Why is the speed test for gigabit broadband WiFi only 600 Mbps? Bandwidth, negotiation speed, and actual speed are the three main accounts, and I spent the whole night checking them

my cousin's misconception lies here: he thinks 1201 is " WiFi speeds, then 300 Mbps broadband should be maxed out. In reality, 1201 is just the theoretical upper limit between the phone and router, regardless of the bandwidth or bandwidth, and it will never be reached.

First ledger: Is bandwidth the lane, or is it shared by the whole family

Bandwidth is the easiest to understand, but two details are often overlooked. First, unit conversion: the operating trademark is measured in megabits (Mbps), while the speed test software and download display in megabytes (MB/s) should be divided by 8. The theoretical download limit for 300Mbps broadband is 37.5MB/s, and for gigabit bandwidth, 125MB/s. how many megabytes are enough for your home I calculated the math in that post. The difference between movies and gigabits at 300 megapixels feels much less than the price difference in daily experience.

second is the word "sharing." The lane width is like a pool where all the family's devices walk together: the TV occupies 25 megabytes, the phone playing videos takes up several megabytes, someone backing up photos takes up about ten megabytes, and the speed test was just a ladle from the pool. There's another lesser-known thing: the bandwidth referred to in the operating trademark refers to the downlink to your fiber modem, while the uplink is another ledger (300 Mbps broadband uplink usually only has 30 to 50 Mbps). Whether NAS remote or video call cards lag depends on the uplink, not the downlink.

in this bandwidth ledger, a small portion of the "not running the rated value" is normal: the gigabit bandwidth connected directly to the optical modem is measured at around 940 Mbps, and the remaining 6% is protocol overhead, which is normal loss. That's how I remember my home's Range: the setup that ran the entire topology for two years, wired 940 capped out, but I've never seen a 1250.

second notebook: The negotiated speed is the handshake price, not the speed

The 1201 on my cousin's phone is the standard level negotiated when two spatial streams and 80MHz bandwidth on a WiFi6 phone. I've broken down this numerical algorithm in the 2x2 and 4x4 spatial stream articles: bandwidth multiplied by spatial flow multiplied by encoding efficiency. 867 is WiFi 5's 2x2 80MHz, 1201 is WiFi6's 2x2 80MHz, and 2402 is the 160MHz mode. whether the bandwidth is set to 20 or 160 directly determines which price you are in and your hand price is set.

you want to check this number yourself, Android has a "Connection Speed" section in the WiFi details; In Windows, open the command line and type netsh wlan show interfaces; the two-line send/receive rate is the negotiated price; Apple won't let you see it, so you can only rewind it through speed testing software. By the way, the speed shown in the router's background device list is also a negotiated value, so don't complain about it as the device's real-time speed.

but handshake prices have three pitfalls. First, focus on the shortcomings: routers 4x4, phones 2x2. If you negotiate to use 2x2, then a 2402 router connected to a 2x2 phone is the 1201. Second, it's dynamic: when the phone moves from the living room to the bedroom, the signal drops by one level, the negotiation rate is recalculated on the spot, and 1201 might drop to 866 and then to 288, and all you see is the handshake price in "that one second." Third, it has nothing to do with broadband: even if you unplug the cable, your phone and router can still negotiate 1201—no matter how wide the road between the two village entrances, the national highway outside the village only has two lanes, so you can't run on the highway with speed limits.

blaming broadband for negotiated rates is the wrong approach. It only answers "How much WiFi can theoretically be installed," not "How much can it actually run."

third ledger: measured take-home wages, with taxes deducted at every level

from negotiation speed to measured speed, there are at least two layers of taxes in between. The first step is WiFi air port efficiency: WiFi is a half-duplex walkie-talkie, and managing frames, beacons, retransmission, and waiting for devices to take turns talking all require space space. The industry consensus is that the utilization rate of the space is 50% to 70%, which is the 1201 negotiation. The unobstructed room in the same house measured around 650 Mbps is completely normal—this is already a 54% discount, not a fault.

second is environment and concurrency. Separated by a load-bearing wall, we negotiated gear drops and tested them to drop along with them; The more devices you have at home, the more old devices you talk about slowly, and the market share of new phones is diluted; On the 2.4G band, neighbors have a dozen or so networks crowded together, and the physical gap between 2.4G and 5G is obvious. Older devices negotiate 288 Mbps on 2.4G, and actual measured tens of megawatts is the norm.

This chart is my home test log—every row is in its own layer, which is 'normal.' My cousin's 187 Mbps is the same: 300 Mbps broadband, phones separated by a wall in the bedroom, negotiations dropped from 1201 to 866, then 60% of the calls were empty, 187 Mbps matched perfectly, no one lied to him.

there's another incident that easily misjudges the equipment. By the way, speed testing software itself also has a discrepancy in diameter. The web version speed test is limited by the browser's single-threaded process, so the numbers are generally lower than those in the app; The same app tested at different times shows varying levels of exit congestion; a 10% lower peak than morning is normal for the price to fluctuate. When reconciling, I always use the same app and test three times in each time period to get the median. Only then can the numbers be comparable. If you compare the web version's 940 yuan in the morning to the app's 680 yuan at night, the score will never be fair.

reconciliation order: first lock the variables, then stack them

if the three numbers don't match, the biggest mistake is blaming the router or calling for repairs right away. My order was to plug in the network cable first: wired straight to the optical modem, network port speed , wired 940, broadband layer is fine, slow to stay indoors; Wired is only 93 Mbps. First, check if you have plugged in the optical modem's 100 Mbps port or IPTV port. how do you distinguish between gigabit and 100 Mbps ports I've encountered this pitfall. Now let's see if the internal network cable is quadruple-core.

wired is normal but WiFi is slow, then it's time for the wireless layer: stand one meter from the router and test once, achieving the "best wireless layer" result. If it can reach 600 or 700 kilometers from one meter, it means the router is fine. Slowness is a matter of distance, walls, and channels. Relocation, channel switching, and adding a Mesh node if necessary; Even at a meter away, it's only one or two hundred. Check if the router is in power-saving mode and if the phone is stuck to 2.4G. If the mismatch is "fast and slow," not "always slow," then it's normal for peak hours to drop during the day, which is congestion at shared exits, which is broadband issues. the tiered approach from Gigabit speed checks and game latency more effective than switching devices.

conclusion after reconciliation: Don't chase the three numbers equally

After calculating for my cousin, he actually felt more at ease. A reasonable expectation is: gigabit broadband, cable 940; WiFi 6 phones face to face 600 to 900; Fold it in half over a wall; 2.4G with tens of megawatts is no big deal. Any floor that falls within this range means the equipment is intact and the money is well spent.

really want to improve, the goal isn't to "make WiFi run all gigabits"—unless you go for hard metrics like 2x2 160MHz or 2.5G ports, no one can avoid the air port efficiency tax. Cost-performance from high to low: first lock your phone to the 5G band (zero cost), then move the router out of the weak current box (zero cost), and only then replace the device and add nodes. If the order is reversed, the money spent may not change the numbers. Only after reconciling the accounts do you act without wasting money.