I was trying increase performance of our sites and found some strange thing. When I bundled our js files pagespeed had given me better results, but is strange, because our site working on http2 and this protocol should working better with many small files then with one bundled file.
I checked our graylog and what I saw:

As you can see all requests is Http1. Why?
Links:
Our sute
Checking http2 support
For the time being, PSI does not support h2. We even currently disable an audit that specifically checks for h2.
@patrickhulce think we should add this source of variability between channels in the variability docs?
@kusler Maybe we could add a disclaimer about h2 support in PSI somewhere?
There's currently no timeframe for support, but I've just knocked on the appropriate team's inbox to get an update.
Internal to googlers: b/32022322
Holy cow. I can't believe it.
Could it be reason why Google chrome lighthouse give better score then pagespeed insights?
Does it mean that we have to bundle our files if we wanna good score in pagespeed insights?
Here's a test page: https://www.babycenter.com/0_signs-of-labor_181.bc
I get vastly different results on PSI/web.dev (6-7 range) vs. running locally or on WPT ( 45-55 range) in a consistent fashion. They make heavy usage of h2 multiplexing, and I was wondering if PSI not supporting H2 is causing the difference here.
@patrickhulce bump on whether this is documented anywhere? thanks!
@patrickhulce bump on whether this is documented anywhere? thanks!
Definite opportunity for some PSI docs. @paulirish do we have/want to make a PSI FAQ? This would definitely make it on there.
Can I get a clarification on the reason Page Speed Insights doesn't appear to support HTTP 2 even though 90% of internet browsers actively support it?
https://caniuse.com/#feat=http2
This kind of blows my mind a bit lol.
Is there any update on a timeline for this fix to enable HTTP/2 support in PSI? It feels a bit awkward that running Lighthouse directly gives you a different performance score than the one you get when running PSI. Especially since the PSI result page clearly indicates that "The speed score is based on the lab data analyzed by Lighthouse."
It feels a bit awkward that running Lighthouse directly gives you a different performance score than the one you get when running PSI.
yeah :(
I can confirm it's now on the roadmap!
I can confirm it's now on the roadmap!
Great news! sites relaying on http2 multiplex using granular assets will not be penalized more. Also usging http2 push feature properly increases the performance and should be also audited. Thanks.
I can confirm it's now on the roadmap!
Great news! sites relaying on http2 multiplex using granular assets will not be penalized more. Also usging http2 push feature properly increases the performance and should be also audited. Thanks.
Great to see that Lighthouse is supporting this feature now, have we any updates about HTTP2 support from the PSI?
nope. hit the subscribe button, I'll comment on this thread when there's something to update.
I will be also very interested in knowing about this support for PSI. It's hard to convince clients that low scores are because of lack of HTTP/2 support ;)
Any update on this and support of HTTP/2 server push on https://web.dev/measure/ and https://developers.google.com/speed/pagespeed/insights/ ?
Any update on Whether h2 is implemented in the latest PSI ?
There are no updates. See https://github.com/GoogleChrome/lighthouse/issues/7326#issuecomment-573749810 :)
Most helpful comment
I will be also very interested in knowing about this support for PSI. It's hard to convince clients that low scores are because of lack of HTTP/2 support ;)