Randomly ordered details to mind when benchmarking a page load time:
- The averaged timing is valid if and only if it is labelled valid.
- The tool relies completely on a properly fired
load
event to measure when a page has completely loaded.
- This means if a browser improperly fires the
load
event, this could lead to incorrect results when comparing two different browsers.
- If comparing two or more blockers:
- Be sure the benchmarked blocker is the only active extension.
- To benchmark another blocker, disable the current one, enable the other blocker, then ...
- Chromium: open the page load speed tool in a new tab.
- Firefox: restart the browser.
- Configure them in a similar way.
- Clicking the Benchmark button will launch an automated benchmark session which will end when the average page load time is marked as "valid".
- If you want to bypass the browser cache, open de developer console and check the "disable cache when dev console is opened" setting of your browser.
- Bypassing the cache can lead to more erratic results though, as the measurements become more sensitive to arbitrary network latency.
- There is a timeout for when a page never completes loading: 30 seconds.
- The result will be discarded in such case.
- This page load speed tool works only for web pages which accept to be embedded in a frame.
- Since the web page to measure is inside a frame, it may not rendered correctly if you block 3rd-party cookies (a good idea in general).
- This tool can also be used to compare different configurations for the same blocker. For examples:
- When you share the results, you should provide all the information:
- Browser, version, relevant settings (click-to-play, block 3rd-party cookies, etc.)
- Blocker, version, filter lists, settings.
- Which URL was benchmarked.
- Disclose the whole average result of the benchmark, for example:
Average (valid): 832.01 ms (10/13 iterations)
.
- Etc.
- Keep in mind uBlock Origin disables pre-fetching by default -- most other blockers do not modify this browser setting.
- I haven't confirmed so far that this makes a significant difference in page load speed.
- The page load speed tool is completely self-contained, just look at the page source to see how it works.
- Being self-contained means you can save a local copy onto your computer and launch from there.
- There is a prototypical feel to the code, I emphasized on it working, not being elegant. I may work more on it as time allows.