CT_00_029 -- Improve constant synthesis

Owned by Jeff Law
Last updated: Jun 05, 2024
2 min read

When the Zbkb extension is enabled an arbitrary 64bit constant can be loaded in a maximum of 5 instructions.  A lui+addi for the upper and lower 32 bits and a pack to merge them.   High performance uarchs will execute this in 2c and even a simplistic uarch would probably take a maximum of 5c.   Contrast to the what we do now where we push complex constants into the constant pool.  That's probably 3 instructions and 5c on most uarchs.  Naturally we would like to see constant synthesis improved when Zbkb is enabled.  The basic support for Zbkb is development-complete.  So the effort here is really just to use it in constant synthesis.


Known areas to improve:

  1. Use bseti when profitable.  Many cases of poor constant synthesis are failure to use bseti.
  2. Use "uw" variants of shifts/arithmetic instructions when the constant has bit 0x80000000 set.
  3. Use shNadd for constants evenly divisible by 9, 5 or 3.
  4. Use blcri in conjunction with lui or addi to clear a small number of bits, particularly in the high 33 bits of a 64bit word
  5. Adjust constant so low 13 bits are 0x1800, recursively re-synthesize and restore low bits with trailing addi.
  6. Synthesize C' from C using bit inversion.  Synthesize C', then invert the result.  If that's better than synthesizing C directly, then use the inversion sequence.
  7. Use pack for repeating constants
    1. Depends on reassociation of constant in logical ops with shifts
    2. Depends on basic Zbkb support
  8. Use pack as synthesis of last resort.


Stakeholders/Partners

RISE:

Ventana: Jeff Law – general oversight / guidance and implementation work.

External:

                    RAU: Sevak Sargsyan & Lyut Nersisyan – Lyut did the initial Zbkb work under contract to Ventana.


Dependencies



Status

Development

COMPLETE


Development Timeline1H2024
Upstreaming

COMPLETE



Upstream Version

gcc-15 (target)

(Spring 2025)





Contacts

Jeff Law (Ventana)


DependenciesNone


Updates

  • Project done. 

  •  zbkb as synthesis of last support submitted upstream for CI testing
  • Additional ideas for improvements in constant synthesis pushed to 2H2024.

 

  • Pack for repeating constants is integrated.

 

  • Lyut's patterns to allow generation of pack, packw, packh automatically have been integrated.

 

  • Reassociation of constants in logical ops with shifts integrated upstream
  • Using pack to handle constants with equal high/low halves implemented
  • Synthesizing the inverted constant, then inverting the result integrated upstream
  • Adjusting the constant to have 0s in the low bits, synthesizing, then restoring low bits with addi integrated upstream.

  • bclr method for generating constants integrated upstream
  • Adjustment of low 13 bits to make easier to synthesize constant development submitted upstream

 

  • Updated list of various deficiencies in GCC's constant synthesis
  • Note that some of these are recently fixed (first week of May)
  • Hoping to have this done within the next week or so.

 

  • Basic Zbkb patterns written.  Haven't started on synthesis changes though.
  • Noted as a 1H2024 work item.