I have Cox Internet service in Southern California: "INTERNET ULTIMATE 300", which promises speeds up to 300 Mbps download, and 30 Mbps upload. I used Google web search with the phrase "internet speed test" to find several different speed test web pages. The top result was Google's own widget (using "M-Lab"). I also tried "speedtest.com" and the speed test page offered by Cox Communications. I used those three speed test sites, and many others. I tested in two different VirtualBox virtual machines: Windows XP SP3 and Ubuntu 17.10.1 (Linux). (I avoid web browsing directly in my host OS for most sites.) Because I did the tests in VMs, the speeds might be somewhat lower than if the tests were conducted in the host OS. The peak download speed I observed was 348 Mbps (~5:30 A.M. PDT, 40 miles to server), and the peak upload speed I observed was 35 Mbps (for many cities, up to 2,800 miles), with latency of 9-12 ms for servers within 40 miles. However, in late evening, download speeds within 40 miles were generally in the 160 to 200 Mbps range.
I think it is crucial to recognize the dramatic effect of time of day and the day of the week, and also the distance to the server which is being used for the internet speed test. Some speed test sites enable the user to select servers in different cities, and I think it is important to repeat tests with different cities to see just how important distance (and, implicitly, network hops) is for download speed, upload speed, latency, and jitter.
This modem enabled me to achieve download and upload speeds notably higher than the maximum speeds promised by Cox for my service tier (300 Mbps download, 30 Mbps upload). So, my ISP service tier is the limiting factor.
I feel compelled to mention that a few isolated internet speed test runs performed relatively poorly -- like a single instance of a download speed of 60 Mbps (slow!), ten seconds after a result of 160 Mbps, and twenty seconds before a result of 194 Mbps. These isolated slow test runs are uncommon and random; and, so, it is important to repeat tests several times to make sure that incidental connection issues, which have nothing to do with the cable modem or "last-mile" ISP link, are recognized.
I did many test runs from a Linux VM to a server within 40 miles, and all download speeds exceeded 319 Mbps, and routinely hit 346+ Mbps, around 5:30 A.M. local time, notably faster than the 300 Mbps promised by my ISP service tier.
One final note: Although disconnecting my previous cable modem, and connecting the NETGEAR CM1000, was sufficient to access the Cox "cable modem activation" web page, their web page encountered an error, and I was forced to call Cox on the phone to complete activation (with modem serial number and MAC address). It took several "reset" commands from Cox before they detected the modem, and a couple of power cycles, all during the phone call, before it worked. Now, I am very pleased with the speed, and it is tempting to upgrade to Cox's Gigablast tier.
| Brand | NETGEAR |
|---|---|
| Series | C7000 |
| Item model number | CM1000-1AZNAS |
| Item Weight | 2.14 pounds |
| Product Dimensions | 8.8 x 5.4 x 5.9 inches |
| Item Dimensions LxWxH | 8.8 x 5.4 x 5.9 inches |
| Color | Black |
| Voltage | 120240 Volts |
| Manufacturer | Netgear |
| ASIN | B0781VN7W5 |
| Is Discontinued By Manufacturer | No |
| Date First Available | December 15, 2017 |
