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.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@120960 91177308-0d34-0410-b5e6-96231b3b80d8
"-mattr=+vfp3" is specified. However, this will not work for hardware that
only supports 16 registers. Add a new flag to support -"mattr=+vfp3,+d16".
Patch by Jan Voung!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@116310 91177308-0d34-0410-b5e6-96231b3b80d8
LDM/STM instructions can run one cycle faster on some ARM processors if the
memory address is 64-bit aligned. Radar 8489376.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@115047 91177308-0d34-0410-b5e6-96231b3b80d8
cost modeling for if-conversion. Now if only we had a way to estimate the misprediction probability.
Adjsut CodeGen/ARM/ifcvt10.ll. The pipeline on Cortex-A8 is long enough that it is still profitable
to predicate an ldm, but the shorter pipeline on Cortex-A9 makes it unprofitable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@114995 91177308-0d34-0410-b5e6-96231b3b80d8
llc now recognizes the "intent" to support MC/obj emission for ARM, but
given that they are all stubs, it asserts on --filetype=obj --march=arm
Patch by Jason Kim.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@114856 91177308-0d34-0410-b5e6-96231b3b80d8
take multiple cycles to decode.
For the current if-converter clients (actually only ARM), the instructions that
are predicated on false are not nops. They would still take machine cycles to
decode. Micro-coded instructions such as LDM / STM can potentially take multiple
cycles to decode. If-converter should take treat them as non-micro-coded
simple instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@113570 91177308-0d34-0410-b5e6-96231b3b80d8
memory and synchronization barrier dmb and dsb instructions.
- Change instruction names to something more sensible (matching name of actual
instructions).
- Added tests for memory barrier codegen.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@110785 91177308-0d34-0410-b5e6-96231b3b80d8
Jordy <snhjordy@gmail.com>.
Followup patches will add some tests and adjust to use Subtarget features
for the instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103119 91177308-0d34-0410-b5e6-96231b3b80d8
for all the processors where I have tried it, and even when it might not help
performance, the cost is quite low. The opportunities for duplicating
indirect branches are limited by other factors so code size does not change
much due to tail duplicating indirect branches aggressively.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@90144 91177308-0d34-0410-b5e6-96231b3b80d8
than doing the same via constpool:
1. Load from constpool costs 3 cycles on A9, movt/movw pair - just 2.
2. Load from constpool might stall up to 300 cycles due to cache miss.
3. Movt/movw does not use load/store unit.
4. Less constpool entries => better compiler performance.
This is only enabled on ELF systems, since darwin does not have needed
relocations (yet).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89720 91177308-0d34-0410-b5e6-96231b3b80d8
contents of the block to be duplicated. Use this for ARM Cortex A8/9 to
be more aggressive tail duplicating indirect branches, since it makes it
much more likely that they will be predicted in the branch target buffer.
Testcase coming soon.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89187 91177308-0d34-0410-b5e6-96231b3b80d8
Module*.
Also, dropped uses of TargetMachine where unnecessary. The only target which
still takes a TargetMachine& is Mips, I would appreciate it if someone would
normalize this to match other targets.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@77918 91177308-0d34-0410-b5e6-96231b3b80d8
