Leaderboard

The same logic where more people are likely to post negative reviews than positive ones. You're probably right, to a degree. I think there are also plenty of people who test here on sunny days. If that is true, it's true across the board with all providers. They're all being judged using the same method. In my experience, the better providers always perform better and rank higher here. Is any of this perfect, no. Will it ever be perfect, no. Will I try, definitely. After TMN's recent full switch to https (SSL) settles in with the search engines I have an upgrade to the host stats pages you mention. It aggregates a much broader range and I think it better represents the highlighted ISP, city or country's speed because it won't just be showing you the logs like it does today. It will unlock a huge amount of information that TMN has been calculating and storing in private databases for a very long time now. So look for that tool to become higher resolution in the near future.

Hi swantesty, I'm using the correct conversions, please see https://testmy.net/understand-bandwidth. In your details you can see next to the result in Mbps is also the result in MB/s. You may not be getting the same results as other tests because TestMy.net is not the other speed tests. Other speed tests draw resources in multiple streams. Although TestMy.net can test that way (enable the multithread speed test and test again as you normally do, keep in mind it remembers the setting next time) it does not work this way by default. As a community we found right away that the multithread testing method that everyone uses often masks issues. You can achieve higher results in multithread even if you have an issue affecting your bandwidth. You should instead be tuning single thread performance, if that performs at it's full potential the multithread results will always fall in line. Multithread speed test results favor the ISP. They'd rather you had 10 pipes at 10 Mbps than to draw that all down one pipe. In that example you will appear to have 100 Mbps when you aggregate the result into one but in TMN's eyes... you have 10 Mbps. TMN isn't here to make it easy, it's here to help you improve. It will only call it how it sees it. I've seen it on my own connection and my own mother has seen it on hers. Both times (separate instances) we knew we had an issue at the time so we went out and used a bunch of other speed tests to compare what members had been telling us for years. speedtest.net and others reported 10X higher result when we had modem issues that were physically limiting the single thread performance. If we had only used the other tests we could have thought, "well it must just be the websites I'm visiting." --- The only test that gave indication to the issue was TestMy.net. Modems were swapped and both times and instantly performance returned on TMN results. Re-testing other speed tests had pretty much the same results as before so again, no indication. We had many other factors that were tested outside of speed tests to prove the performance was in line with what TMN was reporting. That was probably over 7 years ago, the same is true now. The methodology here is the same as it has been since TMN's inception in 2001.

I ran an automatic test and the provider service was so bad at times that the test timed out. But this is not reflected in the results. I copied the results into Excel and made my own graph with the results including zero speed when the test timed out. (See included image of results with zero results included) The graph with those zero results included shows a true picture of the service while the website graph show a better level of service. I think zero speed (timeout) results should be included... Noel

Let this sink in for a moment. Now have a look at setting in squid locally:

It is ironic that you can only get 1990's era speeds at a luxury $400/night hotel, that is a short days drive from silicon valley. The only place I've had worse internet recently was in Cuba. (The food is just as overpriced and substandard as well BTW)

If ISPs decide to throttle traffic, they would need to do this all services to be effective, in which case it would probably be easier for them to charge by maximum speed like many ISPs did in the post, e.g. one price for 10Mbps, a higher price for 50Mbps and so on. Otherwise, it would be relatively straight forward to overcome with the use of a VPN. Let's say an ISP prioritises port 8080 to deliver fast speed tests while throttling everything else, just use something like OpenVPN over port 8080 with a VPN privacy service. To the ISP, all your traffic would be seen and treated as speed test traffic. For example, about two years ago the Irish cellular networks Three and Vodafone were doing something similar, i.e. throttling most traffic over the standard web ports (e.g. HTTP port 80), while letting port 8080 run at full speed to deliver fast speed tests, at least with the well-known Ookla Speedtest App. At the time, it meant one could get 4G speed tests over 20Mbps, yet faced slow browsing speeds similar to a 1Mbps connection. All I had to do was make a VPN connection over port 8080 (same port # as Ookla uses for its speed tests) and everything performed a heck of a lot better. A few months later, Three changed their tactic by prioritising certain services such as YouTube when the network is congested. This means YouTube can potentially play 4K fine even when the speed tests (including Ookla) deliver low test result figures.

Okay, so I was just going through forums topics and profiles, and just realized,.. It's @CA3LE's birthday today! Happy Birthday Damon! Hope you have a good one, I don't know if you celebrate or not but I wanted to wish you a happy birthday and show some gratitude! It's on a Friday, so this might be a fun weekend for you, if not ... MAKE IT ONE! Wohoo!

The fact that I have realized the power of GIF here... I am going to be having fun. Trolololo Haha yea I read the PM first, habit of reading PMs before Posts on Forums Wait. I JUST realized YOU MADE MY LABEL "this guy's cool" LOL

@barney79, Welcome to testmy.net forum. If your testing to a specific server such as the testmy.net Australian server, and the test is originated from the same continent, that is where the test is run. Full stop. The reporting server has no bearing on the test results. The test has already been completed before results are sent anywhere, the Australian test server simply reports findings to the main testmy.net server, records them into the database, and reports back to you, after the test has been taken. If you were to run the multi threaded test ( selecting more than one server, then yes there would obviously be more variables involved) but that is not your question or case it seems. Not sure where the provider is getting this information, but generally speaking, that theory would not be viable in any test environment. As you may well already be aware of.

Thank you!! Now I see the download test packets payload with alphanumeric characters and no-transform directive in the http response.

It would make life easier if the date and time for each test record was separated. That is, when the csv file is loaded into a spreadsheet, the date and time loaded into separate columns. This would allow ready examination of connection performance by time. Otherwise, extremely valuable tool.

