Use a more reliable speed test site. The correct answer is you can’t, after some overhead a gigabit port caps out around 940mbps. Anything being reported over that is a bad measurement or a bug.
Well, while what you are saying is not false, it’s not the way some isp work. Comcrap, for example, you have a “1 gig” package, but in reality, the package is provisioned for 1.25g, so getting 1.2 is actually very normal. Not saying fast.com is accurate or the op has comcast, just providing a example.
The 1 gig wan interface on his router will not run at 1.2gigs. It does not matter if he is over provisioned on the ISP side, that is not the cause of this speedtest saying 1.2.
I respectfully disagree with this. Not all ISPs are scumbags. Every service I build, I overbuild for this exact reason. For example, if I built you a 500 Mbps service today, I would “fudge” your service to 500240, in an attempt to make sure you’re getting what you pay for. I understand I’m the exception and not the rule, but the “good guys” are still out here doing what we can.
At 500mbps it’s not an issue, you get 500mbps. But lots of ISP’s are claiming gigabit and then only giving routers or ONT’s with gigabit ports, gigabit Ethernet can only do ~940mbps after overheads
His router has a 1 gig wan interface. This has nothing to do with over-provisioning.
This is a gross oversimplification and, resultantly, an overall false statement.
Lol, yeah sure buddy.
Good one. Couldn’t have lived without this input….
Please explain.
Ignoring the use of fiber here, the IEEE 802.3ab defines Gigabit Ethernet over UTP at 1000Base-T over 100 meters. But, this standard was created around the use of CAT-5 cabling. As tech has progressed, it became easily possible to achieve faster data transfer rates with what is still considered “Gigabit Capable” ports and cables. Just because something is rated as Gigabit, does not in any way mean that is the max it is capable of. MystiCom was able to use CAT-5 for 10Gb/s transfer back in 2002. Notwithstanding that you completely omitted any other hardware capabilities/limitations such as backplanes or line cards, you simply saying “you can’t” and that gigabit ports “cap out around 940mbps” is simply incorrect. What you seem to be referring to are the data rates that ISP’s offer in their brochures where they throw up an asterisk and say that over head may reduce their gigabit plan to 940mbps max. So yes, while the above is not incredibly in-depth, you way oversimplified what gigabit entails and what impacts that performance. And telling someone that “anything reported over that is a bad measurement or a bug” is just incorrect information.
I think you need to remember we are in /HomeNetworking and not /Networking. I am very aware that spec does not dictate hardware capability. I am very aware that you can run 10gig over cat 5 (and other various configs beyond the “spec”) in certain cases. I am very aware that IPSs overprovision. What I am NOT aware of is anyone who has gone to best buy, purchased a modem and wifi-router with gigabit NICs, gone home to their gigabit ISP plan, plugged and played, and magically gotten 1.2Gbps to their client device. Simplifying for the purposes of home networking to say that given a standard Gig NIC > Cat whatever cable > Gig NIC negotiated at 1000Base-T will run ~940Mbps is perfectly valid, especially in the context of this post where the user has one data point showing 1.2Gbps, all other sites showing ~940Mbps, and his set up is what I described above. You are reading way to far into my (I’ll agree here) aggressive use of the word “cant” and trying to apply it en masse to all possible situations. In the context of home networking, the OPs issue, and anyone else experiencing similar things, this is quite clearly a bad measurement or a bug. Also, I am not referring to data rates in the brochure. Go run some Iperf tests on the setup I described above and let me know what you get. Verizon does clearly say 940, but I suspect this is because they know when they drop off their ONT with a Gig interface, they know dam well its expected to be ~940Mbps on the top end. They didn’t pull this number out of their ass.
Some of the ISPs over provision the circuits to avoid issues with customers running speed tests and getting a lower speed than they are paying for. I’m paying for 500mb internet but actually get close to 600.
Source I worked for an ISP.
ISPs typically provision their modems to provide a little more speed than what is actually listed. Not always true. When I worked at spectrum, 200mbps connections would show over 200 when I ran tests. Selecting different servers would yield wildly different results.
I have FiOS and mine reads faster than the speed I pay for by about 100mbps both ways. The tech said they deliberately over-provision so that you’ll get at least the advertised speed on wireless. I’m not sure who your provider is so this may not apply to you.
it’s the fact that he has a 1 gig router. with overhead only like 940 Mbps is possible
That can’t be right because I’ve seen higher speeds than that during downloads with a download manager on my 1GBs infrastructure https://imgur.com/a/aIaVpi8
9k. somehow you have less overhead. or that counts the overhead into the total transfer speed. you still have less then 1 gbps
That’s windows task manager measuring the bandwidth at the network card. Not sure how or why it would incorporate switching overhead.
His router is rated for 1 gig, meaning it’ll do minimum that amount.
It’s not like they’re going to set a software rate limit to make sure he doesn’t go over that amount in ideal circumstances.
Unless he owns a Cisco router and hasn’t purchased a performance license.
“His router is rated for 1gig, meaning it’ll do minimum that amount”
Uh, No? What are you on about lol.
Your ping seems to be a little high, are you on coax? What time of day was this test run? Are you running this test from the modem or internal network?
Do you have the Pro 6E mesh? That system has 2.5GbE ports according to their website. Either way Fast tends to be a little generous with the results they show, did you try nperf or ookla to see if you get the same result?
