v5.53 (c) Sébastien WILLEMIJNS


Version française        disponible ici


  Welcome to my speedtest page,

  Classical method who has been known and developped to test your Internet speed (download a testfile from a webserver) can not be sufficient for high speed because this kind of servers does not "hold out the load" of a lot of users...

  This speedtest use best and efficient method to test your DOWNLOAD speed with the "mirror method" who tries to download in same time a lot of big files from best FTP mirror servers you can access...

  Note this software works well under all Windows versions (Windows XP to Windows 10).


HADOPI - Le Net en France : black-out

Main page Webhosting is offered by

Thanks a lot to these VivilProject partners, too

How it works ?

  First, every FTP server (*) are tested during about 13 seconds to known her individual speedtest...

  Second, the eight FTP servers who gives you biggest data volume with the first test will be used during 2 minutes for a multithread FTP download to known your real speedtest...

- After unzipping the software, launch "go" or "go.bat",
- Validate default choices by typing on "ENTER" keystroke (please read all the text before validations ;),
- Unlock/authorize ARIA2/ARIA2C software if your firewall ask you about this,
- Wait the elapsed time displayed and read the result which will be present on the end of the text report.

Some interesting values:

- VoIP use 12 KBytes/s IP,
- Classical TV DSL use 2,1 MBits/s IP for cartoons to 3,8 MBits/s for movies or TV shows,
- HDTV DSL use 8,1 MBits/s IP for movies or TV shows.

General F.A.Q about this software:

Q: I have unzipped your software on my USB device, it is slow !
A: I/O disk access on this mode are slow with this kind of stuff and can give you lower result than the reality. As explained on the first menus no guarantee are given also if you have a ADSL2+ (or more) connexion ! Installing my software on a real HD let NO TRACES after deleting the main directory !

Q: Can i detect Gigabyte offers with your software ?
A: Yes but you must mandatory read the specific FAQ.

Q: I use a 64 bits version of Windows, it this software works ?
A: Yes.

Q: Is this software is free ?
A: Yes, core is based to open-source projects and all my batch + the servers lists are freeware.

Q: How long this test durates ?
A: between 15 and 25 minutes for guaranteed tests.

Q: My antivirus tells me your software can be dangerous !
A: This software contains a lot of batch commands... some "high sensibility" antivirus can not like this kind of computer language...

Q: What is your guarantee to obtain real result ?
A: In a theorical manner, this test mainly use university networks and FTP servers all around the world which is all reputable more performant than a lot of commercial services as ISP or others multinationals. In a practical manner, this software has been sucessfully tested on several 100 Mbit/s FTTH network lines. I think this software can not actually (in July 2010) detect speed more than 1 Gbit/s for download...

Q: What is his advantage ?
A: It is the only software from 400 (four hundreds) i have succesfully tested to displaying real download result from line over 10 MBit/s... if you prefer to choose beautiful and unprecise speedtest WEB applications it is you choice but not mine sic !

Q: In my case, i use a small Internet offer as POTS or a small DSL offer as 2 Mbit/s) and all WEB tests i use give me same value as your test ! what is the interest of your software for my case ?
A: It is in fact the opposite observation of preceding question and i will give you the same answer: my software has a guarantee ;)

Q: How to unistall this software ?
A: You can safely delete these files and the directory... This software is now 100% portable so they do not let traces on your computer ;)

Q: How to test my UPLOAD speed ?
A: Version 3.0 to 3.31 included an upload test but it can be technically easily detected so i stop to purpose it... Upload speed can be also tested by transferring a big file to a FTP site or by using a upload test site ( here or here ) without obtaining bad result because UPLOAD tests use less bandwidth than DOWNLOAD tests... (in 2007, the majority of servers detect my 1 Mbit/s upload speed but not 50 Mbit/s FTTH uploads speed...)

Q: Is this software is compatible with BartPE ?
A: No because this kind of emulation is too slow (i have tried to install it in my hard disk with no success)

Q: Your tests are too long !
A: but they include a guarantee ;) choose your camp.

