Commit Graph

9 Commits

Author SHA1 Message Date
Jakob Stoklund Olesen
352d352c02 Always normalize spill weights, also for intervals created by spilling.
Moderate the weight given to very small intervals.

The spill weight given to new intervals created when spilling was not
normalized in the same way as the original spill weights calculated by
CalcSpillWeights. That meant that restored registers would tend to hang around
because they had a much higher spill weight that unspilled registers.

This improves the runtime of a few tests by up to 10%, and there are no
significant regressions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96613 91177308-0d34-0410-b5e6-96231b3b80d8
2010-02-18 21:33:05 +00:00
Evan Cheng
014d624a28 Fix a pasto. Also simplify for Bill's benefit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82505 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-22 01:48:19 +00:00
Evan Cheng
cf985a9545 Clean up spill weight computation. Also some changes to give loop induction
variable increment / decrement slighter high priority. 

This has major impact on some micro-benchmarks. On MultiSource/Applications
and spec tests, it's a minor win. It also reduce 256.bzip instruction count
by 8%, 55 on 164.gzip on i386 / Darwin.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82485 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-21 21:12:25 +00:00
Dan Gohman
36a0947820 Eliminate more uses of llvm-as and llvm-dis.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81290 91177308-0d34-0410-b5e6-96231b3b80d8
2009-09-08 23:54:48 +00:00
Dan Gohman
c8db34cb07 Don't use special heuristics for nodes with no data predecessors
unless they actually have data successors, and likewise for nodes
with no data successors unless they actually have data precessors.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@64327 91177308-0d34-0410-b5e6-96231b3b80d8
2009-02-11 21:29:39 +00:00
Evan Cheng
d498c8f2c8 Teach 2addr pass to be do more commuting. If both uses of a two-address instruction are killed, but the first operand has a use before and after the def, commute if the second operand does not suffer from the same issue.
%reg1028<def> = EXTRACT_SUBREG %reg1027<kill>, 1                                                                                                                                     
%reg1029<def> = MOV8rr %reg1028                                                                                                                                                      
%reg1029<def> = SHR8ri %reg1029, 7, %EFLAGS<imp-def,dead>                                                                                                                            
insert => %reg1030<def> = MOV8rr %reg1028                                                                                                                                            
%reg1030<def> = ADD8rr %reg1028<kill>, %reg1029<kill>, %EFLAGS<imp-def,dead>                                                                                                         

In this case, it might not be possible to coalesce the second MOV8rr                                                                                                                 
instruction if the first one is coalesced. So it would be profitable to                                                                                                              
commute it:                                                                                                                                                                          
%reg1028<def> = EXTRACT_SUBREG %reg1027<kill>, 1                                                                                                                                     
%reg1029<def> = MOV8rr %reg1028                                                                                                                                                      
%reg1029<def> = SHR8ri %reg1029, 7, %EFLAGS<imp-def,dead>                                                                                                                            
insert => %reg1030<def> = MOV8rr %reg1029                                                                                                                                            
%reg1030<def> = ADD8rr %reg1029<kill>, %reg1028<kill>, %EFLAGS<imp-def,dead>


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@62954 91177308-0d34-0410-b5e6-96231b3b80d8
2009-01-25 03:53:59 +00:00
Evan Cheng
4a03775777 For whatever the reason, x86 CallingConv::Fast (i.e. fastcc) was not passing scalar arguments in registers. This patch defines a new fastcc CC which is slightly different from the FastCall CC. In addition to passing integer arguments in ECX and EDX, it also specify doubles are passed in 8-byte slots which are 8-byte aligned (instead of 4-byte aligned). This avoids a potential performance hazard where doubles span cacheline boundaries.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@55807 91177308-0d34-0410-b5e6-96231b3b80d8
2008-09-04 22:59:58 +00:00
Evan Cheng
d36531249a Spiller now remove unused spill slots.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47657 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-27 03:04:06 +00:00
Evan Cheng
249ded3fa8 Rematerialization logic was overly conservative when it comes to loads from fixed stack slots.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47529 91177308-0d34-0410-b5e6-96231b3b80d8
2008-02-23 03:38:34 +00:00