This patch introduces generating VP intrinsics in the Loop Vectorizer.
Currently the Loop Vectorizer supports vector predication in a very
limited capacity via tail-folding and masked load/store/gather/scatter
intrinsics. However, this does not let architectures with active vector
length predication support take advantage of their capabilities.
Architectures with general masked predication support also can only take
advantage of predication on memory operations. By having a way for the
Loop Vectorizer to generate Vector Predication intrinsics, which (will)
provide a target-independent way to model predicated vector
instructions. These architectures can make better use of their
predication capabilities.
Our first approach (implemented in this patch) builds on top of the
existing tail-folding mechanism in the LV (just adds a new tail-folding
mode using EVL), but instead of generating masked intrinsics for memory
operations it generates VP intrinsics for loads/stores instructions. The
patch adds a new VPlanTransforms to replace the wide header predicate
compare with EVL and updates codegen for load/stores to use VP
store/load with EVL.
Other important part of this approach is how the Explicit Vector Length
is computed. (VP intrinsics define this vector length parameter as
Explicit Vector Length (EVL)). We use an experimental intrinsic
`get_vector_length`, that can be lowered to architecture specific
instruction(s) to compute EVL.
Also, added a new recipe to emit instructions for computing EVL. Using
VPlan in this way will eventually help build and compare VPlans
corresponding to different strategies and alternatives.
Differential Revision: https://reviews.llvm.org/D99750
After some discussion and experimentation, we have seen that changing the default number of vector register bits to LMUL=2 strikes a sweet spot.
Whilst we could be clever here and make the vectorizer smarter about dynamically selecting an LMUL that
a) Doesn't affect register pressure
b) Suitable for the microarchitecture
we would need to teach its heuristics about RISC-V register grouping specifics.
Instead this just does the easy, pragmatic thing by changing the default to a safe value that doesn't affect register pressure signifcantly[1], but should increase throughput and unlock more interleaving.
[1] Register spilling when compiling sqlite at various levels of `-riscv-v-register-bit-width-lmul`:
LMUL=1 2573 spills
LMUL=2 2583 spills
LMUL=4 2819 spills
LMUL=8 3256 spills
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D143723
It's less clear with scalable vectors than fixed length vectors that
interleaving exposes more ILP, as scalable vectors can be thought of a
sort of hardware form of interleaving, especially with larger LMULs.
This also addresses the unexpected additional unrolling that occurs when
using larger LMULs in the loop vectorizer.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D144485
Fixed issue where 'ConstantInt::get(IndextTy, -Part)' was executed with the wrong type for Part,
e.g. IndexTy was i64, but Part was 'unsigned', which led to things like 'mul i64 .., 4294967292',
which was obviously wrong.
Also changed sve-vector-reverse.ll to be vectorized with UF>1 to test this.
This reverts commit 1f01cdda68614dba12af3cc3aff38541d0abcc6b.
This is specifically relevant for loops that vectorize using a scalable VF,
where the code results in:
%vscale = call i32 llvm.vscale.i32()
%vf.part1 = mul i32 %vscale, 4
%gep = getelementptr ..., i32 %vf.part1
Which InstCombine then changes into:
%vscale = call i32 llvm.vscale.i32()
%vf.part1 = mul i32 %vscale, 4
%vf.part1.zext = sext i32 %vf.part1 to i64
%gep = getelementptr ..., i32 %vf.part1.zext
D143016 tried to remove these extends, but that only works when
the call to llvm.vscale.i32() has a single use. After doing any
kind of CSE on these calls the combine no longer kicks in.
It seems more sensible to ask DataLayout what type to use, rather
than relying on InstCombine to insert the extend and hoping it can
fold it away.
I've only changed this for indices that are not constant, because
I vaguely remember there was a reason for sticking with i32. It
would also mean patching up loads more tests.
Reviewed By: paulwalker-arm
Differential Revision: https://reviews.llvm.org/D143267
