difficult on current ARM implementations for a few reasons.
1. Even though a single vmla has latency that is one cycle shorter than a pair
of vmul + vadd, a RAW hazard during the first (4? on Cortex-a8) can cause
additional pipeline stall. So it's frequently better to single codegen
vmul + vadd.
2. A vmla folowed by a vmul, vmadd, or vsub causes the second fp instruction to
stall for 4 cycles. We need to schedule them apart.
3. A vmla followed vmla is a special case. Obvious issuing back to back RAW
vmla + vmla is very bad. But this isn't ideal either:
vmul
vadd
vmla
Instead, we want to expand the second vmla:
vmla
vmul
vadd
Even with the 4 cycle vmul stall, the second sequence is still 2 cycles
faster.
Up to now, isel simply avoid codegen'ing fp vmla / vmls. This works well enough
but it isn't the optimial solution. This patch attempts to make it possible to
use vmla / vmls in cases where it is profitable.
A. Add missing isel predicates which cause vmla to be codegen'ed.
B. Make sure the fmul in (fadd (fmul)) has a single use. We don't want to
compute a fmul and a fmla.
C. Add additional isel checks for vmla, avoid cases where vmla is feeding into
fp instructions (except for the #3 exceptional case).
D. Add ARM hazard recognizer to model the vmla / vmls hazards.
E. Add a special pre-regalloc case to expand vmla / vmls when it's likely the
vmla / vmls will trigger one of the special hazards.
Work in progress, only A+B are enabled.
llvm-svn: 120960
The only reasonable way I could find to do this is to provide an alternate
version of the addrmode6 operand with a different encoding function. Use it
for all the VLD-dup instructions for the sake of consistency.
llvm-svn: 120358
The op11_8 field is the same for all of them so put it in the instruction
classes instead of specifying it separately for each instruction.
llvm-svn: 120234
'db', 'ib', 'da') instead of having that mode as a separate field in the
instruction. It's more convenient for the asm parser and much more readable for
humans.
<rdar://problem/8654088>
llvm-svn: 119310
all of the different element sizes are pseudo instructions that map down to vext.8 underneath, with
the immediate shifted left to reflect the increased element size.
llvm-svn: 118183
half of the Q register), rather than with just regno. This allows us to unify the encodings for
a lot of different NEON instrucitons that differ only in whether they have Q or D register operands.
llvm-svn: 117056
1. Cortex-A8 load / store multiplies can only issue on ALU0.
2. Eliminate A8_Issue, A8_LSPipe will correctly limit the load / store issues.
3. Correctly model all vld1 and vld2 variants.
llvm-svn: 116134
"The register specified for a dregpair is the corresponding Q register, so to
get the pair, we need to look up the sub-regs based on the qreg. Create a
lookup function since we don't have access to TargetRegisterInfo here to
be able to use getSubReg(ARM::dsub_[01])."
Additionaly, fix the NEON VLD1* and VST1* instruction patterns not to use
the dregpair modifier for the 2xdreg versions. Explicitly specifying the two
registers as operands is more correct and more consistent with the other
instruction patterns. This enables further cleanup of special case code in the
disassembler as a nice side-effect.
llvm-svn: 113903
an argument, so that we can distinguish instructions with the same register
classes but different numbers of registers (e.g., vld3 and vld4). Fix some
of the non-pseudo NEON ld/st instruction itineraries to reflect the number
of registers loaded or stored, not just the opcode name.
llvm-svn: 113854
instructions prior to regalloc. Since it's getting a little close to
the 2.8 branch deadline, I'll have to leave the rest of the instructions
handled by the NEONPreAllocPass for now, but I didn't want to leave half
of the VLD instructions converted and the other half not.
llvm-svn: 112983
vabd intrinsic and add and/or zext operations. In the case of vaba, this
also avoids the need for a DAG combine pattern to combine vabd with add.
Update tests. Auto-upgrade the old intrinsics.
llvm-svn: 112941
add, and subtract operations with zero-extended or sign-extended vectors.
Update tests. Add auto-upgrade support for the old intrinsics.
llvm-svn: 112773
all the other LDM/STM instructions. This fixes asm printer crashes when
compiling with -O0. I've changed one of the NEON tests (vst3.ll) to run
with -O0 to check this in the future.
Prior to this change VLDM/VSTM used addressing mode #5, but not really.
The offset field was used to hold a count of the number of registers being
loaded or stored, and the AM5 opcode field was expanded to specify the IA
or DB mode, instead of the standard ADD/SUB specifier. Much of the backend
was not aware of these special cases. The crashes occured when rewriting
a frameindex caused the AM5 offset field to be changed so that it did not
have a valid submode. I don't know exactly what changed to expose this now.
Maybe we've never done much with -O0 and NEON. Regardless, there's no longer
any reason to keep a count of the VLDM/VSTM registers, so we can use
addressing mode #4 and clean things up in a lot of places.
llvm-svn: 112322
Bob Wilson