Chrome 135 results

Results for chrome 135 are published.

Due to the problems with chrome for testing we're still using the chrome desktop version.
Results are once again a bit faster than for 134. Seems like create and replace are about 0.5 msecs faster for the fastest frameworks.

Happy to see that variance in the results is again pretty low and I'm even happier that this run just worked without any issues.

Attached you find a binary build of all frameworks such that you can run the benchmark without installing and building all frameworks (see README for how to use the build.zip)

Chrome 134 results

Results for chrome 134 are published.

Due to the problems with chrome for testing we used the chrome desktop version.
Results are also a bit faster than for 133. Seems like create, replace, remove and append are about 0.5 msecs faster for the fastest frameworks.

Happy to see that variance in the results is again pretty low.

Attached you find a binary build of all frameworks such that you can run the benchmark without installing and building all frameworks (see README for how to use the build.zip)

Chrome 133 results

I just saw that I already grumbled that the chrome 132 run wasn't straight forward. Obviously I didn't know what expected me for chrome 133 😄 .

Some time ago I decided to switch to the chrome for testing (https://github.com/GoogleChromeLabs/chrome-for-testing#json-api-endpoints). I hoped incremental runs were better if I used a fixed chrome version without the security updates. However as it turned out there appears to be an issue with the chrome traces for chrome for testing (#1838 ). A solution would have been to restart the chrome process for each and every benchmark loop. However this creates such noisy results that make no sense to publish (and the whole benchmark run takes too long then).

Luckily the issue does not happen for the normal chrome edition. So we're back to that and I'm absolutely happy to publish the results for chrome 133: https://krausest.github.io/js-framework-benchmark/index.html

Other than that I didn't see big changes.

Chrome 132 results

Results for chrome 132 are published.

We're using Chrome for Testing 132.0.6834.83 for this run.

The run was a little bumpy due to several things:

• total duration for the benchmark is just pretty long. This together with the varying results for each run (see #1721) makes it hard to publish official results. I changed the order of the benchmark such that the outer loop iterates over the benchmarks and the inner over all frameworks. This way I hope to be able to interrupt a benchmark run and continue from that benchmark. Results should be pretty okay, as long as the full benchmark is repeated for all frameworks.
• It took three runs to get complete results: CPU results for keyed in the first night, all benchmarks for non-keyed in the second night and memory and startup for keyed the third day
• There was an error for non-keyed/skruv-liten such that I had to run that seperately after the benchmarks
• Just before publishing the results we found one framework that performs the event logic in the pointerup event and thus before the click event. Thus we didn't count the full duration for that framework. We've added some code to allow selecting which event starts the measurement in the package.json.
• This was the first run on my new MacBook M4 Pro such that it can't be compared with earlier chrome results. Maybe the bigger number of performance cores helps with #1721.

Attached you find a binary build of all frameworks such that you can run the benchmark without installing and building all frameworks (see README for how to use the build.zip)

Chrome 131 results

Results for chrome 131 are published.

We're using Chrome for Testing 131.0.6778.85.html for this run and it seems to be a pretty boring result regarding this benchmark.

Attached you find a binary build of all frameworks such that you can run the benchmark without installing and building all frameworks (see README for how to use the build.zip)

Chrome 130 results

Results for chrome 130 are published.

We're using Chrome for Testing 130.0.6723.58 for this run.
Seems like no big changes for this benchmark.

Attached you find a binary build of all frameworks such that you can run the benchmark without installing and building all frameworks (see README for how to use the build.zip)

Chrome 129 results

Results for chrome 129 are published.

We're using Chrome for Testing 129.0.6668.58 for this run.

Attached you find a binary build of all frameworks such that you can run the benchmark without installing and building all frameworks (see README for how to use the build.zip)

Chrome 128 results

Results for chrome 128 are published. Very close to chrome 127, except for first paint. It might be safer to interpret this first paint metric with some caution.

We're using Chrome for Testing 128.0.6613.86 for this run.

Attached you find a binary build of all frameworks such that you can run the benchmark without installing and building all frameworks (see README for how to use the build.zip)

Chrome 127 results

Results for chrome 127 are published. Once again it looks a bit faster than chrome 126!

There a still a lot of alternative vanillajs versions in this run. Cleaning up is still on the TODO list.

We're using Chrome for Testing 127.0.6533.72 for this run.

Attached you find a binary build of all frameworks such that you can run the benchmark without installing and building all frameworks (see README for how to use the build.zip)

Chrome 126 results

Results for chrome 126 are published. Looks faster than chrome 125!

There a still a lot of alternative vanillajs versions in this run. Cleaning up is still on the TODO list.

We're using Chrome for Testing 126.0.6478.55 for this run.

Attached you find a binary build of all frameworks such that you can run the benchmark without installing and building all frameworks (see README for how to use the build.zip)