A 1 Gbps Ethernet reading describes your computer’s local link—not the whole internet path. If downloads stay around 100 Mbps, check the units, confirm the active adapter, inspect upstream links, and repeat a controlled comparison. That speed pattern alone does not identify a faulty cable or prove ISP throttling.[1][2]
1. Are you comparing Mbps with MB/s?
Check the exact unit beside the download number. A lowercase b means bits; an uppercase B means bytes. A byte contains eight bits. Valve documents this distinction because download applications and network measurements can display different units.[1]
Swipe or scroll horizontally to read all three columns.
| Displayed download | Equivalent in Mbps | What to conclude |
|---|---|---|
| 12 MB/s | 96 Mbps | Roughly the throughput range this guide investigates. |
| 100 MB/s | 800 Mbps | Not a 100 Mbps ceiling. |
| 100 Mbps | 100 Mbps | Already expressed in the same unit as most internet plans. |
Arithmetic examples using decimal MB and Mbps—not measurements. Applications displaying MiB/s use a different byte unit.
A 100 Mbps link also does not deliver its entire raw rate as file data: protocol overhead reduces usable throughput. A repeatable result around 94–95 Mbps is therefore compatible with a 100 Mbps segment, but it is a clue, not a unique signature. A configured limit or other constraint can produce a similar ceiling.[6]
2. What does the 1 Gbps Ethernet reading actually confirm?
It reports the negotiated link between that Ethernet adapter and the device immediately connected to it. If the cable goes to a switch, it is not a measurement of the switch’s connection to your router. Valve explicitly warns that the local network speed shown in connection properties is not your internet speed.[1]
On Windows 11, open Settings → Network & internet → Ethernet and look for the link-speed information. Labels can vary by Windows version and driver. Another option is the Ethernet connection’s Status window in Control Panel; Intel illustrates that local link-speed reading in its troubleshooting documentation.[3]
If you prefer PowerShell, this read-only command lists visible physical adapters and selected properties:
Get-NetAdapter -Physical |
Select-Object Name, InterfaceDescription, Status, LinkSpeedMicrosoft documents Get-NetAdapter for reading adapter properties. Identify the connected Ethernet adapter—not a disconnected port, Wi-Fi adapter, or virtual interface. The command does not run an internet speed test.[4]
For a local wired comparison, temporarily disconnect Wi-Fi if you can safely do so. Do not interrupt a remote support session or alter a managed work device without permission.
If the active Ethernet link actually reads 100 Mbps: investigate that local connection first. If it reads 1 Gbps, keep that observation and continue along the path. A negotiated gigabit link does not certify the cable as error-free under load.[3][6]
3. Where can another 100 Mbps link be hiding?
Write down the devices the traffic crosses. A simple home setup might be:
Computer → switch → router → modem or fiber ONT → ISP.
An ONT is the device that terminates a fiber connection. A combined gateway may handle several roles in one box. Your setup might also include wall jacks, a dock, powerline adapters, or a mesh satellite.
Inspect both port capability and current negotiated speed where the device exposes them. A port capable of gigabit operation may currently connect more slowly. Check the manual before interpreting link-light colors; their meaning varies by model.[6]
TP-Link’s U.S. documentation explains that a router’s 100 Mbps WAN port limits internet traffic through that port, regardless of its advertised Wi-Fi rate. The computer’s gigabit LAN connection cannot remove that upstream limit.[2]
4. What if the switch or WAN link reports only 100 Mbps?
First check whether the exact port supports gigabit operation. A label such as 10/100 differs from 10/100/1000. Check the specific hardware revision and port, not just the product’s headline wireless speed.[2]
If the relevant ports support gigabit but the connection reports 100 Mbps, compare that segment with a short, known-good Cat5e or Cat6 patch cable. Try another suitable port when practical. A new cable between your computer and switch does not test a different cable between the switch and router.[6]
Leave speed and duplex on the manufacturer-recommended automatic setting unless your administrator specifies otherwise. Intel specifically documents Auto Negotiation for its Killer adapters. Forcing a gigabit setting is not a repair for an unsupported port or damaged connection.[3][6]
Do not replace working Cat5e solely because of its category. Properly manufactured and installed Cat5e supports gigabit Ethernet. The condition of the cable and its terminations still matters.[6]
