Jump to content

Cox 150/25 performing at 50/25


CA3LE

Recommended Posts

A week ago Cox Cox upgraded my internet to 150mbps download and 25mbps upload.

Your download speed test results are still only 50mbps after many attempts over several days, Other speedtest sites do reflect the new download speed.

Your upload speed test does refect the new speed of 25mbps!!

What gives?

Cheers

Marty

Hi Marty,

Long email... hopefully you take time and read it. If you do, you'll understand speed tests better and how the results are all in how you look at it.

First thing to know is that TestMy.net works unlike any other speed test. Especially at extreme speeds. Once you get into higher speeds TMN becomes more than an Internet speed test... it becomes a computer benchmark that can passively detect countless issues that slow you down. Seriously, I've seen it all. From faulty modems, drivers, network cards, hard drives --- but most often (for Windows users) the TCP stack isn't optimized for your new extreme speed. For that I recommend TCP Optimizer.

Another thing to check for is routing. Here's what I do to thoroughly test a connection. In the next version I hope to do this all with one click.

  • Run TestMy.net on default settings, save yourself time.. a simple download test will do. For the speed of your 150 Mbps connection 50 - 100 MB tests should suffice. Personally, I manually select on each of these steps so that the benchmarks have less variables between them. Apples to apples. I also will select smaller sizes like 25MB or use the express button between performance adjustments... saves time and bandwidth. Then, when I feel like I've got it right I'll run a long 200 MB test (or what the connection I'm on can manage over 10-30 seconds). If you ask me, since early on one of TMNs advantage over other speed tests has been the manual size selection and large, long (but not overly long) tests.

    My result for classic linear 110.8 Mbps

  • Run a multithread speed test, you'll see a "MULTI" link to this at the top of the page... use that link to disable/enable and know that it's on or off. I manually select the same size I selected in the linear speed test above. You may wonder how to do this because there is no size selection on the multithread speed test page. Once you've enabled multithreading just navigate to the download test page and manually select the size. It's not a different test... it's a transformation of the original. For your first test select only the first checkbox, this will test your connection across the US. (TestMy.net is the only speed test that's able to do this). If your browser has a status bar during this test you may see one of the addresses hanging, you can take that address, type it in your browser (it will switcher to that server) and then test again directly to that server using the linear method... if that happens just run a SmarTest because the size it will need to select will likely be smaller because the speed is degraded to that locale.

    My result for multithread 108.8 Mbps

  • Run similar tests on all the official mirrors, the only variable between the official mirrors is the physical location... differences between them can help you better understand your routing and your performance across the Internet.

    My other results -- Dallas (retested at 100MB) 100.6 Mbps, DC 56.7 Mbps, Seattle 79.1 Mbps, San Jose 79.1 Mbps, Europe 25.4 Mbps, Asia 13.8 Mbps (I consider that a great result for a 105 Mbps rated connection. Yeah, I'm not getting that speed everywhere but it performs great for me and I'm happy. And if I was on a slower computer the results would be slower than my connections max because the computer can't process as fast as the connection can deliver... you have to take things like that into account.)

  • Now, take your best linear result from above and select multi again... tell the mulithread test to concentrate on that server (sorry, this does not work with Dallas, if that's your best select the 2nd best) by selecting only the one check box... make sure you hit save then navigate to the download test and again... run a test the same size. I'd bet that will be your best result. But does that make any of the other results wrong? ... no. As you can see clearly by using my various methods, it can be viewed in many ways. My competition wants you to see only the best case scenario... I don't think that helps anyone, it hinders . I built TestMy.net first for myself and continue to develop for my own use. Other speed tests for my purpose are useless.

    couple examples My multithread result to San Jose 103.6 Mbps - multithread result to Asia 53.4 Mbps

So I was only able to pull 79.1 Mbps to San Jose using a single thread but if I multithread to the same locale I max out my connection. With TCP tuning I could probably get that single thread number up but like I said.. I'm happy. Nearly 4X speed when I multithread to Asia. Multithreading makes the result look better... ideally, if your connection is good you should be able to get the same speed either way. This is one way other speed tests make your connection seem better than it is, ISPs love those tests. Makes customers feel good. If you can't achieve similar numbers you should ask yourself, "If others can, why can't I?".

If you perform much better on multithread than you do on the default then that's an indication that you have an issue. The issue may or may not be related to your Internet. I've been amazed by all of the issues that can affect my results... trust the results, the number shown is the way it is for a reason.

Other speeds tests try to eliminate the variables you want to test. Important factors that really affect the results. That's why your results differ, I do it a different way. Up to you which you think is right... but you may want to take a close look at the benchmark you use. Think about if the person who designed it stands to benefit from altering your results or tweaking the process to give you the better results. .... TestMy.net doesn't. A comment from someone able to regularly pull over 900 Mbps on my tests (he has powerful, optimized servers) put TMN under the microscope, I was flattered to read, "I can say with 100% certainty that you will not find a testing methodology more accurate than testmy.net" --- and if you have a program or operating system that lets you read the network interface and see your transfer rate I encourage you to use it and watch it as the tests run. Note: TiP data in your results will differ slightly because local programs poll on a per second basis and the TiP is calculated based on transfer percent. Using that in conjunction with TMN can provide better insight into what's happening during the test.

Having said ALL of that. I had Cox 150/25 in Phoenix up until I moved to Colorado in October. I NEVER saw 150 Mbps, I was only getting little over 100 Mbps. (Before I got that package they had always over delivered.) Speed tests, usenet, bittorrent... I couldn't squeeze any more out of it. I taxed it so hard that it caused pandora to skip, the connection had no more to give. Guess what, the numbers I read off my network interface directly correlated with the results on TestMy.net. As you would expect, I was never able to sustain an upload or download faster than my test results. My new connection is rated lower than my old Cox connection but performs better. It's a little bit more consistent. But to be honest, I'm impressed by both Cox and Comcast's networks. Cox Communications, I'm a little bias because I worked for them for 5 years and I know how awesome the company is internally and how meticulous the high level back end techs are. Very organized, very clean and the company isn't afraid to invest heavily in the future of their technology.

Hopefully this helps you, let me know how it goes. As if this wasn't already enough, read more about why results differ. There is much more to this than the average person considers. Help me by spreading the word.

- Kind Regards, Happy Testing!

- Damon - TestMy.net

Link to comment
Share on other sites

×
×
  • Create New...