sfx2000
Part of the Furniture
Test of the PCE-AC68 (with Windows drivers) and the EA-AC87
The test made this morning has been of great help. I've decided to download an ISO of Windows 10 from the Microsoft website and burn a live system in an USB hard disk (using WinToUSB). Then, I've boot the desktop computer with it, I've installed the drivers from the ASUS website, I've run IPerf3 and... Bingo! I've got 440 Mb/sec on download, and 392 Mb/sec on upload! These results are closer to what I've seen in others' benchmarks.
This means that the problem is in the Linux drivers for the PCE-AC68, definitively
That's what I'm kinda thinking - rate adaptation on linux uses Minstrel - which selects the MCS based on packet error rates/time - the driver might be optimistic on it's estimated, based on DL stats, but when you see UL, it's clear that packet errors are likely higher...
iperf3 can show this by running things both ways, and look at the retransmissions, weakness in the 802.11 will show up in the network layer (iperf3)...
Quick question - on linux - which driver is being used? There are at least three...
Driver Description
brcm80211 Kernel driver mainline version - since it's PCI-e, should be brcmsmac
b43 Kernel driver reverse-engineered version
broadcom-wl Broadcom driver with restricted license
Broadcom's FOSS driver (brcm80211) might have issues in 5GHz, the wl closed source might not...
Truth be told, Broadcom NIC's are a PITA compared to other vendors on Linux, but for add-in boards, the selection is somewhat limited, so one has to work with what is available...
