This is just for open discussion and FYI. Check out this comparison of page rendering speeds between the announcement forums of vBulletin.com, XenForo.com and InvisionPower.com. I thought the differences are interesting and say a bit about each software's performance. The charts you'll find below the rendering frames are really interesting. I didn't think IPB would be worse than vB in overall rendering speed, but according to this, it is. But then, we know vBulletin has a Varnish cache server sitting in front of their web server. It would be interesting to know, if IPB and XenForo do too. I don't think XF has any and that says a whole lot, especially since XF blows them both out of the water! Scott
Just to be fair, test server's geographic location in comparison to target site's server and web host datacenter location also play big part in page load speeds Dulles, Virginia webpagetest location is closer to Xenforo's Softlayer major datacenter in Houston, Dallas etc than vBulletin's on the far West coast. Just do individual tests on webpagetest.org and try selection California test server and see some differences. And IPB is actually behind Cloudlfare proxy which is known to sometimes slow down sites rather than speed up. Also number of posts in each respective Announcement forum Xenforo Announcement forum = 233 messages vBulletin.com Announcement forum = 2,359 posts IPBAnnouncementforum = 424 topics My public vB 5.0.4 test forum with custom style http://nginx.vbconnected.com/504/ on low end 512MB memory based VPS server. Hosted in San Francisco. But no one said vB5 performance is ideal.. still some way to go for better performance.
Well, geographic location within the US shouldn't make too big a difference. Well, not of all of them have the same number of HTTP requests. Being vB and IPB need twice and many as XenForo, that does make a difference though. As for the threads queried. If there were a couple million to a couple of hundred, I'd agree with you, but a couple thousand to a couple of hundred? MySql should spit such queries out almost at the same time. I just tested San Jose and vBulletin was actually a lot worse. First viewed element was at 4.5 seconds. Maybe we should just leave this at something nice to look at and to take a pinch of salt with and maybe make note of the tool pagespeedtest as something cool to use. I'd also say IPB needs to look at their system a bit more in depth, especially now that you say they do have a front-end proxy too. Scott
Yes number of http requests matter too. As to variance, you might need to run first and repeat view and set test runs to like 6 or something to average them out. Some of webpagetest.org's test servers are heavily loaded at certain times of the day. With webpagetest.org focus on speed index value too - time to visually render the page. Also try gtmetrix.com too with advanced option to create video as well
When I upgraded my site to xenForo, a while ago now there was a definite speed increase. xenForo really does rule in that regard.
