Skip to content

Are results correct? #21

Description

@fjmorant

In this new benchmark as opposed to https://github.com/tj-mc/styled-components-native-perf-reproducer we have this:

UI 1 2 3 4 5 6 Avg % Slowdown
Native 696 738 711 712 726 684 711.27 0.00%
Restyle 612 622 654 621 621 623 625.50 -12.05%
Emotion 661 657 659 660 661 685 663.83 -6.66%
Tamagui 667 666 753 647 648 701 680.33 -4.34%
Styled v6 772 709 616 614 647 725 680.50 -4.31%
Unistyles 668 698 693 691 712 722 697.33 -1.96%
NativeWind 734 672 699 677 706 743 705.17 -0.84%
fast-styles 725 705 706 730 715 683 710.67 -0.07%
twrnc 659 763 745 776 693 735 728.50 2.44%
Zephyr 743 747 739 713 749 701 732.00 2.93%
Dripsy 774 804 740 875 781 819 798.83 12.33%

All libs seems to be faster than the Native (StyleSheet) apart from the last 3 that are slower and if I check the old benchmark we have this:

1 2 3 4 5 6 Avg % Slowdown
Native 289 281 286 296 280 268 283 0
Styled v5 440 426 424 428 429 432 429 41.0
Styled v6 400 404 401 395 404 400 400 34.2

In the old benchmark for example, Styled v6 was 67.65% and in the new is faster -4.31%, how come it is so different??

Maybe this benchmark was done in a less accurate environment????

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions