Why internet speeds differ by PCs?

[zamar29]

I got a Gigabit sub, but my PC can only download at 250 down & 30 up Mbps over a Cat5e cable connection from modem direct to PC 1Gb LAN port. The cable was supplied with the modem. The provider's tech came to check the speeds, and on his laptop same cable direct connection gave speeds 650 & 30 Mbps to the 1st hope (provider's speed test server). Why such huge difference?

My PC is running Win 10 64-bit 1703 latest built, the Gigabit card is 2-port PCI-E server card. About the same speed I'm getting through a standard Gigabit port on the PC Mobo. I tried to play with various settings of both adapters, but no changes notably affected the result. Both adapters run default Windows drivers, there is also A/V software running on PC among other, but overall CPU load is not that high during browser speed tests (25% browser, and 71% total, while about 30% total after the test stopped with multiple browser tabs still open and usual other workload). I tried to copy large files btw 2 PCs on LAN via the modem & router Hitron CODA-4582, and the same PC adapters gave file copy throughput 600 Mbps, probably limited only by SATA1 drive read & write speed in the laptop. So it seems like only internet speeds are slow, but LAN speed is much faster using the same equipment.

I repeated the browser speed tests after switching A/V and other PC loads off. The result was 315 & 30 Mbps with browser 21% and total 60% CPU load during the test, and RAM at 40% total used. So its not that different compare to testing with usual PC load. Browser cache is saved on SSD drive, but I assume the test cache is small and kept in RAM.

My question is, how I can troubleshoot what causes such slow internet speeds on my particular PC compare to the tech's laptop? May be resetting TCP stock will help, or some Reg settings? What are the usual culprits? What is different btw loading test file from the web during speed test and copying a file from one PC to another via LAN that causes such speed drop? Can you suggest any good links for relevant reading?

 

[jec6613]

The tech's laptop may be using 802.1x or even just basic MAC filtering to enable him to have uncapped speeds. What speed are you paying for from your provider? If you're getting that, you're fine.

 

[zamar29]

The tech's laptop may be using 802.1x or even just basic MAC filtering to enable him to have uncapped speeds. What speed are you paying for from your provider? If you're getting that, you're fine.

I'm paying for Gigabit sub 1000 down & 30 up Mbps, and getting 300 & 30. The tech's laptop was hooked to the same modem-router, so could you explain in detail how it can get 650 & 30?

Also, are you saying that the tech's laptop hooked to an Ethernet port of the cable modem-router can get higher speeds above the limits set in the modem config file?

 

[jec6613]

Yes, a tech's laptop could authenticate to use different channels or configurations. With enough time and effort, you can do pretty much anything to that sort of setup.

If you're paying for 1000 down and 30 up, you're probably just running into the problem that the speedtest site that you're using requires a certain minimum upstream to measure. 10:1 is pretty normal in TCP type traffic, where you can only hit 10x your upload as your max download.

Have you tried the one over at dslreports? I've found that's the most accurate in terms of actual usable speed.

 

[TonyH]

Is the laptop and your PC about same vintage?

 

[zamar29]

@TonyH
No, laptop is older with weaker CPU, and the provider's and DSLR tests show about the same wired speed test result for it: 150 & 30 with A/V switched off. Nonetheless it shows 600 Mbps download speed at copying files to its internal Sata 1 disk over LAN. Why is that?

@jec6613
I measured the speed at provider's test server (I doubt they are linking downstream and upstream speeds, since this config is rather standard for their Gig customers) and DSLR. The provider's results center around 300 / 30 all the time, while DSLR around 180 / 30 daytime, and 250 / 30 deep at night. Your feedback raises the doubt why modem user GUI usually hides the config Down and Up limits - they actually can be set 2-3 times lower than the subscriber is paying for, and the provider just says "what you expect, its up to 1 Gbps in the Agreement, we never seen it matching closely", which is a total BS.

Another issue is non-optimal and overloaded routing setup in the headend for consumer segment. Meaning, the provider's 1st hope speed server shows high speeds for subscribers like 600 / 30, but actual internet surfing speeds reflected by DSLR test are much slower. While file download speeds from clouds may be faster. It looks like a marketing gimmick, a fraud, since they solicit 1 Gbps subs without giving users pull a dedicated internet channel matching that speed, so their traffic is channeled together through busy lower priced lines at the provider's 1st hub with all other traffic from customers with lower 30 / 5 subs.

The question is, apart from fixing what seems to be the modem config set erroneous speed limits, is their a way to make provider reroute the surfing traffic differently, so DSLR results would reflect the sub speeds paid for? Can DSLR results be an argument in filing a CCTS complaint in Canada?

 

[Trikein]

Who is your ISP? You need very specific hardware, including a PC with the latest CPU and a SSD.

https://www.xfinity.com/support/internet/requirements-to-run-xfinity-internet-speeds-over-1-gbps/

"Nonetheless it shows 600 Mbps download speed at copying files to its internal Sata 1 disk over LAN. Why is that?"

Because your AV is slowing it down as it scans it. Also, your downloading locally will always be faster then downloading from the internet. The local PC is in your house, while the test server could be hundreds if not thousands of miles away.