5. How can you test the path without resetting the network?
When the router is accessible, compare your usual wired path with a direct connection to a confirmed gigabit LAN port on that same router. This can bypass a switch, wall run, coupler, or other intermediate section.[5][6]
- Record the starting setup. Note the device, active adapter, local link speed, test service, server if shown, and result.
- Prepare a quiet comparison. Pause downloads or backups you recognize, where appropriate. Avoid disrupting other people’s work.
- Connect directly to the router’s LAN port. Use a short, known-good cable. Record the new local link speed.
- Repeat the same test. Keep the device, test service, and server selection consistent where possible.
- Restore the original route and repeat. If the earlier result returns, record the pattern before replacing anything.
If the direct path repeatedly performs better, the bypassed section deserves investigation. It does not identify which component failed. If performance does not change, the original switch is not automatically cleared; the same downstream limit may still constrain both paths.
6. Is the ceiling present across devices and destinations?
Once the physical path is documented, compare the scope of the problem. Repeat measurements under similar conditions rather than running multiple speed tests simultaneously.
- Same computer, another test destination: does the ceiling persist, or is one service different?
- Another capable computer, same wired path: does the pattern follow the path or only the original computer?
- Browser test versus file download: is the internet test also limited, or only the application?
VelocityVerify’s methodology states that its results depend on the route to its endpoint, the device, the local network, and current conditions. A result is useful context; it is not a direct reading of every Ethernet port or every download server.[7]
A faster Wi-Fi result can be informative, but Wi-Fi may bypass a wired switch, use another device, or change other conditions. Document those differences. For a broader comparison, see Is It Your Wi-Fi, Router, or ISP?
7. What if every visible link reports gigabit?
Gigabit links do not promise gigabit internet throughput. TP-Link distinguishes link rates, internet service, and router processing capability; Valve also separates local link speed from internet download speed.[2][1]
Review the settings you can inspect safely:
- Internet plan: confirm the current subscribed download tier, not the Ethernet adapter’s rating.
- Application: check the displayed unit, bandwidth limit, and whether it is downloading or performing other work.
- Router: look for configured device limits or traffic-management limits. A setting is a lead, not proof it is responsible.
- Computer: note an active VPN and recognized background transfers. Compare only changes you are permitted to make.
If a reversible limit is clearly misconfigured, record it before changing it and repeat the original comparison. Do not disable the firewall, antivirus, or workplace controls to chase a higher number. If you cannot inspect a setting safely, include it as an unanswered question for support.
If only one application remains slow while browser tests exceed the suspected ceiling, investigate that application and destination before treating the entire internet connection as capped. If website loading is the actual complaint, use the separate fast internet but slow website loading guide.
8. What does your comparison establish?
Use the pattern to choose a next check—not to declare a culprit.
Swipe or scroll horizontally to read all three columns.
| Repeatable observation | Most useful next check | What it does not prove |
|---|---|---|
| The application shows about 100 MB/s. | Convert its units before diagnosing a 100 Mbps limit. | That the connection is limited to 100 Mbps. |
| The computer links at 1 Gbps; a switch uplink reports 100 Mbps. | Check that uplink’s ports, cable, and terminations. | That the computer’s own cable caused it. |
| The router-LAN comparison improves; the original path restores the ceiling. | Reintroduce bypassed sections separately when practical. | That every bypassed component is faulty. |
| Several devices are limited through the same router, including direct LAN connections. | Review WAN negotiation, service tier, router settings, and provider evidence. | That the ISP is deliberately throttling traffic. |
| One device is limited; another exceeds the ceiling on the same path. | Compare adapter, software, workload, and application conditions. | That the first computer needs replacement. |
| The difference disappears on repetition. | Retain the notes and repeat later under comparable conditions. | That a permanent fix has been found. |
9. Frequently asked questions
Does 1 Gbps in Windows mean I have gigabit internet?
