88d31b3f0c
Add a note about a missed cmov -> sbb opportunity.
...
llvm-svn: 153741
2012-03-30 13:02:58 +00:00
f210619d08
Fix typo in the X86 backend readme. Patch from Jaeden Amero.
...
llvm-svn: 147739
2012-01-07 20:35:21 +00:00
767bbe48c1
Chandler fixed this.
...
llvm-svn: 147247
2011-12-24 11:23:32 +00:00
de7ab801cc
Add a natural stack alignment field to TargetData, and prevent InstCombine from
...
promoting allocas to preferred alignments that exceed the natural
alignment. This avoids some potentially expensive dynamic stack realignments.
The natural stack alignment is set in target data strings via the "S<size>"
option. Size is in bits and must be a multiple of 8. The natural stack alignment
defaults to "unspecified" (represented by a zero value), and the "unspecified"
value does not prevent any alignment promotions. Target maintainers that care
about avoiding promotions should explicitly add the "S<size>" option to their
target data strings.
llvm-svn: 141599
2011-10-10 23:42:08 +00:00
42c0330a79
X86: Add patterns for the movbe instruction (mov + bswap, only available on atom)
...
llvm-svn: 141563
2011-10-10 18:34:56 +00:00
124ac2b997
Add a neat little two's complement hack for x86.
...
On x86 we can't encode an immediate LHS of a sub directly. If the RHS comes from a XOR with a constant we can
fold the negation into the xor and add one to the immediate of the sub. Then we can turn the sub into an add,
which can be commuted and encoded efficiently.
This code is generated for __builtin_clz and friends.
llvm-svn: 136167
2011-07-26 22:42:13 +00:00
c956033947
Add a note about efficient codegen for binary log.
...
llvm-svn: 135996
2011-07-25 22:30:00 +00:00
1e81f57bf0
add a note
...
llvm-svn: 131455
2011-05-17 07:22:33 +00:00
6708499b6d
This is done.
...
llvm-svn: 130499
2011-04-29 14:09:57 +00:00
2a75c72e1c
move PR9803 to this readme.
...
llvm-svn: 130385
2011-04-28 05:33:16 +00:00
55f7bf3289
Remove working entry from README.
...
llvm-svn: 129654
2011-04-17 02:36:27 +00:00
0ab5e2cded
Fix a ton of comment typos found by codespell. Patch by
...
Luis Felipe Strano Moraes!
llvm-svn: 129558
2011-04-15 05:18:47 +00:00
6f195469b1
move PR9661 out to here.
...
llvm-svn: 129527
2011-04-14 18:47:18 +00:00
b4dd95b4f9
Add another case we are not optimizing.
...
llvm-svn: 129012
2011-04-06 17:35:32 +00:00
7a3b244d45
The original issue has been fixed by not doing unnecessary sign extensions.
...
Change the test to force a sign extension and expose the problem again.
llvm-svn: 129011
2011-04-06 17:19:35 +00:00
e8f2be0c10
A couple new README entries.
...
llvm-svn: 127786
2011-03-17 01:22:09 +00:00
5237febf0c
a serious "compare CSE" issue that is nontrivial to get right,
...
but which is responsible for us doing really bad things to 256.bzip2.
llvm-svn: 126126
2011-02-21 17:03:47 +00:00
39314bdbc8
A lo/hi mul has higher latency than an imul r,ri, e.g. 5 cycles compared to 3
...
on Core 2 and Nehalem, so the code we generate is better than GCC's here.
llvm-svn: 126100
2011-02-21 01:29:32 +00:00
8731d0cc83
The signed version of our "magic number" computation for the integer approximation
...
of a constant had a minor typo introduced when copying it from the book, which
caused it to favor negative approximations over positive approximations in many
cases. Positive approximations require fewer operations beyond the multiplication.
In the case of division by 3, we still generate code that is a single instruction
larger than GCC's code.
llvm-svn: 126097
2011-02-21 00:22:02 +00:00
78b9851a3a
Minor x86 README updates.
...
llvm-svn: 126054
2011-02-19 21:54:28 +00:00
0281731cc2
add a poor division by constant case.
...
llvm-svn: 125832
2011-02-18 05:35:49 +00:00
51415d26f1
update a bunch of entries.
...
llvm-svn: 122700
2011-01-02 18:31:38 +00:00
f9e0a56b94
fix some sort of weird pasto
...
llvm-svn: 122560
2010-12-26 12:05:11 +00:00
424de3498b
add a note
...
llvm-svn: 122559
2010-12-26 03:53:31 +00:00
b37ae33125
Remove some obsolete README items, add a new one off the top of my head.
...
llvm-svn: 122495
2010-12-23 15:07:02 +00:00
6020ed9d99
X86: Lower a select directly to a setcc_carry if possible.
