json Digest calculated in batches of sizeof(size

Digest calculated in batches of sizeof(size_t)

Open grisumbras opened this issue 2 years ago • 13 comments

Fixes #601

Aug 08 '21 14:08 grisumbras

Codecov Report

Merging #606 (c291598) into develop (8957955) will decrease coverage by 0.05%. The diff coverage is 80.76%.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop     #606      +/-   ##
===========================================
- Coverage    92.98%   92.94%   -0.05%     
===========================================
  Files           85       85              
  Lines         8058     8080      +22     
===========================================
+ Hits          7493     7510      +17     
- Misses         565      570       +5

Impacted Files	Coverage Δ
include/boost/json/detail/digest.hpp	`82.75% <80.76%> (-17.25%)`	:arrow_down:

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 8957955...c291598. Read the comment docs.

Aug 08 '21 14:08 codecov[bot]

Benchmark test results. More info at https://benchmark.cppalliance.org/jsonbenchmarks-pullrequests/606/pullrequest.html

Aug 08 '21 15:08 cppalliance-bot

What is the status of this?

Oct 05 '21 15:10 vinniefalco

I've made 6 variants of this and benched them. Here are the results: https://docs.google.com/spreadsheets/d/1WEfTLx5vEb5TduzzHt-OkAj6HnS4ku4Y4Y7C3YGIgc8/edit?usp=sharing The 6th approach seems most promising, fairly small regression for a few tests and huge improvements for many tests. But my local benchmarking disagrees with CI as you can see on the picture posted above by the bot.

Oct 10 '21 09:10 grisumbras

yeah the automated performance CI bot is not accurate

Oct 10 '21 10:10 vinniefalco

Benchmark test results. More info at https://benchmark.cppalliance.org/jsonbenchmarks-pullrequests/606/pullrequest.html