Weird Results?

Hi,

Im having the toughest time, here are my results :

http://www.webpagetest.org/result/131204_MG_735/

Pingdom Website Speed Test results : load time 3.08s

GTmetrix :

http://gtmetrix.com/reports/hermanfurniture-online.com/vcsRclYX

all the testing differ vastly, can someone please enlighten me of whats going on?

I’m usings the basic plan on hostwinds, should i change my web hosting altogether?

Thanks,

Cheers

The test results differ because the variables differ. On WebPageTest you’ve selected the Singapore test location and IE8 browser, at GTMetrix you’re testing with Firefox from a Vancouver, Canada location. With the huge amount of resources you’re loading, the added latency is going to make a big difference. A test from Denver, CO (closest to Vancouver) will be more similar to the GTMetrix results:
http://www.webpagetest.org/result/131204_W1_DP7/

But if you look closely at your WPT results from Singapore, you’ll also notice most of the time is spent processing the page by the CPU (graph at the bottom of the waterfall). So that test machine is having a tough time rendering all your page elements.

As for Pingdom, those results look rather unreliable; more of an estimate of page load time than a measurement, it seems.

In short, all WPT/GTMetrix tests are run on computers with varying hardware configurations, so it is to be expected that results will be different between sites and test locations. They may also run different test software and interpret the results differently.

Hi Robzilla,

Thanks for the info, i overlooked it! My client is from Singapore, how can i know what the loading speed in singapore is? right now, the test results in singapore on my site is :http://www.webpagetest.org/result/131204_WK_GAZ/

75 secs is horrid. I’m using the basic plan on hostwinds, should i change my web hosting altogether?

Ideally, you would host your site closest to where your client’s clients (i.e. most of the site’s visitors) are located. If that’s Singapore, then you should probably host your site with a company in Singapore. If it’s a worldwide audience, Dallas, TX should be fine.

As I mentioned earlier, the long load time in your Singapore test results is caused by CPU processing, so it doesn’t necessarily take as long for everyone who visits the site. It may simply be that the test location was experiencing high loads at the time, or is hosted on slower hardware than other test locations. The biggest win for you is to cut back the number of HTTP requests (images mostly).

btw, the best way to know how the site is performing for the actual users is to look at some real-user perfomance reporting. Google analytics and SOASTA botha have free offerings. Tools like WebPagetest are good for diagnosing “why” it’s slow.

That said, 10MB of images isn’t going to be fast for anybody. It looks like you have a bunch of the product images as png’s and that’s where the bulk of the issue is coming from. Saving them as JPEG’s (which is what they should be since they are photos) should eliminate most of that bloat.

Thanks for the both of you for taking a look at my problem.

Regarding about what Pmeenan said : Saving them as JPEG’s (which is what they should be since they are photos) should eliminate most of that bloat.

Sounds like a good idea, but how much savings if i convert them into Jpeg?
Changing the extension from .png to .jpg, that means i will not be able to replace all of the files. Will have to start all over again?

cheers,

There are 10MB of images that should drop to under 1MB when you compress them correctly so it’s VERY worth the effort. It looks like there are also some background png’s that are quite huge but only 2-3 color - saving them as palletized png’s will save a bunch.

Basically look at every png on the site.

What do you mean by start all over? Technically you could save the jpeg’s and keep the file name as png since the browsers all do content sniffing to figure out the image formats but you’d be better off with correct extensions just for your sanity. You should be able to automate the conversion and do it all as a batch.

i see… Thanks. :slight_smile: any software that can do this : “Technically you could save the jpeg’s and keep the file name as png…” ?

My .pngs are optimized already, so i’ll save it as .jpg?

-edit- i found Imagemagick can do the trick. http://www.imagemagick.org/Usage/basics/#mogrify but I’m not familiar with command line. anybody can chime in here? perhaps Pat?

Cheers,

You probably want to reach out to whoever does the dev on the site because you might break things.

I’d probably do it with a shell script but there are command-line utilities on Linux that do a good job of the conversion for individual files.

i.e.

mogrify -format jpg *.png

will create jpeg versions of all the png images in a directory (using default quality settings).

Imagemagick has a steep learning curve.

The easiest (and free) program I know of is Faststone
http://www.faststone.org/FSViewerDetail.htm

Just view the image, move the cursor to the left hand side of the window and a menu will appear. Click save as (a shorthand key Ctrl-S will take you to this point too) , select jpg, then click options and reduce the compression to 70-75%. While you are in the options window it will show you the compressed result and output size in bytes as you change the compression.
[hr]
A missing or incorrect IMG WIDTH HEIGHT is considered major. That is why GTmetix emphasizes this factor. The Browser starts with a wire-frame rendering leaving the area for images allocated based on the HTML WIDTH and HEIGHT. when the image is retrieved it is simply render in the allocated area. If the WxH of the HTML do not match the image the Browser then take the time to scale. If there was no HTML WxH it is very likely the Browser is going to have to re-start rendering to fit in the image.

Hi,

Kudo to Patrick, Robzilla, iSpeedlink, i managed to lower my speed to 12 sec :frowning:

here is my results : http://www.webpagetest.org/result/131210_X2_VNK/

The test result shows grade A is almost every aspects but my time is far to slow, 12secs. and Ive change my host to a localize host (site5 - singapore). do you guy think, i should upgrade my server from basic plan to VPS or ?

And is there any way i can test my site accurately in singapore? my site is meant for the singaporean customers only?

Thanks in advance.

Cheers,

Your problem is still that you’re loading too many images on that one page. So many, in fact, that it freezes my browser until it’s done. And most of the images aren’t even visible on the page without scrolling left and right. Best to load those only when needed.

Robzilla,

“Best to load those only when needed” How do i do this ? only load when the image is visible on the homepage, it will definitely speed thing up… then the rest of the images can load when is needed.
any plugins or solutions?

Cheers,

You’d probably have to customize the plug-in you currently use to create the image sliders, reduce the number of images, or replace it with a different slider. Do you really need to display all those images in your sliders? Won’t a small sampling of your furniture suffice?

Robzilla,

Hmm… i’ll have to reduce the images loaded on my homepage. :frowning: Thanks for your advice.

You are still using png for photographic images. When the image is a photo you should almost always use jpg.

The latest results you post today were for IE8 Browser. Why? Your page does not render or function correctly in IE8. IE8 accounts for 7% of Singapore users.

http://ispeedlink.com/ie8.jpg

That’s not the worst of it. Once the page loads, on the one gallery slider that does render the "Next: arrow does not work.

The Opera Browser while not very popular is a good benchmark of a well designed web page. A well designed page will render perfectly in Opera. You page does not render anywhere close to acceptable in Opera. here is a screen shot from Opera.

http://ispeedlink.com/opera.jpg

You ask if there is a plug-in solution. Plug-ins are at the root of your problems. A 12 second page load is the best you can expect when you use Word Press and too many plug-ins.

Yes you have five grades of A and one grade F. The five A’s are important, but your one F is so serious the five A’s will do you no good. You had the same five A’s when you had a 66 second page load.

A well designed website will not use javaScript and will only use HTML and CSS for the first view. Your site will not render or function without javaScript.

You problem is the fundamental design is seriously flawed.

Go back to the drawing board. Remove all plug-ins. Get the site functional with no plug-ins.