[go: nahoru, domu]
Skip to content
This repository has been archived by the owner on Sep 5, 2023. It is now read-only.

Murcake/optimize array load #160

Merged
merged 6 commits into from
Dec 19, 2021
Merged

Murcake/optimize array load #160

merged 6 commits into from
Dec 19, 2021

Commits on Dec 18, 2021

  1. make length/size naming consistent for packed arrays 

    Now 'length' or 'len' refers to the number of cells in an
     array. 'size' refers to the number of bytes. We used to mix them
     and it made things confusing.
    @temyurchenko
    temyurchenko committed Dec 18, 2021
    Configuration menu
    Copy the full SHA
    859aa50 View commit details
    Browse the repository at this point in the history

  2. fix shifted array copy 

    It was just wrong and it just never came up for some reason. Added
     some tests to detect in the future.
    @temyurchenko
    temyurchenko committed Dec 18, 2021
    Configuration menu
    Copy the full SHA
    8ad8dbf View commit details
    Browse the repository at this point in the history

  3. store calldata shifted 12 bytes to the right 

    The reason for the shift is to optimize for the hot-path. Most of the
     calldata read positions follow the same pattern: 'n * 32 +
     4' (the first four bytes in calldata in Solidity represent a
     function selector). Thus, most of the calldata reads are
     /unaligned/ in regards to bit packing. In order to align them, we
     perform the 12-byte shift.
    @temyurchenko
    temyurchenko committed Dec 18, 2021
    Configuration menu
    Copy the full SHA
    b81e863 View commit details
    Browse the repository at this point in the history

  4. validate the number of bytes in calldata

    @temyurchenko
    temyurchenko committed Dec 18, 2021
    Configuration menu
    Copy the full SHA
    8ee513a View commit details
    Browse the repository at this point in the history

  5. optimize 'array_load' to have fewer comparisons and extractions

    @temyurchenko
    temyurchenko committed Dec 18, 2021
    Configuration menu
    Copy the full SHA
    39ba3e9 View commit details
    Browse the repository at this point in the history

  6. optimize 'copy_from_memory_aligned' 

    remove usage of comparisons and inefficient bit manipulations
    @temyurchenko
    temyurchenko committed Dec 18, 2021
    Configuration menu
    Copy the full SHA
    1395734 View commit details
    Browse the repository at this point in the history