdiff
|
tree
2016-07-06
wil
c
o
Fix prototy
p
e in vs
t
1Q_l
a
neu64-1
.
c t
o
u
nsigne
d
char*
.
commit
|
commitdiff
|
tree
2016-06-30
wilco
This patc
h
sets the
b
ranch cos
t
to the same
most optimal
.
.
.
commit
|
commitdiff
|
tree
2016-06-29
w
i
lco
Increase loop al
i
gnment
on Cortex
cor
e
s
to 8 a
n
d set
.
.
.
commit
|
commitdiff
|
tree
2016-06-28
wilco
This p
a
t
ch
f
ixes a
b
ug
i
n
the
b
swap pass
.
In
big-e
n
dian
.
.
.
commit
|
commitdiff
|
tree
2016-06-21
wi
l
co
T
he recently added
g
c
c
.
target/aarch64/advsimd-int
r
i
nsi
c
s
.
.
.
commit
|
commitdiff
|
tree
2016-06-21
wilco
Fix tree-ssa/attr-hotcold-2
.
c
failu
r
es now that the
.
.
.
commit
|
commitdiff
|
tree
2016-06-21
wil
c
o
Due to r
e
cent impro
v
ements to the vectoriz
e
r
,
the n
u
m
ber
.
.
.
commit
|
commitdiff
|
tree
2016-06-20
w
ilco
Th
i
s
patch cleans up
t
he -mp
c
-relat
i
ve
-
loads
o
ption
.
.
.
commit
|
commitdiff
|
tree
2016-06-20
w
ilco
Improve modes_tieable by
r
eturni
n
g true in more ca
s
e
s
.
.
.
commit
|
commitdiff
|
tree
2016-06-20
wilco
The
C
ortex-A5
7
s
c
heduler
i
s missin
g
f
csel, so add it
.
commit
|
commitdiff
|
tree
2016-05-27
wilco
Remov
e
aarch64_cannot_chan
g
e
_
m
o
de_cla
s
s as the underlying
.
.
.
commit
|
commitdiff
|
tree
2016-05-26
wilco
GCC expan
d
s switch st
a
t
ements in a ver
y
simplist
i
c
.
.
.
commit
|
commitdiff
|
tree
2016-05-26
wilco
SIMD operations like combine prefer to
h
ave
their o
p
erands
.
.
.
commit
|
commitdiff
|
tree
2016-05-16
wilco
Update
d
ocumenta
t
i
o
n of AArch64 opt
i
o
n
s for GC
C
6 t
o
.
.
.
commit
|
commitdiff
|
tree
2016-05-16
w
ilco
Some pattern
s
are
usi
n
g '%
w
2
'
f
o
r immediate ope
r
a
n
ds
.
.
.
commit
|
commitdiff
|
tree
2016-05-16
wilco
This pat
c
h fi
x
es the
a
ttribut
e
s o
f
int
e
ger imm
e
diate
.
.
.
commit
|
commitdiff
|
tree
2016-02-17
wilco
A
s
part of the final clea
n
u
p
o
f
a
a
rch64_in
t
e
rnal_mov_imme
d
i
a
.
.
.
commit
|
commitdiff
|
tree
2016-02-10
wilco
Enable instruc
t
ion fusion of
d
ependent AESE; AE
S
M
C
.
.
.
commit
|
commitdiff
|
tree
2016-02-10
wilco
Ena
b
le instructi
o
n fusion of AES in
s
t
r
u
c
t
i
o
n
s
on ARM
.
.
.
commit
|
commitdiff
|
tree
2016-02-04
wilco
T
h
is
p
atc
h
fixes an expo
n
ential issue in c
c
mp
.
c
.
Wh
e
n
.
.
.
commit
|
commitdiff
|
tree
2016-02-03
wil
c
o
F
ix the
c
cm
p
_1
.
c test
b
ack to use
'
0'
a
s re
g
ular expre
s
s
i
ons
.
.
.
commit
|
commitdiff
|
tree
2016-02-02
wil
c
o
Improve TARGET_IRA_C
H
ANGE_PSEUDO_ALLOCNO
_
CL
A
SS t
a
rg
e
t
.
.
.
commit
|
commitdiff
|
tree
2016-02-02
w
ilco
T
h
is patc
h
adds support for the TARGET_IRA_C
H
ANGE_PSEUDO_ALL
.
.
.
commit
|
commitdiff
|
tree
2016-01-28
wilco
A recent chan
g
e a
d
ded
U
NSPEC to the CCMP patterns to
.
.
.
commit
|
commitdiff
|
tree
2016-01-28
wilco
Add support for ve
c
tor
p
ermu
t
e
c
ost
s
i
n
c
e
v
ar
i
ous p
e
rmutes
.
.
.
commit
|
commitdiff
|
tree
2016-01-28
wilco
Several instruct
i
ons
d
is
a
ssemble a zero imm
e
diate
a
s
.
.
.
commit
|
commitdiff
|
tree
2016-01-27
wil
c
o
2015-01
-
2
7
Wilco Dijkstra <wdijkstr@arm
.
com>
commit
|
commitdiff
|
tree