Commit Graph

19721 Commits

Author SHA1 Message Date
Chris Lattner
c44fff472c Keep the killed/dead sets sorted, so that "KillsRegister" can do a quick
binary search to test for membership.  This speeds up LLC a bit more on KC++,
e.g. on itanium from 16.6974s to 14.8272s, PPC from 11.4926s to 10.7089s and
X86 from 10.8128s to 9.7943s, with no difference in generated code (like all
of the RA patches).

With these changes, isel is the slowest pass for PPC/X86, but linscan+live
intervals is still > 50% of the compile time for itanium.  More work could
be done, but this is the last for now.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22993 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-24 00:09:02 +00:00
Chris Lattner
44b94c2185 adjust to new live variables interface
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22992 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 23:42:17 +00:00
Chris Lattner
d177579860 Adjust to new livevars interface
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22991 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 23:41:14 +00:00
Chris Lattner
e0cbf970ac Change live variables from using multimaps to using maps of vectors and
rearrange some of the accessors to be more efficient.

This makes it much more efficient to iterate over all of the things with the
same value.  This speeds up liveintervals analysis from 8.63s to 3.79s with
a release build of llc on kc++ with -march=ia64.  This also speeds up live
var from 1.66s -> 0.87s as well, reducing total llc time from 20.1s->15.2s.

This also speeds up other targets slightly, e.g. llc time on X86 from 16.84
-> 16.45s, and PPC from 17.64->17.03s.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22990 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 23:40:41 +00:00
Chris Lattner
ab4b66d4c2 Simplify this code by using higher-level LiveVariables methods
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22989 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 22:51:41 +00:00
Chris Lattner
76eb08bcda Simplify this code by using LiveVariables::KillsRegister
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22988 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 22:49:55 +00:00
Chris Lattner
20647a55bf Add RegisterDefIsDead to correspond to KillsRegister, mark both const
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22987 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 22:43:24 +00:00
Chris Lattner
b980578b41 Keep track of which registers are related to which other registers.
Use this information to avoid doing expensive interval intersections for
registers that could not possible be interesting.  This speeds up linscan
on ia64 compiling kc++ in release mode from taking 7.82s to 4.8s(!), total
itanium llc time on this program is 27.3s now.  This marginally speeds up
PPC and X86, but they appear to be limited by other parts of linscan, not
this code.

On this program, on itanium, live intervals now takes 41% of llc time.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22986 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 22:27:31 +00:00
Chris Lattner
ef35ba4f64 add a method
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22985 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 21:45:31 +00:00
Chris Lattner
57da25231a Fix PR618 and Regression/CodeGen/CBackend/2005-08-23-Fmod.ll by not emitting
x%y for 'rem' on fp values.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22984 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 20:22:50 +00:00
Chris Lattner
c544cece40 New testcase for PR618
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22983 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 20:22:07 +00:00
Chris Lattner
424dcbdb3b add a note
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22982 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 06:27:59 +00:00
Nate Begeman
8ca5693c51 Ack, typo
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22981 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 05:45:10 +00:00
Nate Begeman
a50d53e5ef Add an option to make SetCC illegal as a beta option
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22979 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 05:42:36 +00:00
Nate Begeman
7042f15bde Teach the SelectionDAG how to transform select_cc eq, X, 0, 1, 0 into
either seteq X, 0 or srl (ctlz X), size(X-1), depending on what's legal
for the target.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22978 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 05:41:12 +00:00
Nate Begeman
b942a3dd5f Teach Legalize how to turn setcc into select_cc
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22977 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 04:29:48 +00:00
Nate Begeman
354df0ae6d Remove some instructions we no longer generate
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22976 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-23 01:16:46 +00:00
Chris Lattner
1f24df6324 Remove some regs that are not used.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22975 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 22:32:13 +00:00
Chris Lattner
9dc4d3cbac Nate noticed that 30% of the malloc/frees in llc come from calls to LowercaseString
in the asmprinter.  This changes the .td files to use lower case register names,
avoiding the need to do this call.  This speeds up the asmprinter from 1.52s
to 1.06s on kc++ in a release build.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22974 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 22:00:02 +00:00
Chris Lattner
0a00bece0d Fix a crash I introduced into the IA64 backend with my copyfromreg change.
It used to crash on any function that took float arguments.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22973 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 21:33:11 +00:00
Chris Lattner
a411cbca5c Try to avoid scanning the fixed list. On architectures with a non-stupid
number of regs (e.g. most riscs), many functions won't need to use callee
clobbered registers.  Do a speculative check to see if we can get a free
register without processing the fixed list (which has all of these).  This
saves a lot of time on machines with lots of callee clobbered regs (e.g.
ppc and itanium, also x86).

This reduces ppc llc compile time from 184s -> 172s on kc++.  This is probably
worth FAR FAR more on itanium though.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22972 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 20:59:30 +00:00
Chris Lattner
a6c1750362 Move some code in the register assignment case that only needs to happen if
we spill out of the fast path.  The scan of active_ and the calls to
updateSpillWeights don't need to happen unless a spill occurs.  This reduces
debug llc time of kc++ with ppc from 187.3s to 183.2s.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22971 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 20:20:42 +00:00
Chris Lattner
47c0889419 Add a pass name for -time-passes output
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22970 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 18:28:09 +00:00
Chris Lattner
8f03405ee5 Fix a problem where constant expr shifts would not have their shift amount
promoted to the right type.  This fixes: IA64/2005-08-22-LegalizerCrash.ll