Technical F.A.Q about this software:

Q: What is the theorical limit of this software when i use long version ?
A: This software can give bad results when the sum of average speed from eight best tested FTP servers is smaller than your maximal theorical speed... When i wrote this sentence (July 2010) and since the France, the safety margin for my software to guarantee the result when the Internet network works well was about 3000% (yes 3000%) on a DSL 28 Mbit/s ATM and about 450% on a 100 MBit/s IP network line.

Q: I actually notice than my Internet connexion in some countries is dramatically slow but your test gives me my full speed ! why ?
A: The given final speed by this software for international tests is only a GLOBAL parameter which is often reached even if some Internet network part of your ISP are down or overloaded. In others words, my international tests (with no report) can not permit us to detect some networks problems at your ISP but ONLY on a signifiant part or the totality of his network !  "1" menu from extra tools purpose to list all tested servers and calculate a "speedsum" which permits to you to counterbalance the main result if you suppose they are a temporary problem...

Q: Why your quick test is not accurate than long test ?
A: (FYI this test does still exist on new v5.0 core) Because all FTP servers are tested by "eight groups" in accordance with list order: the hazard can do that an average speed FTP server (tested with 7 down or lazy others servers) can be better classified for the final test than 8 best FTP servers tested together in same time...
  3.30 Version now purposes, with an one minute additionnal time (compared to old quick test) to increase his accuracy by testing again the 20 best FTP servers tested... in a practical manner, this test well increase the quick test but real best and good test will always be the long test because too few FTP servers (in February 2007) reachs speed more than 50 Mbit/s...

Q: Why you does not use HTTP test servers for downloads ?
A: Only because my test can be detected (with "User-Agent" and "Referer" info)

Q: Why you use 8 simultaneous threads for this kind of test ?
A: My internal tests shows we must use more than 3 threads to guarantee and saturate an Internet link. The other advantage is to increase safety margin (see below) because more you use threads (in a clever way), less FTP servers you use can be "ultra-powerful" to guarantee result...

Q: Why some WEB speedtest can give me a value greater than your software ? A: All WEB speedtest installed in some big servers -AND- which using very small testfile (which give you result in less than 3 seconds for high speed) can give you in about 10% of case some greater values who can be higher than 25% of your maximal average speed because this delay is too short for the ISP hardware to stabilize your commercial rate...>BR>
Q: Why your software does not test all FTP servers in same time?
A: Because it is really unnecessary and disrepectful to use on very test during 2 minutes some faraway servers from your country which in all cases will does not "hold out the load"...

Q: I have quotas at my ISP. How many megabytes your software use per test ?
A: A classical test at 20 Mbit/s use 920 MB in download and 15 MB in upload. A quick test (always at 20 Mbytes/s) use 220 MB in download and 7 MB in upload. In 2011, this values are near proportional until 100 Mbytes/s and this proportionnality decrease when you reach Gigabytes/s connexions. I can not know over sorry.

F.A.Q about speed test:

Q: How to optimize my PC ?

Q: I'm afraid my computer is too old !
A: FYI 6 Mbit/s IP speed (7,2 Mbit/s ATM) has been very hard to optain under a Pentium II 233Mhz Windows 98SE without firewall, antivirus and others spywares ;) We can write here without tell you some big mistakes that a Pentium I is the minimum to surf on 512 Kbits, a Pentium 3 for 8 Mbit/s and a P4 to surf on 20 Mbit/s (i only suppose for the last thing)

Q: I'd like to compare values you display with real commercials offers...
A: See below:

