This summary of the video was created by an AI. It might contain some inaccuracies.
00:00:00 – 00:22:25
In the video, Marc explores the selection and performance of various routers compatible with OpenWrt. He evaluates the D-Link DIR-2660, Linksys WRT3200ACM, and TP-Link Archer C7 based on factors like memory, CPU power, Ethernet, and Wi-Fi performance. Emphasizing the importance of hardware offloading for high-speed internet tasks, he demonstrates how different routers handle network traffic and tests both Ethernet and Wi-Fi performance. The Linksys router shines in routing performance without offloading, whereas the D-Link excels in hardware offloading but impacts features like Quality of Service and VPN capabilities.
Marc also provides comprehensive tips on flashing OpenWrt firmware, underlining the choice of the factory image for the initial setup and sysupgrade images for updates. He covers detailed steps for flashing D-Link devices using a Recovery GUI. Through practical performance tests, he shows variations in CPU usage and network speeds depending on the enabled offloading features.
For Wi-Fi, the D-Link DIR-2660 stands out, though it shows mixed results with multiple clients. The WRT3200ACM is noted for robust VPN performance but has limitations with one of its Wi-Fi adapters, and the Archer C7 is recognized for value in lower-speed scenarios. Marc wraps up by discussing the potential of upcoming MT7622 chips and his project involving custom router boards like the Banana Pi R64 and Mikrotik Routerboard, highlighting their flexibility and advanced features.
Overall, the video serves as a guide for choosing routers based on specific needs, performance criteria, and the practical use of OpenWrt firmware, encouraging viewers to check device compatibility and performance details on the OpenWrt website.
00:00:00
In this part of the video, Marc addresses a common question about recommending specific router hardware for OpenWrt. He notes that the answer depends on various factors and introduces a comparison of three routers: the D-Link DIR-2660, the Linksys WRT3200ACM, and the TP-Link Archer C7. He explains his selection criteria, which include personal experience with the TP-Link Archer C7 and a need for more powerful replacements, focusing on memory and CPU power.
Marc emphasizes that all three routers are well-supported by OpenWrt and will assess their Ethernet and Wi-Fi performance, ease of installation, technical specifications, price, potential for extensions, power consumption, and best use cases. He advises viewers to check the OpenWrt website for device compatibility and details on flashing OpenWrt, including debricking and reverting to stock firmware. He also highlights the OpenWrt firmware selector for easy firmware downloads.
00:03:00
In this part of the video, the presenter provides tips and steps for selecting the correct OpenWrt version and flashing it onto the router. They emphasize choosing the factory image for the initial OpenWrt installation and the sysupgrade image for subsequent upgrades. The flashing process is highlighted, particularly noting that most routers can be flashed via their web interface, but D-Link devices have a unique Recovery GUI accessible via a specific method involving the reset button and browsing to 192.168.0.1. Additionally, the presenter suggests checking the OpenWrt device page for compatibility and popularity statistics to guide purchasing decisions. Transitioning into performance testing, the presenter sets up an evaluation of Ethernet performance by testing switch performance and NAT/Routing performance using a D-Link router. They demonstrate accessing router status information through the Luci interface and detailing the test setup involving two connected test machines.
00:06:00
In this part of the video, the presenter examines the CPU usage during network performance tests on a D-Link router with a Mediatek CPU. Initially, the test shows low CPU usage and a stable 940 Mbps speed when transferring data over a switch. Moving one PC to the WAN zone reduces speed to 650 Mbps with one CPU fully utilized. The presenter then enables hardware and software offloading to see if it reduces CPU load. Enabling these offloads yields approximately 935 Mbps with 0% CPU utilization, confirming effective hardware offloading on the D-Link router. Disabling hardware offloading but keeping software offloading results in unchanged Ethernet performance with the load distributed across two CPUs. Next, the presenter tests routing without NAT by defining a second interface and observes different performance metrics with and without offloading. Again, hardware offloading achieves near Gigabit speeds with minimal CPU usage. Finally, the presenter begins evaluating a Linksys router, noting its more powerful ARMv7 processor and additional RAM.
