regs. This is the only change in this checkin that may affects the
default scheduler. With better register tracking and heuristics, it
doesn't make sense to artificially lower the register limit so much.
Added -sched-high-latency-cycles and X86InstrInfo::isHighLatencyDef to
give the scheduler a way to account for div and sqrt on targets that
don't have an itinerary. It is currently defaults to 10 (the actual
number doesn't matter much), but only takes effect on non-default
schedulers: list-hybrid and list-ilp.
Added several heuristics that can be individually disabled for the
non-default sched=list-ilp mode. This helps us determine how much
better we can do on a given benchmark than the default
scheduler. Certain compute intensive loops run much faster in this
mode with the right set of heuristics, and it doesn't seem to have
much negative impact elsewhere. Not all of the heuristics are needed,
but we still need to experiment to decide which should be disabled by
default for sched=list-ilp.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@127067 91177308-0d34-0410-b5e6-96231b3b80d8
"long latency" enough to hoist even if it may increase spilling. Reloading
a value from spill slot is often cheaper than performing an expensive
computation in the loop. For X86, that means machine LICM will hoist
SQRT, DIV, etc. ARM will be somewhat aggressive with VFP and NEON
instructions.
- Enable register pressure aware machine LICM by default.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@116781 91177308-0d34-0410-b5e6-96231b3b80d8
else in X86), and add support for pavgusb. This is apparently the
only instruction (other than movsx) that is preventing ffmpeg from building
with clang.
If someone else is interested in banging out the rest of the 3DNow!
instructions, it should be quite easy now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@115466 91177308-0d34-0410-b5e6-96231b3b80d8
Since mem2reg isn't run at -O0, we get a ton of reloads from the stack,
for example, before, this code:
int foo(int x, int y, int z) {
return x+y+z;
}
used to compile into:
_foo: ## @foo
subq $12, %rsp
movl %edi, 8(%rsp)
movl %esi, 4(%rsp)
movl %edx, (%rsp)
movl 8(%rsp), %edx
movl 4(%rsp), %esi
addl %edx, %esi
movl (%rsp), %edx
addl %esi, %edx
movl %edx, %eax
addq $12, %rsp
ret
Now we produce:
_foo: ## @foo
subq $12, %rsp
movl %edi, 8(%rsp)
movl %esi, 4(%rsp)
movl %edx, (%rsp)
movl 8(%rsp), %edx
addl 4(%rsp), %edx ## Folded load
addl (%rsp), %edx ## Folded load
movl %edx, %eax
addq $12, %rsp
ret
Fewer instructions and less register use = faster compiles.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113102 91177308-0d34-0410-b5e6-96231b3b80d8
rip out the implementation of X86InstrInfo::GetInstSizeInBytes.
The code being ripped out just implemented a copy and hacked up
version of the (old) instruction encoder, and is buggy and
terrible in other ways. Since "GetInstSizeInBytes" is really
only there to support the JIT's "NeedsExactSize" hook (which
noone is using), just rip out the code. I will rip out the
NeedsExactSize hook next.
This resolves rdar://7617809 - switch X86InstrInfo::GetInstSizeInBytes to use X86MCCodeEmitter
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109149 91177308-0d34-0410-b5e6-96231b3b80d8
returns the start of the memory operand for an instruction.
Introduce a new "X86AddrSegment" enum to reduce # magic numbers
referring to X86 memory operand layout.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107916 91177308-0d34-0410-b5e6-96231b3b80d8
- Add encode bits for VEX_W
- All 128-bit SSE 1 & SSE2 instructions that are described
in the .td file now have a AVX encoded form already working.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@107365 91177308-0d34-0410-b5e6-96231b3b80d8
addresses a longstanding deficiency noted in many FIXMEs scattered
across all the targets.
This effectively moves the problem up one level, replacing eleven
FIXMEs in the targets with eight FIXMEs in CodeGen, plus one path
through FastISel where we actually supply a DebugLoc, fixing Radar
7421831.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@106243 91177308-0d34-0410-b5e6-96231b3b80d8
In file included from X86InstrInfo.cpp:16:
X86GenInstrInfo.inc:2789: error: integer constant is too large for 'long' type
X86GenInstrInfo.inc:2790: error: integer constant is too large for 'long' type
X86GenInstrInfo.inc:2792: error: integer constant is too large for 'long' type
X86GenInstrInfo.inc:2793: error: integer constant is too large for 'long' type
X86GenInstrInfo.inc:2808: error: integer constant is too large for 'long' type
X86GenInstrInfo.inc:2809: error: integer constant is too large for 'long' type
X86GenInstrInfo.inc:2816: error: integer constant is too large for 'long' type
X86GenInstrInfo.inc:2817: error: integer constant is too large for 'long' type
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@105524 91177308-0d34-0410-b5e6-96231b3b80d8
instruction defines subregisters.
Any existing subreg indices on the original instruction are preserved or
composed with the new subreg index.
Also substitute multiple operands mentioning the original register by using the
new MachineInstr::substituteRegister() function. This is necessary because there
will soon be <imp-def> operands added to non read-modify-write partial
definitions. This instruction:
%reg1234:foo = FLAP %reg1234<imp-def>
will reMaterialize(%reg3333, bar) like this:
%reg3333:bar-foo = FLAP %reg333:bar<imp-def>
Finally, replace the TargetRegisterInfo pointer argument with a reference to
indicate that it cannot be NULL.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@105358 91177308-0d34-0410-b5e6-96231b3b80d8
otherwise labels get incorrectly merged. We handled this by emitting a
".byte 0", but this isn't correct on thumb/arm targets where the text segment
needs to be a multiple of 2/4 bytes. Handle this by emitting a noop. This
is more gross than it should be because arm/ppc are not fully mc'ized yet.
This fixes rdar://7908505
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102400 91177308-0d34-0410-b5e6-96231b3b80d8
SSEDomainFix will collapse to the domain with the lower number when it has a
choice. The SSEPackedSingle domain often has smaller instructions, so prefer
that.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99952 91177308-0d34-0410-b5e6-96231b3b80d8
On Nehalem and newer CPUs there is a 2 cycle latency penalty on using a register
in a different domain than where it was defined. Some instructions have
equvivalents for different domains, like por/orps/orpd.
The SSEDomainFix pass tries to minimize the number of domain crossings by
changing between equvivalent opcodes where possible.
This is a work in progress, in particular the pass doesn't do anything yet. SSE
instructions are tagged with their execution domain in TableGen using the last
two bits of TSFlags. Note that not all instructions are tagged correctly. Life
just isn't that simple.
The SSE execution domain issue is very similar to the ARM NEON/VFP pipeline
issue handled by NEONMoveFixPass. This pass may become target independent to
handle both.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99524 91177308-0d34-0410-b5e6-96231b3b80d8
This is work in progress. So far, SSE execution domain tables are added to
X86InstrInfo, and a skeleton pass is enabled with -sse-domain-fix.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99345 91177308-0d34-0410-b5e6-96231b3b80d8