No. It describes the local link associated with that adapter. Your subscribed service and the rest of the path are separate constraints.[1][2]
Does a result around 94 Mbps prove a cable is bad?
No. It is compatible with a 100 Mbps segment, but does not reveal its location or cause. Port capability, negotiation, and configured limits need separate checks.[6]
Can Windows show 1 Gbps when the router’s internet port runs at 100 Mbps?
Yes. The computer’s local Ethernet link and the router’s WAN link are different connections. Inspect the WAN port’s capability and current negotiation separately.[2]
Should I force “1.0 Gbps Full Duplex”?
Not as a generic fix. Follow the adapter and network manufacturer’s guidance. Intel specifies Auto Negotiation for the Killer adapters covered by its support article.[3]
Can a browser speed test identify the slow Ethernet port?
Not by itself. It measures throughput to a test endpoint. You need device link information and controlled path comparisons to investigate a particular segment.[7]
Should I upgrade my internet plan first?
First confirm the plan and record the limiting observation. A higher tier will not remove a 100 Mbps physical segment in the path. If the plan itself is the limit, an upgrade is a separate purchasing decision.[2]
10. Your next step: record the first unresolved segment
Start with the units and the active adapter. At the next convenient maintenance window, perform one safe path comparison. Stop once you have a repeatable observation worth investigating.
Copy this note and fill it with actual readings:
Date, local time, and time zone:
Subscribed download tier:
Download value and exact unit:
Active Ethernet adapter and local link speed:
Usual path, including intermediate devices:
Switch uplink speed, if visible:
Router WAN speed, if visible:
Test service and server, if shown:
Original-path result:
Direct-to-router-LAN result:
Original-path result after restoring it:
Other device or destination result, if checked:
First unresolved segment or setting:Use the note to choose one next check. Send a repeatable shared-path problem to the router vendor, ISP, or network administrator with the relevant topology and readings. Remove passwords, public IP addresses, device identifiers, and unrelated private details from screenshots before sharing them.
The useful outcome is a narrower question: “This link reports gigabit; this other segment is unverified; this comparison changes the result.” That is a better starting point than resetting Windows and hoping the number changes.
Sources and scope
This is a researched diagnostic guide, not an original Ethernet benchmark or a report of the author’s personal tests. Numerical conversions are arithmetic examples. The network diagram is illustrative.
- Valve: Slow Downloads and Connection to Content Servers. Local link speed versus internet speed; bits and bytes. Publication date not stated.
- TP-Link: Router Internet Speed FAQs—Mbps, Link Speed, and Wi-Fi Explained. Updated June 10, 2026. Used for port capabilities, WAN limits, and distinct speed measurements.
- Intel: Ethernet Link Speed Capped at 100 Mbps. Last reviewed July 8, 2021. Used for the link-speed reading and Killer-specific Auto Negotiation guidance, not its broad cable-replacement claims.
- Microsoft Learn: Get-NetAdapter. Page dated May 14, 2025 in the retrieved source. Documents the read-only adapter query; not a test performed on a Windows computer for this article.
- TP-Link: Slow Network Speed When Connected to an Unmanaged Switch. Retrieved page dated July 16, 2026. Used for local-link and direct-router comparison principles; not a recommendation to cycle forced duplex settings.
- Computer Cable Store: Ethernet Stuck at 100 Mbps? How to Restore Gigabit Speed. Retrieved page dated July 18, 2026. Commercial technical guide used for Cat5e capability, overhead, physical-path comparisons, and interpretation limits.
- VelocityVerify: Testing Methodology and Editorial Integrity. Measurement endpoints and limits. Publication date not stated.
About the Author: Younes Kirat
Younes publishes and maintains VelocityVerify, focusing on broadband diagnostics, latency reduction, and consumer networking explainers. Read more about his work on his author profile.