Very useful site. Good to hear time/date is being thought about.

I wanted to follow the steps provided on the Sticky, however It's from 2004 and has windows Vista? Get with the times Just looking for info on Win 10 x64

The speed test probably isn't the best tool for testing that. I think you need to be monitoring ping instead. Running TestMy.net's speed test on all of your machines on a regular basis will use up a lot of your network bandwidth and may cause problems depending on the pipe you have. I wouldn't do it. When I use the automatic speed test I do so to test fluctuations, not uptime. One thing you can do is use TMN's automatic response time testing. Every time you load a page on TMN's frontend (any page outside of the forums) it will test that machine's response time. This action only uses bytes of data. It's possible to call the page up and have it test automatically on an interval, I'll just have to dig into the program a little to find the commands. Currently if it encounters a timeout it isn't logged but in the future this will be reported in your RT database results. If you want to differentiate the machines you can select a unique identifier for each machine. If you have it set to test on a regular interval you be able to see holes in the results when you click details. As long as you're signed in your RT will be tested each time a page load occurs. You can do this even without using TMN to reload the page. Make a simple HTML meta refresh script, make it refresh a minimal page on TestMy.net. https://testmy.net/blank.php is already there for this purpose. Before starting the script just open a browser window, log in to TMN and select an identifier so you can tell which machine it is. The frequency shouldn't be set lower than about 15 seconds... the time it takes for the RT test to run and send the request to log it's result to the database. I hope this helps.

iperf is working pretty well, thanks for the pointers!

@Avneez , !dea is a large cellular provider in India, over 38 million subscribers. I would check upload speeds at different times of the day, as the network might be very congested when you are trying to upload something. Work to find the best times the content can be transferred.

Keep in mind the adjustments to the 2010 Open Internet Order now require ISPs to be transparent about any content blocking, throttling or paid prioritization. If you ever feel like that's happening and they aren't being transparent, please post it on the Internet for others to see. I don't expect ISPs to prioritize or de-prioritize TMN traffic any more than they did prior to 2015 when Net Neutrality was enacted. What I do expect to see is premium access to things that we shouldn't have to pay for... and extra billing for bandwidth for those services if you don't pay your ISP for a premium lane to those services. And THAT is BS. --- but I wouldn't expect to see that in 2018, they're just setting up the prerequisites. https://www.fcc.gov/restoring-internet-freedom

I'm checking on different companies in the area of Cottonwood AZ and someone said stay away from Hughes Network System. They will rip you off and won't let you out of the contract even if your speeds are so bad. Need advice. Thanks.

I pay Virgin Media £140 a month for TV VIP , including 2 X V6 boxes; land line and Broadband. My headline download speed is 300Mb but I always get well over that. With Speedtest.net I get 380Mb.

I pay 680 czk (= 32 usd = £24 )

It was working for me because it was pulling up old results. I may have my search field expanded to earlier dates on top of that.

Actually this is the exact reason for various testing sites, it is performing precisely as intended. Showing you where the issue is. There is a routing issue between your ISP, and the New York test server, likely due to the fact the ISP has changed their schema to IPV6 on your modem, and they are likely working out new routes for this block. Of course you could have gotten bumped to a different ISP level network for various other reasons as well. Therefore the tool is doing it's job, letting you know your connection is not proper from the ISP, to the New York area, yet proper to Florida. See, if a site has a CDN (Content Delivery Network) you will not notice pageload issues, considering you will pull from the available resource in line, on the other hand, if a site is not static (like a blog) and is in the New York area, you will load slowly, if at all. If the 'blog' is located in Florida, and not on any CDN, then you'll load quickly. Testmy.net is a tool not just to check for throughput, or what speeds you are paying for, but for geographical connectivity as well.

I have now tested this on an Eir 150Mbps FTTH connection which provides full IPv6 routing and get it looks like this issue is indeed IPv6 related:

As ridiculous as it sounds, most computer components such as the motherboard, PSU and certain expansion cards can survive a full dishwasher cycle. Obviously that's not how I recommend cleaning a PC. The following 8-year old video is an example. The internals were covered in cigarette smoke residue, so he decided to dismantle the components and put them through the dish washer, with the exception of the case, battery and the hard disk. I also enjoy watching videos on old hardware such as restoration videos. I have come ones where such hobbyists put components through a dishwasher cycle to clean extensive debris build-up from being left for 10+ years in storage such as a dusty shed. As long as no moisture remains when the PC is powered up, it should be fine. I generally clean my PC with an air duster, which is basically an aerosol can filled with propellant gas. Here in Ireland, humidity is usually on the high side, e.g. I run a dehumidifier to keep the indoor level below 60%. The only component I may wash is the keyboard. I have a Corsair Cherry Red Mechanical keyboard that I accidentally knocked a glass of cider on. I immediately unplugged it, tried wiping off what I could and let it dry. Once dry, many of the letters were sticking. When I started removing the keys to try cleaning below, I realised the mechanical switches were jamming. With what appeared to be a ruined keyboard that I only purchased a few months before the incident, I figured I'll try giving it a bath as we don't have a dishwasher. I partially filled a wide container with water from our dehumidifier (since it's effectively distilled), soaked the keyboard and pushed each key multiples times to force water through the switches. I left it to dry for about a week. It's fully functional again, all keys work and no sticking or other issue since.

Useless is a little strong. The results plots are very useful as is. You can select it and get dates and times. The fact that the x axis doesn't use fixed intervals isn't a big negative. It's a record, for as many days as you choose. I'm rarely interested in what happens every x minutes. Certainly an option to scale the x-axis to some user defined interval might be helpful. You do have the option to select individual TIP plots where the x-axis is scaled.