Comparison table between software values and commercials offers
IP value
Real ATM value Name of
commercial offer
4 Kbytes/s
"56K dialup"
17 Kbytes/s
134 Kbit/s
160 Kbit/s
"128Kb DSL"
67 Kbytes/s
538 Kbit/s
640 Kbit/s
"512Kb DSL"
84 Kbytes/s
673 Kbit/s
800 Kbit/s
128 Kbytes/s
972 Kbit/s
1216 Kbit/s
"1 Mb DSL" (IP)
256 Kbytes/s
1944 Kbit/s
2432 Kbit/s
"2 Mb DSL" (IP)
431 Koctets/s
3449 Kbit/s
4096 Kbit/s
"4 Mb DSL" (ATM)
862 Kbytes/s
6898 Kbit/s
8192 Kbit/s
"8 Mb DSL" (ATM)
1077 Kbytes/s
8623 Kbit/s
10240 Kbit/s
"10 Mb DSL" (ATM)
1618 Koctets/s
12935 Kbit/s
15390 Kbit/s
"15 Mb DSL" (ATM)
2160 Kbytes/s
17246 Kbit/s
20520 Kbit/s
"20 Mb DSL" (ATM)
2371 Kbytes/s
18970 Kbit/s
22528 Kbit/s
"22 Mb DSL" (ATM)
2586 Kbytes/s
20695 Kbit/s
24576 Kbit/s
"24 Mb DSL" (ATM)
3018 Kbytes/s
24144 Kbit/s
28672 Kbit/s
"28 Mb DSL" (ATM)
100 Mbit/s
"100 Mb FTTH" (ATM) (1)

(1) DSL and FTTH are not the same technology / It is too early to have some real values because they are too few customers in France ;)

Q: Every workdays between 5pm and 10 pm and every week-ends, my bandwidth can have 10/20% differences at a few minutes of interval, can it be logical ?
R: Yes if your ISP has non-optimized liaisons...

F.A.Q about Gigabits connexions and my software compatibility:

- As i spoken above in this (huge) text, the software depends too of the CPU+memory+BW "power" of the remote servers we will use. In 2010, i think it is be very no so easy to look for servers.

- This software has just been tested on a 1 Gbps/s connexion. A temporary tip (not guarantee) is to change (for the moment) the default number 8 to 54 for "NumberOfServersForFinalTest" in vpst.ini file. This tip only works with default menu (choice number 1+1) else the tip will not work and will display you an error !!!

- FYI, in March 2013 in France on a 1 Gbps/s line, i have an average of 500 Mbps/s (18 times the max for a DSL French line) for the best server (8/9 have an average peak of 500/350 Mbps/s, 10/15 have an average peak of 100/350 Mbps/s and the rest stays from 0 to 100 Mbps/s). Never forget your server and the remote server must hold the load each other to have the best results...

Patchs (useful to delete some "boards effects"):

- DEDIBOX (ILIAD / FREE.FR) users: place this file (version 1.01) on the same directory than "go.bat" file to do not have the error than one or severals files has been anormally quickly downloaded ;) (this patch temporarly removes from the list the IMAGE.ISO file from TEST-DEBIT.FREE.FR)

- OVH / OVH.NET users: place this file (version 1.01) on the same directory than "go.bat" file to do not have the error than one or several files has been anormally quickly downloaded ;) (this patch temporarly removes from the list the GRENOUILLE.COM test files and OVH mirrors which can be used on the future)

Known bug:
- If you want to use ARIA2C up to 1.17.1 to resolve the abnormal DNS resolution time in numbers 1 and 6 menues:
- you must use my v5.53 and modify "vpst.ini" file by using 26 instead of 13 on DebugStopAfterSSecond line. The duration of the test can be twice the time displayed.
- or use v5.54 with this patch
or this
which to be placed on the main directory.

If you discover a bug:

No software or programmer is perfect so please give your report by mail with this informations:
- Used software version number ?
- Used Windows version (if you use Vista, do you use yet UAC ?) ?
- What firewall and antivirus do you use ?
- Please describe with a lot of precision the problem, copy/paste the error or send me screenshot (via imageshack.us if possible)...

Thanks in advance for your help...