00:09:00
In this part of the video, the speaker conducts performance tests on various routers. The switch test shows 940 Mbps with 0% CPU usage for the D-Link router. With NAT enabled, the average speed is 933 Mbps at 45% CPU, slightly reduced to 35-40% with software offloading, indicating minimal change. The concept of flow offloading is explained, noting its limitation with Smart Queue Management. The Archer C7 (v2) is tested next, showing 940 Mbps for switch performance with low CPU. However, with NAT, it achieves 280 Mbps with maxed CPU. Software offloading significantly improves the speed to 935 Mbps but still maxes out the CPU. The summary reveals that all devices manage Gigabit Ethernet within the LAN without CPU load, but only the Linksys can route Gigabit Ethernet without offloading. The D-Link supports hardware offloading, impacting Quality of Service and related features like VPN client or server usage that require a robust CPU.
00:12:00
In this part of the video, the speaker discusses the impact of internet speed on device performance, emphasizing the importance of cooling for routers handling high CPU loads, especially when dealing with Quality of Service (QoS) on fast internet connections. The segment then shifts to testing Wi-Fi performance using a 5 GHz network with WPA2-Preshared key encryption. The test involves running two instances of iperf3 on a wired PC, with an iPhone as one Wi-Fi client and another device joining later.
For the D-Link router, initial tests show a single client achieving 600 Mbits with about 60-75% CPU load. Adding a second client results in approximately 300 Mbits for the first and 250 Mbits for the second, which was less than expected despite the router’s 4 antennas and the clients’ 2 antennas each. The speaker speculates on the potential for better performance with MU-MIMO or SU-MIMO but lacks the necessary equipment for further testing. The takeaway is that the connection remains stable and sufficiently fast for practical use. The segment concludes with a brief reference to testing a Linksys router in a similar setup.
00:15:00
In this segment, the video discusses the performance and configuration of the WRT3200ACM router when used as an all-in-one device, serving as both a router and a Wi-Fi access point. The speaker highlights that while the router has three Wi-Fi adapters, the third one (adapter 8887) is not very useful due to its limitations in power and capacity, mainly intended for weather radar detection. They recommend turning off this adapter for more stable performance. The speaker also touches on issues with the Marvell chipset in achieving fast roaming on OpenWrt 21, advising to use the stable OpenWrt 19.7 release for better reliability. Key configuration tips include ensuring the country code or regulatory domain of the two main radios is set to the same value. Additionally, attempts to replace Wi-Fi hardware were futile since components are soldered to the PCB. Lastly, the Archer C7 router is mentioned as having good support with the Atheros ath10k driver for hardware versions above v1.
00:18:00
In this part of the video, the speaker discusses the performance and features of various Wi-Fi devices. They highlight that the Archer C7 router, despite being old and having only 3×3 MIMO antennas, offers good value for lower-speed internet connections (up to 150 Mbit) due to its price point. The D-Link DIR-2660 is identified as the best option for Wi-Fi performance, while the WRT3200ACM is recommended for strong VPN performance without Wi-Fi needs. The prices of these devices are mentioned, along with specific considerations such as POE compatibility, power consumption, and USB ports. The speaker emphasizes the importance of hardware over the brand, suggesting checking OpenWrt site statistics for more information.
00:21:00
In this segment, the speaker discusses the potential of MT7622 chips to become popular and explains their decision to choose the D-Link DIR-2660 router. The speaker also mentions other compatible models like the 1960, 2640, and 3060, suggesting there are opportunities to find these routers affordably. They share personal success in buying a 3060 model second-hand on eBay for $50 and express satisfaction with its Wi-Fi performance.
Furthermore, the speaker reveals they are working on assembling a custom router using different router boards, specifically the Banana Pi R64, Mikrotik Routerboard RBM33G, and PC Engines APU D4D board, highlighting their features such as mpcie slots for Wi-Fi and SIM card slots for 4G LTE hardware. The segment concludes with a request for viewers to subscribe and set reminders for the upcoming episode and thanks the audience for their support.