Once we have the encoder and encoder benchmmark, we'll want to benchmark the protoboeuf MVP. This will be used in my RubyKaigi talk, but can also be used for a blog post. Graphs can be produced with Google sheets.
We can probably just use our rake bench command for this, but we'll want to either do it on AWS, or at least do multiple runs to ensure consistency.
It may also be interesting to benchmark protoboeuf on 3.3 so that we can do a comparison, show the optimizations we've added in 3.4. This would be the icing on the cake, if we have enough time.
This is also done. I measured a 9.45x speedup over Google's protobuf on the decode+read benchmark. We're slower on encoding, which is not optimized yet.
Once we have the encoder and encoder benchmmark, we'll want to benchmark the protoboeuf MVP. This will be used in my RubyKaigi talk, but can also be used for a blog post. Graphs can be produced with Google sheets.
We can probably just use our rake bench command for this, but we'll want to either do it on AWS, or at least do multiple runs to ensure consistency.
It may also be interesting to benchmark protoboeuf on 3.3 so that we can do a comparison, show the optimizations we've added in 3.4. This would be the icing on the cake, if we have enough time.