WPT load times vastly different from other speed tests

Support
1

Hi,

I’ve been using your tool for some time, big thanks.

I’ve just switched over my site to the Genesis framework & theme. My test site on a cheap server is running blisteringly fast measured with WPT: http://www.webpagetest.org/result/151005_TS_C0C/

However my live site, VPS server, Genesis accelerator service is running like a dog measured with WPT:
http://www.webpagetest.org/result/151005_Q4_DSQ/

However load times with Pingdom, GTmetrix, and Dotcom-tools looks ok.

https://www.dotcom-tools.com/DetailView.aspx?id=5682ECB5BFAB4783943D56BF65BF9FAF:1

Any reason for this big difference, I’m confused which is right.

Thanks
Kevin

2

The first thing that jumped out at me is your test site is smaller than the live site, fully loaded: 880 KB (test), 1,423 KB (live). To go along with that number of requests on the live site is 82, the test site 48.

Those two data points make this an unequal comparison.

3

Hi Joseph,

Thanks for the insight, my apologies for the different sites.

Here’s a better example:

On my live site I’m getting a page load of 4.9 secs with webpagetest:

http://www.webpagetest.org/result/151008_HC_T7M/

whilst the test with gtmetrix shows a page load of 1.3 secs
925ms with Pingdom: Website Speed Test | Pingdom Tools

https://gtmetrix.com/reports/www.vintageheirloom.com/bsj4Hci0

Testing a few times I get the same big difference. I’ll assume webpagetest is correct. there is still a slight difference in the page size, which seems strange.

Any ideas why webpagetest is different, perhaps the others are just inaccurate?

thanks

4

There is no one “correct” performance result. The WebPageTest result you linked to was run from Ireland, Pingdom was from Amsterdam, and GT was from London.

To break it down even further, the WebPageTest result was for a throttled “cable” connection ( 5Mbps down, 1Mbps up, 28ms RTT ). I didn’t see an indication for Pingdom or GT on what, if any, network throttling was used.

Instead of comparing results of tests done under wildly different conditions I’d recommend using one set of conditions consistently over time.

5

Thanks Josepth

Those were the closest zones available to test against without a long wait. I’m trying again but only webpagetest and gtmetrix have London.

I’m using (5/1 Mbps, 30ms) with GTmetrix - this looks very similar (is that 5.1 Mbsp ? ).

webpagetest is so popular in london, long wait…

I’ll get back, but am I now comparing like for like?

thanks
Kevin

6

I’d recommend picking just one tool ( I’m a fan of WebPageTest, so I’d vote for that one ), they periodically run the same test, under the same conditions, and track the results over time.

7

Yup you were right much much closer comparing both against London:

London webpagetest 2.9 sec: http://www.webpagetest.org/result/151008_5C_ZTV/
London gtmetrix 1.9 secs

thanks