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22969 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 17:28:31 +00:00
Chris Lattner
4c23c5b330 Testcase for a crash in the legalizer on ia64. This is reduced from kc++.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22968 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 17:16:49 +00:00
Chris Lattner
ea966d714f Add a long-overdue itanium regression test dir: hint hint Duraid :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22967 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 17:15:41 +00:00
Chris Lattner
f8355d9846 Speed up this loop a bit, based on some observations that Nate made, and
add some comments.  This loop really needs to be reevaluated!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22966 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 16:55:22 +00:00
Chris Lattner
ce5fcc30ae Revert my patch which changed the code to not work.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22965 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 16:24:25 +00:00
Chris Lattner
730e15eb74 Make the example a bit easier to understand, suggested by Jim.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22964 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 16:11:46 +00:00
Chris Lattner
f7f2255505 Implement stores.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22963 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 01:27:59 +00:00
Chris Lattner
23553cfb4a Add a fast-path for register values. Add support for constant pool entries,
allowing us to compile this:

float %test2(float* %P) {
        %Q = load float* %P
        %R = add float %Q, 10.1
        ret float %R
}

to this:

_test2:
        lfs r2, 0(r3)
        lis r3, ha16(.CPI_test2_0)
        lfs r3, lo16(.CPI_test2_0)(r3)
        fadds f1, r2, r3
        blr


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22962 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 01:04:32 +00:00
Chris Lattner
ddf3e7dfd7 Fix compilation of:
float %test2(float* %P) {
        %Q = load float* %P
        %R = add float %Q, %Q
        ret float %R
}

By returning the right result.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22961 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 00:59:14 +00:00
Chris Lattner
b30ee6a41a Make sure expressions only have one use before emitting them into a place that is conditionally executed
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22960 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-22 00:47:28 +00:00
Chris Lattner
9944b76cfe Implement most of load support. There is still a bug though.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22959 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 22:31:09 +00:00
Chris Lattner
4fbde4d871 add a method
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22958 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 22:30:42 +00:00
Chris Lattner
c975e1dedc add anew method
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22957 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 22:30:30 +00:00
Chris Lattner
81e72b13d7 Add support for frame index nodes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22956 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 19:56:04 +00:00
Chris Lattner
99baddaf57 add a method
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22955 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 19:48:59 +00:00
Chris Lattner
c3aea74ae5 ADd a method
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22954 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 19:48:53 +00:00
Chris Lattner
1db1adbdee Don't print out the MBB label for the entry mbb
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22953 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 19:09:33 +00:00
Chris Lattner
f913d3f91c Simplify the logic for BRTWOWAY_CC handling. The isel code already
simplifies BRTWOWAY into BR if one of the results is a fall-through.
Unless I'm missing something, there is no reason to duplicate this
in the target-specific code.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22952 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 19:03:28 +00:00
Chris Lattner
2fbb4579d6 Implement selection for branches.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22951 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 18:50:37 +00:00
Chris Lattner
d54209fcf9 Add 5-operand version of SelectNodeTo
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22950 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 18:49:58 +00:00
Chris Lattner
6b09a29604 add a method
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22949 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 18:49:33 +00:00
Chris Lattner
f85ab15acf Add support for basic blocks, fix a bug in result # computation
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22948 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 18:49:29 +00:00
Chris Lattner
550b1e59c4 When legalizing brcond ->brcc or select -> selectcc, make sure to truncate
the old condition to a one bit value.  The incoming value must have been
promoted, and the top bits are undefined.  This causes us to generate:

_test:
        rlwinm r2, r3, 0, 31, 31
        li r3, 17
        cmpwi cr0, r2, 0
        bne .LBB_test_2 ;
.LBB_test_1:    ;
        li r3, 1
.LBB_test_2:    ;
        blr

instead of:

_test:
        rlwinm r2, r3, 0, 31, 31
        li r2, 17
        cmpwi cr0, r3, 0
        bne .LBB_test_2 ;
.LBB_test_1:    ;
        li r2, 1
.LBB_test_2:    ;
        or r3, r2, r2
        blr

for:

int %test(bool %c) {
        %retval = select bool %c, int 17, int 1
        ret int %retval
}


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22947 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 18:03:09 +00:00
Chris Lattner
4dd4a2d278 If the false value for a select_cc is really simple (has no inputs), evaluate
it in the block.  This codegens:

int %test(bool %c) {
        %retval = select bool %c, int 17, int 1
        ret int %retval
}

as:

_test:
        rlwinm r2, r3, 0, 31, 31
        li r2, 17
        cmpwi cr0, r3, 0
        bne .LBB_test_2 ;
.LBB_test_1:    ;
        li r2, 1
.LBB_test_2:    ;
        or r3, r2, r2
        blr

instead of:

_test:
        rlwinm r2, r3, 0, 31, 31
        li r2, 17
        li r4, 1
        cmpwi cr0, r3, 0
        bne .LBB_test_2 ;
.LBB_test_1:    ;
        or r2, r4, r4
.LBB_test_2:    ;
        or r3, r2, r2
        blr

... which is one fewer instruction.  The savings are more significant for
global address and constantfp nodes.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22946 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 17:41:11 +00:00
Chris Lattner
2cfd128bec Fix a bug in the RUN line
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22945 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 16:37:36 +00:00
Duraid Madina
7340dd5d2e reenable collapse of loadimm+AND -> dep.z (thanks guys)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22944 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-21 15:43:53 +00:00
Chris Lattner
fe0c2c8721 fix bogus warning
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22943 91177308-0d34-0410-b5e6-96231b3b80d8
2005-08-20 18:07:27 +00:00