|
M_Gargantua posted:For anything longer than a few feet using copper for more than 2.5G is inefficient, and using it for more than 10G is wasteful. For anything local, DAC is the answer. For anything else Cat6a is a good and useful standard cable. Run Cat6a when you're doing new installation. Decision point #1: You don't need 10G, stop it. Decision point #2 (if you ignored #1): Just run fiber. 
|
#
¿
Apr 2, 2024 18:23
|
|
|
|
| # ¿ May 20, 2024 23:46 |
|
|
A internet speed testing website that, at least in my experinence gets better speeds, is Tele2's speedtest. And the bufferbloat test probably also needs to be mentioned. If anyone's using iperf, be aware that version 2 and version 3 are not equivalent - version 2 uses multithreading, version 3 uses singlethreading. You kinda need both.
|
|
#
¿
Apr 17, 2024 17:38
|
|
|
SamDabbers posted:60Ghz 802.11ad/ay gear exists in the wild but mostly for inter-building bridging rather than directly to client endpoints.
|
|
#
¿
Apr 25, 2024 20:10
|
|
|
Kibner posted:Thanks! I have tried to search for what packet size Sunshine uses but have not been able to find anything useful. Maybe it changes depending on bit rate? No clue. EDIT: Sure looks like it. Voice payloads tend to be on the order of 30B, so once you account for the UDP datagram and IP header, it only works out to ~250bps per voice stream. The difference is that they're very sensitive to both latency and delta-latency - so you want to ensure that VoIP traffic given a higher priority through your network, and that they're always guaranteed the bandwidth. A lot of gear will advertise "gigabit speeds" but can only achieve it at MTU, whereas gear that can do 1Mpps isn't going to be artificially limited in the same way. BlankSystemDaemon fucked around with this message at 15:30 on May 16, 2024 |
|
#
¿
May 16, 2024 15:18
|
|