Commit Graph

12908 Commits

Author SHA1 Message Date
Evan Cheng
d51425a82d Use movaps / movapd (instead of movss / movsd) to do FR32 / FR64 reg to reg
transfer.

According to the Intel P4 Optimization Manual:

Moves that write a portion of a register can introduce unwanted
dependences. The movsd reg, reg instruction writes only the bottom
64 bits of a register, not to all 128 bits. This introduces a dependence on
the preceding instruction that produces the upper 64 bits (even if those
bits are not longer wanted). The dependence inhibits register renaming,
and thereby reduces parallelism.

Not to mention movaps is shorter than movss.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26226 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-16 01:50:02 +00:00
Chris Lattner
f17c42d409 fix a bug where we unswitched the wrong way
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26225 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-16 01:24:41 +00:00
Evan Cheng
18a8452f3d A bit more memset / memcpy optimization.
Turns them into calls to memset / memcpy if 1) buffer(s) are not DWORD aligned,
2) size is not known to be greater or equal to some minimum value (currently 128).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26224 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-16 00:21:07 +00:00
Chris Lattner
a48654ef23 Implement trivial unswitching for switch stmts. This allows us to trivial
unswitch this loop on 2 before sweating to unswitch on 1/3.

void test4(int N, int i, int C, int*P, int*Q) {
  int j;
  for (j = 0; j < N; ++j) {
    switch (C) {                // general unswitching.
    default: P[i+j] = 0; break;
    case 1: Q[i+j] = 0; break;
    case 3: P[i+j] = Q[i+j]; break;
    case 2: break;              //  TRIVIAL UNSWITCH on C==2
    }
  }
}


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26223 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 22:52:05 +00:00
Evan Cheng
fe1c31204e Remove an entry.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26222 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 22:14:34 +00:00
Evan Cheng
a47876d87a Remove an unused function parameter.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26221 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 22:12:35 +00:00
Chris Lattner
4e1323969c make "trivial" unswitching significantly more general. It can now handle
this for example:

  for (j = 0; j < N; ++j) {     // trivial unswitch
    if (C)
      P[i+j] = 0;
  }

turning it into the obvious code without bothering to duplicate an empty loop.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26220 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 22:03:36 +00:00
Evan Cheng
74d0aa9a4b Turn a memcpy from string constant into a series of stores of constant values.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26219 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 21:59:04 +00:00
Andrew Lenharth
2f9859486a fix a bunch of alpha regressions. see bug 709
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26218 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 21:13:37 +00:00
Chris Lattner
5946fef1db new test
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26217 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 19:52:06 +00:00
Jim Laskey
d0e58e36a9 Should not combine ISD::LOCATIONs until we have scheme to remove from
MachineDebugInfo tables.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26216 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 19:34:44 +00:00
Chris Lattner
3fdde110eb Checking the wrong value. This caused us to emit silly code like
Y = seteq bool X, true
instead of just using X :)


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26215 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 19:05:52 +00:00
Chris Lattner
58af2a1c5f Convert this over to work with the new makefiles
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26206 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 07:22:58 +00:00
Chris Lattner
45090476be Sparc actually *DOES* have a directive for emitting zeros. In fact, it requires
it, because this:

.bss
X:
.byte 0

results in the assembler warning: "initialization in bss segment".  Annoying.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26204 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 07:07:14 +00:00
Chris Lattner
2fecc0f9e2 random lexer change to test the makefile updating stuff
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26203 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 07:02:59 +00:00
Chris Lattner
6fa1f57c66 Fix SingleSource/Regression/C/2004-08-12-InlinerAndAllocas.c on Sparc.
The ABI specifies that there is a register save area at the bottom of the
stack, which means the actual used pointer needs to be an offset from
the subtracted value.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26202 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 06:41:34 +00:00
Evan Cheng
2354f5a46b Remove an entry.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26197 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 01:56:48 +00:00
Evan Cheng
17ef92eda7 Use .zerofill on x86/darwin.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26196 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 01:56:23 +00:00
Evan Cheng
c080d6fb3d Lower memcpy with small constant size operand into a series of load / store
ops.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26195 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 01:54:51 +00:00
Chris Lattner
6d9d13d2e4 more refactoring, no functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26194 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 01:44:42 +00:00
Evan Cheng
f3f0a9c587 cvtsd2ss / cvtss2sd encoding bug.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26193 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 00:31:03 +00:00
Evan Cheng
7335f9beea movaps, movapd encoding bug.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26192 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 00:11:37 +00:00
Chris Lattner
fed5d9dbd3 pull some code out into a function
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26191 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-15 00:07:43 +00:00
Chris Lattner
0ab9f966de Canonicalize inner loops before outer loops. Inner loop canonicalization
can provide work for the outer loop to canonicalize.