What's New:

v 5.00:
- Now the software is compatible with Windows 64 bits OSes,
- Now the software only exist in portable version format,
- Some menus has been deleted.

v 5.01:
- Small enhancements.
v 5.02:
- Emergency update: The software forgot to update translations the 8/2/2010 between 6pm30 and 6pm40 GMT.

v 5.03:
- Second emergency update: only non-guaranteed test worked between 6pm30 and 8pm20 GMT. I has only modified some stuff on test version not on dev ;)

v 5.04:
- The new method used in v5.00 can decrease the warranty result of 0.8% (0.8 to 8% in FTTH or Gigabyte/GB connexions) (FYI in 2010, only Japan and a nordic country purposes GB offers for end-users),
- Lot of (bigs) internal changes. One new module permit to earn from one to three minutes on the longest tests.

v 5.05:
- Some small bugs repaired.

v 5.06:
- Fixed bug than Autoupdate looped forever if used update server was down.

v 5.07:
- Add a patch possibility in case of updates troubles.

v 5.08:
- New ARIA2 software changed internal output format and VPST protection detected abnormal results :-( Thanks Estevan.

v 5.09:
- Oups... the error was not repaired in fact :-( Mandatory changes permit to now lose from one to three minutes on the longest tests grrrrr !!!!

v 5.50:
- A small English translation error,
- Margin error for guaranteed tests has been reduced from 3 to 2 (the lowest fair value i can do !). This is due to a good mix (sic!) of capped/saturated links or remote servers due to the actual economical crisis. Since 1 or 2 years, the result is too abnormal and value is only good between 2 and 7am GMT !!! WHAT IS THE INTEREST TO OFFER HUGE BANDWIDTH ???

v 5.51:
- Better reset option thanks to "reset.bat" file.

v 5.52:
- The 1 Gbps/s test patch only used default number serveurs :-(
- I use now OPENSUSE list... maybe i will add again MAGEIA and MANDRIVA dunnon...
*** 14 december 2013: downtime *and* EN version instability between 12pm and 2pm GMT time. MAGEIA and MANDRIVA had been added on the list.
v 5.53:
- This software is now compatible with ARIA2 1.17.1 and up. the bad news is than, on this case , the test duration is twice :-(.
v 5.54:
- This software is now compatible with ARIA2 1.17.1 with normal speed with a patch.
*** 22 october 2017: small hosting crash during 10 minutes about 10:10pm.
FYI stable v5.08 version include auto-update ;)

(*) One French test FTP server and all MANDRIVA ISO distribution mirror servers in v3.00, all MANDRIVA+MAGEIA+OPENSUSE distribution mirror servers since 14th december 2013.


- Thanks to THF for question.exe and Patrick Vincendon for winlocate.exe,
- Thanks to Pierre, SEKH, stlf, zebu66 and gsrider (beta-testers) ;)

LAST DATE OF CORE DATABASE UPDATE: 22/10/2017_rev2 (DD/MM/YYYY date format)


  Using this kind of FTP servers is not logic (it is not for speedtest all these FTP mirrors servers has been made...) so please be clever for the Internet community to do not abuse to launch this software...

  FTP Admins who wish to do not include his FTP server from all my FTP list can write me and create an empty file in the directory before the mirror directory as "ftp.foo.com/pub/mirror/no-speedtest-here-please.txt"

 Subscribe to these free online services and receive extra storage as me ;)
 souscrivez à ces services de stockage gratuits et gagnez comme moi de l'espace en plus ;)

Follow "vivilproject" (very low traffic about all my projects) on TWITTER
or click here if you prefer FACEBOOK

Be notified of
page updates
it's private
powered by
Get notified when this page changes!
by WatchThatPage.com
Follow This Page

Your suggestions and comments interest me !
To display my e-mail address, complete and validate the sentence
"It's hot outside, it is due to the

Keywords: speedtest speed test gigabyte Gigabyte/s GB GB/s windows reliable FTTB FTTH optical fiber very high speed highspeed stable guarantee guaranteed portable freeware
Bookmark and Share
        Tell a Friend

Mesurez votre audience