ContextQMD
Back to Arangodb

Potential new ops for Highway

3rdParty/highway/g3doc/op_wishlist.md

3.12.9.14.1 KB
Original Source

Potential new ops for Highway

<!--* # Document freshness: For more information, see go/fresh-source. freshness: { owner: 'janwas' reviewed: '2023-04-21' } *-->

[TOC]

Wishlist

Remaining math functions for hwy/contrib/math

atan2, cbrt, cosh, erf, exp2, fmod, hypot, ilogb, lgamma, logb, modf, nextafter, nexttoward, pow, scalbn, tan, tgamma

Remaining STL functions for hwy/contrib/algo

Range coder

Port https://github.com/richgel999/sserangecoding to Highway (~50 instructions).

float64 support for WASM

AfterN

= Not(FirstN()), replaces several instances. WHILEGE on SVE.

52x52=104-bit multiply

For crypto. Native on Icelake+.

Slide1Up/Down

Potentially useful for comparing neighbors e.g. for RLE.

RVV codegen

  • New tuple interface for segment load/store
  • Use new mask<->vec cast instruction, possibly for OddEven, ExpandLoad
  • rgather_vx for broadcasting redsum result?
  • Fix remaining 8-bit table lookups for large vectors (Broadcast, Interleave, LoadDup128): use 64-bit for initial shuffle. For TwoTablesLookupLanes, use 16-bit indices.

SVE codegen

  • SVE2: use XAR for RotateRight
  • CombineShiftRightBytes use TableLookupLanes instead?
  • Shuffle*: use TableLookupLanes instead?
  • Use SME once available: DUP predicate, REVD (rotate 128-bit elements by 64), SCLAMP/UCLAMP, 128-bit TRN/UZP/ZIP (also in F64MM)

emu128 codegen

  • #pragma unroll(1) in all loops to enable autovectorization

Add emu256 target

Reuse same wasm256 file, #if for wasm-specific parts. Use reserved avx slot.

MaxOfLanes, MinOfLanes returning scalar

Avoids extra broadcast.

Reductions for 8-bit

For orthogonality; already done for x86+NEON.

Clear lowest mask bit

For iterating in hash table.

Conflict detection

For hash tables. Use VPCONFLICT on ZEN4.

PromoteToEven

For WidenMul, MinOfLanes.

Add DupEven for 16-bit

Use in MinOfLanes (helps NEON).

Masked add/sub

For tolower (subtract if in range) or hash table probing.

Div (integer division) and Mod

Issue 633.

AddSub

Interval arithmetic?

Dup128TableLookupBytes

Avoids having to add offset on RVV. Table must come from LoadDup128.

LoadPromoteTo

For SVE (svld1sb_u32)+WASM? Compiler can probably already fuse.

Done

  • Signbit
  • ConvertF64<->I32 (math-inl)
  • Copysign (math)
  • CopySignToAbs (math)
  • Neg
  • Compress
  • Mask ops (math)
  • RebindMask
  • Not
  • FP16 conversions
  • Scatter
  • Gather
  • Pause
  • Abs i64
  • FirstN
  • Compare i64
  • AESRound
  • CLMul (GCM)
  • TableLookupBytesOr0 (AES)
  • FindFirstTrue (strlen)
  • NE
  • Combine partial
  • LoadMaskBits (FirstN)
  • MaskedLoad
  • Bf16 promote2
  • ConcatOdd/Even
  • SwapAdjacentBlocks
  • OddEvenBlocks
  • CompressBlendedStore
  • RotateRight (Reverse2 i16)
  • Compare128
  • OrAnd
  • IfNegativeThenElse
  • MulFixedPoint15 (codec)
  • Insert/ExtractLane
  • IsNan
  • IsFinite
  • StoreInterleaved
  • LoadInterleaved (codec)
  • Or3/Xor3
  • NotXor (sort)
  • FindKnownFirstTrue (sort)
  • CompressStore 8-bit
  • ExpandLoad (hash)
  • Zen4 target (sort)
  • SSE2 target - by johnplatts
  • AbsDiff int - by johnplatts
  • Le integer - by johnplatts
  • LeadingZeroCount - by johnplatts in #1276
  • 8-bit Mul
  • (Neg)MulAdd for integer
  • AESRoundInv etc - by johnplatts in #1286
  • OddEven for <64bit lanes: use Set of wider constant 0_1
  • Shl for 8-bit
  • Shr for 8-bit
  • Faster Reverse2 16-bit
  • Add Reverse2 for 8-bit
  • TwoTablesLookupLanes - by johnplatts in #1303
  • Add 8/16-bit TableLookupLanes - by johnplatts in #1303
  • FindLastTrue - by johnplatts in #1308
  • Vec2, Create/Get functions
  • PromoteTo for all types (#915) - by johnplatts in #1387