Code

Dilithium/ML-DSA Stack Optimizations#340

Code

Merged

mkannwischer merged 32 commits intomupq:mastermupq/pqm4:masterfrom dop-amin:dilithium_stackoptdop-amin/pqm4:dilithium_stackoptCopy head branch name to clipboardApr 16, 2024

+5,540-21

Commits

Commits on Apr 15, 2024

• Init dilithium3 stack optimized variant

  

  dop-amin

  authored and

  mkannwischer

  committed
  44e901c

• Start stack optimization [Passing]

  Show description for 80c9e07
  

  dop-amin

  authored and

  mkannwischer

  committed
  80c9e07

• Compress w

  

  dop-amin

  authored and

  mkannwischer

  committed
  5c5b868

• Eliminate z, y

  

  dop-amin

  authored and

  mkannwischer

  committed
  926e957

• Eliminate cp

  

  dop-amin

  authored and

  mkannwischer

  committed
  302f7f2

• Eliminate s1, s2

  

  dop-amin

  authored and

  mkannwischer

  committed
  3c36dbe

• Eliminate second poly needed for A*y

  Show description for f71e025
  

  dop-amin

  authored and

  mkannwischer

  committed
  f71e025

• Inline sampling uniform and uniform_gamma1

  

  dop-amin

  authored and

  mkannwischer

  committed
  deeabab

• Inline hint generation

  

  dop-amin

  authored and

  mkannwischer

  committed
  cbc29cf

• Inline polyw subtraction

  

  dop-amin

  authored and

  mkannwischer

  committed
  8468d60

• Refactor decompose to high/lowbits

  

  dop-amin

  authored and

  mkannwischer

  committed
  b4505e7

• Inline Keccak state

  

  dop-amin

  authored and

  mkannwischer

  committed
  f5a8a65

• Shared buffer for polynomials

  

  dop-amin

  authored and

  mkannwischer

  committed
  10d4766

• rm 257 FFT

  

  dop-amin

  authored and

  mkannwischer

  committed
  2804237

• Union for small and big poly

  

  dop-amin

  authored and

  mkannwischer

  committed
  d30a766

• Eliminate some smaller buffers

  

  dop-amin

  authored and

  mkannwischer

  committed
  a37b5a6

• Remove asym small mul

  

  dop-amin

  authored and

  mkannwischer

  committed
  2bd00ad

• Stack friendly uniform_gamma1 w/o add

  

  dop-amin

  authored and

  mkannwischer

  committed
  77a7572

• Stack optimized Dilithium{2,5}

  

  dop-amin

  authored and

  mkannwischer

  committed
  6609f82

• Switch to Plantard-based 769 NTT

  

  dop-amin

  authored and

  mkannwischer

  committed
  59724a7

• First batch of stack opt for Verify

  Show description for 0dd789b
  

  dop-amin

  authored and

  mkannwischer

  committed
  0dd789b

• On-the-fly unpacking for z, h

  

  dop-amin

  authored and

  mkannwischer

  committed
  a8c993f

• Compress w

  

  dop-amin

  authored and

  mkannwischer

  committed
  b7ded84

• rm tmp poly, subtract on wcomp

  

  dop-amin

  authored and

  mkannwischer

  committed
  e6e164b

• Verify Stack Optimizations

  Show description for 6ef4fbc
  

  dop-amin

  authored and

  mkannwischer

  committed
  6ef4fbc

• rm buffers/unionize in Verify

  

  dop-amin

  authored and

  mkannwischer

  committed
  9870bec

• Stack opt key pair

  Show description for 1d21996
  

  dop-amin

  authored and

  mkannwischer

  committed
  1d21996

• Overlap buffers

  

  dop-amin

  authored and

  mkannwischer

  committed
  76b16c1

• Stack optimized challenge generation

  

  dop-amin

  authored and

  mkannwischer

  committed
  e718f2e

• Match 769 Plantard to m4f code

  

  dop-amin

  authored and

  mkannwischer

  committed
  a37b311

• update skiplist

  mkannwischer

  committed
  d401a15

Commits on Apr 16, 2024

• update benchmarks

  mkannwischer

  committed
  c013920