4 ms·
I am not any more in this business but I was involved at IEEE 802.11 some years ago. * One limit that you should consider is the fact that if you use Wi-Fi wit
by JPLeRouzic 10y ago
I am not any more in this business but I was involved at IEEE 802.11 some years ago.
* One limit that you should consider is the fact that if you use Wi-Fi with more than a few hundred megabits/sec, then the computer busses will be overloaded at the hardware level.
* Then memory chips are not so fast either, with respect to modern Wi-Fi.
* Another problem is the fact there are few free drivers and the one I saw (four years ago) were actually very simplified. For example there was no MIMO, which was supposed to exist since 2007 in IEEE 802.11n, even simple MAC features do not existed in 80211mac.
* And indeed most chips are not free, they load a binary at powerup, and nobody knows what's in there. I suspect that it is quite ugly. What you see in the "free" Linux driver, is just calls to the binary API.
* Another thing is that most kernel and userland software use memory buffers which slow incredibly any transfer.
Obviously things may have improved but I would check those five points.
- gtirloni 10y agoMakes sense. These small ARM-based boards might be to underpowered for my needs (capable home server, not too worried about power consumption, etc). Great insights, thanks!