Check your settings and connections. Ensure the device is properly paired and there are no interference issues.
Check your settings and connections. Ensure the device is properly paired and there are no interference issues.
Immediately after sharing that post, an idea struck me. It wasn’t a highly precise measurement, but I configured iperf3 on two distinct ports—first on the NAS and then on my primary machine. I also tested it using a secondary system with only a gigabit port, running both simultaneously. Initially, I set up Computer 1 (equipped with a 2.5G NIC) to run two server instances of iperf3 in client mode on Computer 2 (same NIC) and Computer 3 (with a 1G NIC). Later, I reversed the roles, making Computer 2 the server while Computer 1 and 3 acted as clients. The results showed that my main computer could handle a full 2.5G connection, achieving around 1.55 Gbps from the NAS while the secondary system managed about 800 Mbps, totaling roughly 2.35 Gbps. On the NAS side, I only reached speeds similar to what I was previously experiencing. Running in reverse mode, both machines consistently hit about 800 Mbps. Unless there’s an error in my setup, this suggests the new adapter on the NAS isn’t delivering its full potential. It seems the issue lies with the NIC itself, meaning a replacement would be necessary. I suspect I made a mistake by opting for a cheaper model and relying on Amazon reviews. It cost me around $12, and the feedback was consistent across both options. The more expensive choice used the identical RTL8125 chipset, so I thought I was saving money. In reality, the card’s quality is crucial. A valuable lesson learned.
I received the new NIC and the output matches perfectly. When I have time, I plan to install it on my secondary machine (i5 9600K, Asrock Z390 Phantom Gaming 4s, Windows 11) to check if full speeds are achieved. If successful, I’ll try Linux LiveUSB to determine whether Ubuntu or the existing hardware is the issue. It’s frustrating in both scenarios, though I’m hoping it’s just an Ubuntu bug that can be resolved with software updates or a different chipset card. If the problem lies with the hardware itself, I’ll focus on upgrading the NAS internals to a more modern and efficient model.
This would have been a suitable choice had Ubuntu been the root cause. However, after testing with my secondary machine, it appears the issue lies elsewhere. It seems to be related to the NAS hardware itself. I managed to verify both NICs and confirmed they consistently achieved the expected 2.35Gbps speed in real-world conditions between two 2.5G connections. No driver updates were necessary in Windows 11—it instantly reached 2.35Gbps in both directions. When using a LiveUSB setup for Ubuntu, the performance remained consistent. The discrepancy of 1.6Gbps to 1.35Gbps suggests an asymmetric bottleneck, though I can't pinpoint the exact source. Currently, my connection is limited to about 1.5Gbps. I plan to upgrade to a more capable platform soon. Fortunately, both NICs were purchased on Amazon during their holiday sale and are eligible for returns until January’s end. If I find a suitable motherboard with a 2.5G interface, I can exchange them. Otherwise, I’ll keep one and continue using it. I’ll likely explore Black Friday offers for various motherboard models to potentially acquire a better option for my custom NAS setup.
Because Realtek can run at full speed across various systems, Intel isn't guaranteed to face identical limitations with your current setup. Each NIC may offer different CPU boost and I/O capabilities, and driver performance varies. Even if PCIe 2.0 x1 becomes a constraint, using a NAS with an x4 port could bypass it with a more powerful card like one from Aquantia NBASE-T. They might cause issues on Windows but perform reliably on Linux according to my experience.
One of the NICs functions as a PCIe 2.0 x1 card, performing well in any setup unless it lacks 1.0 speed support. This connection can handle up to 500MB/s, which comfortably exceeds the capabilities of a 2.5Gbps link that only achieves around 312.5MB/s in theory and about 295MB/s in practice. The feedback I've encountered on cheaper cards using the Intel i226-V chipset has raised some concerns. Interestingly, some negative reviews suggested these cards actually used RealTek technology, while others mentioned driver issues on Linux and FreeBSD. I'm open to trying one, but since I plan to upgrade for improved power efficiency soon, it's unclear if the effort is worthwhile. The old LGA1156 system serves more as a proof of concept for me—my goal is to explore what I can accomplish with a NAS to help make a stronger decision later.
I've experienced the situation firsthand with several i225-V B3 models, all functioning properly. My router supports i226 as well without issues. However, my managed switches let me disable energy-efficient Ethernet, which has consistently caused problems across various NICs since its inception. Mixing different client speeds can also lead to complications. For some time, I had to enable Flow Control on my switch or accept significantly slower communication with my NAS that uses 10Gbps. It appears the issue was addressed either through an OS update or a switch firmware patch, but it highlighted the importance of a smart-managed switch when handling diverse client speeds. Many users face difficulties using i225/6 at Gigabit instead of 2.5Gbit, likely because the common scenario involves chipset configurations on their motherboards.
I just enhanced my home network to 2.5Gbps fully. Since you brought up iperf3, I decided to give it a try. I found the speeds were lower in one direction, but I got the full rate when switching from IPv4 to IPv6. I don’t understand why this happened, but updating the driver for my Intel NIC helped align the results with other setups.