Vector Predication Roadmap

Motivation

This proposal defines a roadmap towards native vector predication in LLVM, specifically for vector instructions with a mask and/or an explicit vector length. LLVM currently has no target-independent means to model predicated vector instructions for modern SIMD ISAs such as AVX512, ARM SVE, the RISC-V V extension and NEC SX-Aurora. Only some predicated vector operations, such as masked loads and stores, are available through intrinsics [MaskedIR].

The Vector Predication (VP) extensions is a concrete RFC and prototype implementation to achieve native vector predication in LLVM. The VP prototype and all related discussions can be found in the VP patch on Phabricator [VPRFC].

Roadmap

1. IR-level VP intrinsics

  • There is a consensus on the semantics/instruction set of VP.

  • VP intrinsics and attributes are available on IR level.

  • TTI has capability flags for VP (supportsVP()?, haveActiveVectorLength()?).

Result: VP usable for IR-level vectorizers (LV, VPlan, RegionVectorizer), potential integration in Clang with builtins.

2. CodeGen support

  • VP intrinsics translate to first-class SDNodes (eg llvm.vp.fdiv.* -> vp_fdiv).

  • VP legalization (legalize explicit vector length to mask (AVX512), legalize VP SDNodes to pre-existing ones (SSE, NEON)).

Result: Backend development based on VP SDNodes.

3. Lift InstSimplify/InstCombine/DAGCombiner to VP

  • Introduce PredicatedInstruction, PredicatedBinaryOperator, .. helper classes that match standard vector IR and VP intrinsics.

  • Add a matcher context to PatternMatch and context-aware IR Builder APIs.

  • Incrementally lift DAGCombiner to work on VP SDNodes as well as on regular vector instructions.

  • Incrementally lift InstCombine/InstSimplify to operate on VP as well as regular IR instructions.

Result: Optimization of VP intrinsics on par with standard vector instructions.

4. Deprecate llvm.masked.* / llvm.experimental.reduce.*

  • Modernize llvm.masked.* / llvm.experimental.reduce* by translating to VP.

  • DCE transitional APIs.

Result: VP has superseded earlier vector intrinsics.

5. Predicated IR Instructions

  • Vector instructions have an optional mask and vector length parameter. These lower to VP SDNodes (from Stage 2).

  • Phase out VP intrinsics, only keeping those that are not equivalent to vectorized scalar instructions (reduce, shuffles, ..)

  • InstCombine/InstSimplify expect predication in regular Instructions (Stage (3) has laid the groundwork).

Result: Native vector predication in IR.

References

MaskedIR

llvm.masked.* intrinsics, https://llvm.org/docs/LangRef.html#masked-vector-load-and-store-intrinsics

VPRFC

RFC: Prototype & Roadmap for vector predication in LLVM, https://reviews.llvm.org/D57504