Fast, while loading quicker than speedtest, can hitch occasionally and give you a higher speed than you’re actually getting. It might read 800mb/s one moment, then realize it was supposed to be 1gb/s, and then compensate by giving false numbers. That number approximated your gig service, you’re fine.
It could be AT&T. They overprovision the circuits. A 1Gbps on Xpon will test 1300.
If the router has 1 gbps ethernet port, the max it will provide is about 940mbps. It’s the PHY limit of a 1gbps NIC.
Math bad.
It’s definitely over provision to be sure you don’t complain lol. 😂
They do it where I am here too
You are right, just kidding. Technically, it is possible to surpass 1gbps if there is compression during transmission. Steam actually does that. It’s usually compress during transmission and it’s decompressed on the user end. I understand it’s not related to this. But it is a known fact.
In this instance yes, its inaccuracies, would recommend using another speed test like Speedtest.net
Some testing software in my experience can give you a higher speed than what you are getting and vice/versa.
For example the Xbox internal test in settings can give me a speed of +600mbps whereas a Google test on a wired connection can give me 300mbps on a 550mbps connection
Additionally, some ISP’s may over provision by whatever % to cover overhead and other things.
This is Exactly what my ISP does. So I don’t know why you are being downvoted. I pay for symmetrical 1Gb. I receive, as tested through speedtest.net, 1060Mbps down and 1040Mbps up. In order to receive those speeds, I had to test from a computer with an ethernet interface faster than 1Gbps. And of course the router would also need to support the higher bandwidth link. My Mac Studio has a 10Gb interface; and my router has a 2.5Gb interface. If you have a router with WiFi 6 or 6e capability, you could also exceed 1Gb with that (providing the device you are testing with also supports it).
Yeah, I’m confused why he’s being down voted. I do find Fast dot com to be inaccurate and likely that’s why he’s seeing this, but I work for an ISP and every customer is overprovisioned. It’s more common than not if testing with a capable device, especially hardwired, they’re getting more than what they pay for.
That is correct. E.g. a 500 Mbps subscription is configured as e.g. 540 Mbps, so that overhead will not impact a speedtest.
But with a 1Gbps connection, setting 1.1Gbps might not be possible as the modems/routers can have a 1Gbps port, so there is a “hard limit” on 1 Gbps. Because of overhead, you will typically only get around 950 Mbps on a 1 Gbps connection.
As for Fast, I have seen the same thing as OP. I can get 1.1Gbps on a 1Gbps connection that normally is “limited” to 950-isj Mbps on other speedtests. The modem internally supports more than 1 Gbps, but is limited by the Ethernet-port.
1 Gbps is computationally 1.24416.
Except you know overhead is a thing
Idk why that comment is getting upvoted so much.
Fast.com uses a crude way to determine speed. It’s obviously some flawed math formula since I’ve gotten some wild speeds before (like over 6-7Gbps) on a 2Gbps connection. Whatever it is is obviously inaccurate.
This is it
With 1 Gbps, most ISPs and router manufacturers mean 1000 Megabits per second. And this site also shows the speed in bits per second. So, the speed shouldn’t really be more than 1000 Mbps or 1 Gbps because their hardware isn’t capable of that, even if their ISP is providing some extra speed (which sometimes they do, especially with servers like Netflix).
This is the typical wrong speed reported by Fast.com and that’s why people shouldn’t use it to measure how fast their connection is, unless they are trying to verify if their connection sucks with Netflix servers and/or video streaming.
Care to explain?
I think they’re getting at the difference between using powers of 10 vs powers of 2.
So 1 kb (kilobyte) = 10^3 = 1,000 bytes, but since computers tend to use powers of 2, sometimes they might use 1 kb (kibibyte) = 2^10 = 1024 bytes
Strictly speaking, kilo always means 10^3, and kibi always means 2^10 so they’re not equal. But in common parlance, the difference barely matters so it is common to use kilo- even when kibi- is strictly correct.
The same holds true for bigger prefixes like giga (10^9) and gibi (2^30).
So I think that guy was getting at those differences even though the number he said makes no sense even if you account for the giga/gibi difference. So maybe I wrote all this for no reason lol
You are still appreciated.
I feel like you’re referring to the 1000 vs 1024 issue, but to get that kind of factor you’d need to be talking on the order of 10^27 bits per second since 2^90 / 10^27 = 1.2379
Thankyou for being able to do math. Seriously.
Except its not…and even if you’re trying to talk in Gibibytes, its still not.
Fast.com is really inaccurate in my experience
Same here. I have 500Mbps service and speedtest more or less agrees. Fast consistently says it’s running at 700Mbps.
I feel like thats a bit ironic because it was made because speedtest wasn’t giving accurate numbers for a time because the site would optimize for certain ISPs like comcast instead of giving realistic speeds
Fun fact: if your company throttles video bandwidth fast.com gets that. We throttle at 2mbs and fast shows exactly that.
I do belive that was the initial reason for the site, it was during the whole ISPs getting caught lying and throttling hence why it’s a Netflix owned site.
Most ISPs will add a buffer to their provisioning.
Ie, 100 Mbps is provisioned at 125, 500, its 550, etc.
Source: work at ISP
its a bug like others said. sometimes i get 800kbps-1mbps on T-Mobile 2G using fast.com which is impossible. most of the time its accurate at 160-190kbps though.