Even with SPDY disabled[1] HTTPS still appears faster (not by much: ~5% in my case). I do agree that enabling SPDY by default for the HTTPS case is unfair, but do not let it eclipse the takeaway: most users with most browsers will observe HTTPS being a bit faster (edit: I should rather say "comparable in speed") in this specific benchmark. How well this result translates to other sites is a different question. For example the benchmark does not measure TLS handshake latency:

  <!-- "pre-load" HTTPS connection to remove TLS handshake latency when switching to HTTPS test. And set the detectio var -->
  <script src="https://www.httpvshttps.com/detect-spdy.js"></script>

For some sites, handshake latency matters very little, so it makes sense to write the benchmark the way they did it. Anyway, the company who made the benchmark probably did it to prove a point: that an image-rich web page can load just as fast on HTTPS as on HTTP.

Also, contrary to what a commenter said below, the site does carefully avoid caching interferences by downloading images with a random query string such as https://www.httpvshttps.com/check.jpg?123.179881

[1] eg. in Chrome pass --use-spdy=off