@kampfader, can you describe how you performed your test?
Was this a file copy under the File Manager?
A linux file copy from an SSH session?
When copying files you always have to be careful which end of the connection you are testing. For this reason you commonly try to make the end you are not testing as fast as possible to minimize its effect on results.
For example, if you are transferring your data out from the test system to another, you might direct the output on the uninteresting destination to a RAM disk or /dev/null to minimize its impact.
Similarly, when you are testing the destination, you try to eliminate the source contribution by using a source RAM disk or /dev/zero.
Doing both tests is particularly useful when you have a performance problem but you don’t know whether it is at one end or both.
The speed is always determined by the slower interface.
In your case the slowest interface may be the USB port or the speed of the network. In theory, a USB 2.0 port has a speed limitation of 35 MB/s. Obviously the SATA interface of your PC’s hard drive has a very high speed and a hard drive too.
mit der nightly von 01.05.19, was ich gerade verwende, habe ich diese rate?
Kopieren von N2 HDD USB3.0 --> DDWRT Router --> PC (LAN) — Max 103 MB/s
andersherum komme ich auf max. 47 MB/s
es war eine 1,9GB .mkv datei
Diese werte wurden mir wären dem Kopiervorgang im Win10 angezeigt.
with the nightly of 01.05.19, what i’m using right now, i have this rate?
Copy from N2 HDD USB3.0 -> DDWRT router -> PC (LAN) — Max 103 MB / s
the other way round, I come to max. 47 MB / s
it was a 1,9GB .mkv file
These values were shown to me when copying on/from Win10.
I would grant you that 10 bits per byte remains a good approximation of the efficiency of transmission from raw bits to usable bytes, though only asynchronous serial communication follows the pattern you described.
However, this is a discussion of disk IO and only bytes (not bits) are relevant. No disk speed tool reports results in bits/sec.
The efficiency of the protocol is not a serious issue on a nominally 5 Gb/s USB3 link with devices that can only hit rates on the close order of 100 MB/s, such as your magnetic disk drives. USB2 was different in that performance was well below the 10 bits to the byte effective rate, coming closer to 16 bits to the byte in practice, i.e. 30MB/s from 480Mb/s.
The fact that a serial transmission medium can operate at a speed of, for example, 80 Mbps (fiber, DSL, USB, etc.) does not mean that it is capable of transmitting 10 MBps, it will always be lower. I always use speed_bytes = speed_bits/10 approach.
Do you allow me to continue using this approach?
Note: For similar reasons in WiFi transmissions I use speed_bytes = link_speed_bits / 20
What does 5G have to do with wifi? That’s two different things.
In any case my gigabit wifi can get 100 mbyte/s tranfers, but only under ideal conditions where there’s no other utilization and the computer is close to the station.
Wifi speed on my network drops down quite a bit if there’s more than one computer transferring large files or if distance to the station becomes large. The hardlines are still much better for shared utilization.