This fixes a case that breaks unswitching.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26189 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 23:06:02 +00:00
Evan Cheng
dea7245997 Doh again!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26188 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 23:05:54 +00:00
Chris Lattner
c27e056d4f When splitting exit edges to canonicalize loops, make sure to put the new
block in the appropriate loop nest.

Third time is the charm, right?


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26187 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 22:34:08 +00:00
Chris Lattner
ae302ab9cb new note
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26186 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 22:19:54 +00:00
Chris Lattner
0d7db6f79a If we have zero initialized data with external linkage, use .zerofill to
emit it (instead of .space), saving a bit of space in the .o file.

For example:
int foo[100];
int bar[100] = {};

when compiled with C++ or -fno-common results in shrinkage from 1160 to 360
bytes of space.  The X86 backend can also do this on darwin.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26185 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 22:18:23 +00:00
Jim Laskey
ae9a84b9be Using wrong DW_FORM.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26184 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 22:01:57 +00:00
Evan Cheng
ee50a1ae27 Don't special case XS, XD prefixes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26183 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 21:52:51 +00:00
Evan Cheng
576c141921 Bug fix: XS, XD prefixes were being emitted twice.
XMM registers were not being handled.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26182 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 21:45:24 +00:00
Chris Lattner
33d5082779 Make sure that weak functions are aligned properly
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26181 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 20:42:33 +00:00
Evan Cheng
ddf7532442 Duh
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26180 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 20:37:37 +00:00
Evan Cheng
968c178172 Remove -disable-x86-sse
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26179 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 20:30:14 +00:00
Chris Lattner
60330ff193 add an assert
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26178 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 20:14:17 +00:00
Evan Cheng
c4f8eee054 Keep to < 80 cols
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26177 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 20:12:38 +00:00
Evan Cheng
ac940ab1bf Missed a break so memcpy cases fell through to memset. Doh.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26176 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 19:45:56 +00:00
Evan Cheng
80e89d7d6c Fixed a build breakage.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26175 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 09:11:59 +00:00
Evan Cheng
a03a5dc7ce Rename maxStoresPerMemSet to maxStoresPerMemset, etc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26174 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 08:38:30 +00:00
Evan Cheng
952b7d6a1e Add a entry.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26173 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 08:25:32 +00:00
Evan Cheng
87ed716d21 Set maxStoresPerMemSet to 16. Ditto for maxStoresPerMemCpy and
maxStoresPerMemMove. Although the last one is not used.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26172 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 08:25:08 +00:00
Evan Cheng
1db92f947c Expand memset dst, c, size to a series of stores if size falls below the
target specific theshold, e.g. 16 for x86.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26171 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 08:22:34 +00:00
Evan Cheng
82eaf628b4 Enable SSE (for the right subtargets)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26169 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 08:07:58 +00:00
Chris Lattner
06a248c9b3 now that libcalls don't suck, we can remove this hack
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26164 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 05:39:35 +00:00
Chris Lattner
32eecb09a2 adjust to new style of handling lexer changes
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26162 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 05:14:46 +00:00
Chris Lattner
3dd4c402de Use statistics to keep track of what flavors of loops we are unswitching
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26157 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 01:01:41 +00:00
Chris Lattner
4b653a0405 Fix a latent bug in the call sequence handling stuff. Some targets (e.g. x86)
create these nodes with flag results.  Remember that we legalized them.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26156 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-14 00:55:02 +00:00
Chris Lattner
2082ad9b41 Implement Instcombine/and.ll:test34
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26155 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-13 23:07:23 +00:00
Chris Lattner
f345fe4d9a If any of the sign extended bits are demanded, the input sign bit is demanded
for a sign extension.

This fixes InstCombine/2006-02-13-DemandedMiscompile.ll and Ptrdist/bc.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26152 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-13 22:41:07 +00:00
Chris Lattner
f9a05329a1 Another hack due to allowing multiple symbols with the same name.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@26150 91177308-0d34-0410-b5e6-96231b3b80d8
2006-02-13 22:22:42 +00:00