Online Benchmarks

General

There are lots of online benchmarks available and found on searching Internet. Most measure download speeds and sometimes upload speeds, often from servers all over the world. This page provides access to the Java version of the Whetstone Benchmark and a series of tests that measure image loading times.

Image loading times can vary, depending on the system hardware, operating system and browser. Of particular note, tests show that Internet Explorer is much slower than Opera and Mozilla Firefox on loading lots of small icon images.

Besides the Whetstone Benchmark and Image Loading Time Tests described here, my off-line and on-line Java Drawing Benchmarks are available. Details and links to the on-line versions are in JavaDraw.htm. Note that, as the benchmark loads files from the server, Java Control Panel Security might need a Medium setting to allow the programs to run. It is not advisable to leave this as a permanent setting.

Whetstone Benchmark - Java Version

The Whetstone Benchmark, introduced in 1972, was the first general purpose benchmark that measured computer performance. Later varieties were written using the Java programming language, the first being produced in 1997. The newer version was produced to overcome problems encountered using Firefox and Linux. The benchmark pages provide links to results on PCs, older computers and using various other programming languages. The first version is started using a Run button, but the second starts on loading the page.

Image Loading Time Tests

Three tests load and display image files each of around one million bytes. Format of these files are uncompressed BMP, 0.33M pixels occupying 1M Bytes of RAM, compressed GIF, 2.33M pixels using 2.33M Bytes and compressed JPG, 3.15M pixels at 9.45M Bytes. A further test loads 400 different GIF icons, each sized approximately 70 Bytes. These are intended for use over broadband connections but others, with lower demands, are provided for testing via dial-up lines.

When images are loaded, they are saved in the browser’s cache on the local PC disk, known as Temporary Internet Files with Internet Explorer. On resubmitting the image page address, or selecting Refresh, the images are likely to be loaded from the cache at high speed. To reload from Internet via Refresh, the image page can be renamed as mypage.php, instead of mypage.htm (see Local Tests below), with the time added to the file name on the remote server. Example HTML to do it - <img src="bmpgif/gif1m.gif?<? echo time(); ?>"> - One problem is that multiple copies of the files can be cached with a different suffix - example gif1m.gif?1232202066 and gif1m.gif?1232364678.

Note that the measured time starts after the page has started loading. Also, the timer does not always start on the first load but normally does so on selecting Refresh.

To Start

Dial-Up Tests

Results - Image Loading Time in Seconds

Image loading times can vary considerably, particularly due to network loading. The following show best case results from a limited number of runs using various systems and browsers. For broadband, they are all connected to Internet via the same router, the laptops via wifi. Overall worst case results are also shown.

In this case, maximum broadband speed is claimed to be 8 Mbps, where speed of loading 1 MByte files might approach a minimum of 1.2 seconds, taking into account overheads. Similarly, 100 kB might be around 0.2 seconds. Four of the systems can load 1 MB files in 1.4 to 1.6 seconds and this is similar with different browsers. The other two take a second longer. The slow laptop might be affected by a USB 1 WiFi connection but there is no apparent reason for the PC using Windows 2000.

High overheads can be expected on loading the 70 byte GIF icons and these appear to be considerably higher using Internet Explorer, compared with the other browsers used. The 70 bytes will be transmitted in larger packets. Performance Monitor results on the P4 indicated around 200,000 bytes received and sent (acknowledgement) for 400 x 70 or 28,000 data bytes. Over one second sampling periods, this showed that Opera was transferring data five times faster, with similar CPU utilisation. IE7 generated twice as many disk input/output requests but this does not appear to be sufficient to cause the performance difference.


               BMP    BMP    GIF    GIF    JPG    100    400   400I
             100kB    1MB  100kB    1MB    1MB  Icons  Icons  Worst
 
 P4 XP IE6     0.2    1.5    0.3    1.5    1.6    2.9   23.1   24.7
 P4 XP IE7     0.3    1.6    0.2    1.6    1.6    3.5   31.5   44.6
 P4 XP Opera   0.2    1.6    0.3    1.5    1.6    1.3    5.8    8.3

 Lap XP IE6    0.3    2.5    0.4    2.5    2.5    4.1   34.9   42.0
 Lap Firefox   0.2    2.5    0.3    2.5    2.5    1.6    5.4   19.8
 Lap Dialup   13.0          23.1                 10.7              

 Ath 2K IE5    0.3    2.6    0.3    2.5    2.6    2.2   10.2   16.9
 Ath Firefox   0.3    2.5    0.3    2.6    2.4    1.2    8.2    8.5
      
 A64 X64 IE7   0.3    1.5    0.3    1.5    1.5    2.1    9.5   17.2

 C2D V64 IE7   0.3    1.5    0.3    1.6    1.5    2.2    9.3   14.2
 C2D V64 IE8   0.4    2.1    0.5    3.2    1.7    1.0    4.4       
 C2D Firefox   0.2    1.4    0.2    1.4    1.7    1.1    4.2    4.3
 C2D Chrome    0.4    1.7    0.3    1.7    2.4    1.3    4.3    4.7

 PHE 764 IE8   0.3    1.5    0.3    1.8    1.7    1.1    5.1    7.5
  
 Lap Vis IE7   0.3    1.4    0.2    1.4    1.4    2.2    9.2   13.5
 Lap Firefox   0.2    1.5    0.2    1.5    1.5    1.0    4.2    5.2

 Worst                                                             
 Broadband     0.7   10.5    1.9    9.2   10.7    5.6   44.6       
 

    P4 XP    1.9 GHz Pentium 4, LAN 100, Windows XP, IE6, Opera V9.00, IE7
    Lap XP   1.8 GHz AMD XP, WiFi USB 1, Windows XP, IE6, Mozilla Firefox 3.0.5
    Ath 2K   2.1 GHz AMD XP, LAN 100, Windows 2000, IE5, Mozilla Firefox 3.0.5
    A64 X64  2.2 GHz Athlon 64 x2, LAN 100, Windows XP x64, IE7
    C2D V64  2.4 GHz Core 2 Duo, LAN 100, Vista 64 Bit, IE7
                            later IE8, Firefox 3.6, Chrome 
    PHE      3.0 HHz Phenom II x4, LAN 100, Windos 7-64, IE8
    Lap Vis  1.8 GHz Core 2 Duo, WiFi, Vista 32 Bit, IE7, Mozilla Firefox 3.0.3

To Start

Local Tests

All the above tests have been saved in a ZIP file with the .PHP files renamed as .HTM. This allows them to be used to run the tests via such as a Local Area Network or USB Flash Drive. The PHP added time statements can be seen via View Source and others included in the document header (that do not appear to affect caching).

Running the tests via a local area network can also involve caching implications. On the first test pass (uncached), the larger files should still load too fast to notice (at 100 Mbps) but loading of numerous small files can produce some unexpected slow performance.

Roy Longbottom October 2014

The Internet Home for my PC Benchmarks is via the link
Roy Longbottom's PC Benchmark Collection