...
int test(unsigned long a, unsigned long b) { return -(a < b); }
compiles to
_test: ## @test
cmpq %rsi, %rdi ## encoding: [0x48,0x39,0xf7]
sbbl %eax, %eax ## encoding: [0x19,0xc0]
ret ## encoding: [0xc3]
instead of
_test: ## @test
xorl %ecx, %ecx ## encoding: [0x31,0xc9]
cmpq %rsi, %rdi ## encoding: [0x48,0x39,0xf7]
movl $-1, %eax ## encoding: [0xb8,0xff,0xff,0xff,0xff]
cmovael %ecx, %eax ## encoding: [0x0f,0x43,0xc1]
ret ## encoding: [0xc3]
llvm-svn: 122451
2010-12-22 23:09:28 +00:00
c8b035d006
Factor the (x & 2^n) ? 2^m : 0 instcombine into its own method and generalize it
...
to catch cases where n != m with a shift.
llvm-svn: 121608
2010-12-11 09:42:59 +00:00
d4b0873c06
Enable sibling call optimization of libcalls which are expanded during
...
legalization time. Since at legalization time there is no mapping from
SDNode back to the corresponding LLVM instruction and the return
SDNode is target specific, this requires a target hook to check for
eligibility. Only x86 and ARM support this form of sibcall optimization
right now.
rdar://8707777
llvm-svn: 120501
2010-11-30 23:55:39 +00:00
5d2262dc76
apparently tailcalls are better on darwin/x86-64 than on linux?
...
llvm-svn: 119947
2010-11-21 18:59:20 +00:00
3c9b5f394b
Don't narrow the load and store in a load+twiddle+store sequence unless
...
there are clearly no stores between the load and the store. This fixes
this miscompile reported as PR7833.
This breaks the test/CodeGen/X86/narrow_op-2.ll optimization, which is
safe, but awkward to prove safe. Move it to X86's README.txt.
llvm-svn: 112861
2010-09-02 21:18:42 +00:00
f75de6eae7
A couple of small missed optimizations.
...
llvm-svn: 112411
2010-08-29 05:07:40 +00:00
9a77382685
Custom lower the memory barrier instructions and add support
...
for lowering without sse2. Add a couple of new testcases.
Fixes a few libgomp tests and latent bugs. Remove a few todos.
llvm-svn: 109078
2010-07-22 02:48:34 +00:00
1f41303260
Remove a fixed item, update a couple partially-fixed items.
...
llvm-svn: 105375
2010-06-03 01:01:48 +00:00
312d604ee2
This is now done.
...
llvm-svn: 97450
2010-03-01 17:43:57 +00:00
4d4c6944e9
A few missed optimizations; the last one could have a significant impact on
...
code with lots of bitfields.
llvm-svn: 95809
2010-02-10 21:26:04 +00:00
3eb76c23dd
this is an SSE-specific issue.
...
llvm-svn: 93373
2010-01-13 23:29:11 +00:00
fb534d97b5
X86 if conversion + tail merging issues from PR6032.
...
llvm-svn: 93372
2010-01-13 23:28:40 +00:00
5d1987f9a0
Remove some README.txt entries which are now implemented.
...
llvm-svn: 92511
2010-01-04 20:55:05 +00:00
dbe2aa91b9
A couple minor README updates.
...
llvm-svn: 91823
2009-12-21 08:03:16 +00:00
fd2730ee8c
Move and clarify note.
...
llvm-svn: 85334
2009-10-27 22:48:31 +00:00
2974f63cb5
Note corrected.
...
llvm-svn: 85332
2009-10-27 22:43:24 +00:00
cd4d148040
Modify note.
...
llvm-svn: 85331
2009-10-27 22:40:45 +00:00
a205402c16
Add a note.
...
llvm-svn: 85329
2009-10-27 22:34:43 +00:00
92df9c3323
Add a note.
...
llvm-svn: 77584
2009-07-30 08:56:19 +00:00
b127c0684f
remove a bogus note.
...
llvm-svn: 74509
2009-06-30 05:22:31 +00:00
5ed255e642
add a note
...
llvm-svn: 74508
2009-06-30 04:20:46 +00:00
0101f45785
another xform that is target-independent (should be done in instcombine).
...
llvm-svn: 73472
2009-06-16 06:15:56 +00:00
aba55a69b1
I think instcombine should unconditionally do this xform.
...
llvm-svn: 73471
2009-06-16 06:11:35 +00:00
32ad5e9c08
Misc x86 README updates: remove a couple of already-fixed issues,
...
add a few suggestions from looking at some assembly code.
llvm-svn: 73210
2009-06-11 23:07:04 +00:00
9b65031add
add some late optimizations that GCC does. It thinks these are a win
...
even on Core2, not just AMD processors which was a surprise to me.
llvm-svn: 72396
2009-05-25 20:28:19 +00:00
Benjamin Kramer