I ran a test and it was
89.89 mbps DL, 5.8 mbps UL, 29ms ping, 1 ms jitter
Another test saw 56 ms latency, 20 ms jitter, 3.1% late packets
Is that decent? Says it's rated as excellent.
Jitter is on the high side. Generally jitter should be below 10ms, above it, you will start having packet losses. 3.1% late packets is also bad. Anything above 1% of lost packets means there will be problems in the connection stability. If I remember correctly, people say that 1-2.5% is "acceptable" for streaming video/audio.
56ms latency is just dependent on geographical position if there are no packets losses or jitter.
For example, I've got 12ms ping from Zagreb, Croatia to Frankfurt, Germany and ~110 to US East and ~200 to US West with 1-3ms jitter and 0% packet loss.
Download/upload speed doesn't mean too much when it comes to stability. It's just the speed of downloading/uploading stuff. Most video games need only about a couple of Mbps Dl/Ul.
Point is, I would not trust anything that has jitter higher than 15ms and packet loss higher than 1%.
Concerning models dropping. I'll speak only from my knowledge gathered during physics studies (had plenty of IT subjects) and general computer knowledge.
I don't see how connection can affect model drops.
Mechanic plays out by pure logics and statistics. You can have the best hardware and connection, it should not matter.
Games run in a server client architecture, which means the server is responsible for running all the game logic max frame rate and sending all the data back to each of the clients. That's added onto the inherent geographical latency of each player. Hence why people tend to have worse server stability when playing with opponents that are all far from each other.
Example: (3v3) I'm from Europe, Croatia and if I play with US players, or Europeans or [north] South Americans, I'll have decent overall server latency because the "TRIANGLE"
[ US(server)-US(player)-Europe(players)-north South America(players)] doesn't have large latency deltas between each location.
If I play with Asians or Australians/Oceania or south Africans or south South Americans, I'll get HORRIBLE server stability because of that nice little thing called the Pacific ocean which is extra large, so the "TRIANGLE" has a massive delta.
That's something I've tested each and every time and concluded that geographical locations of players matter in the overall stability of the game (even though we all connect to the same server).
So, game logic is not something that can be influenced by the latency because the server will sync everything up, no matter what. The only way I can see that the server influences "RNG" is if by some miraculous non-logical way the server "forgets" to remove models because the "damage" is "late"... but then again, you would not see enemy models dropping health.
What could happen theoretically, albeit rarely is that since each model has a fixed HP pool, if there is a significant latency, some models would not start firing on the "closest" model but would spread out damage across all models, dropping HP but not dropping models.
Example: 5 model rifles vs 4 model sturmpios
Rifles engage on charging spios. Under normal circumstances the spio model closest to the [parallel placed] rifle squad would get gunned down by all 5 rifles, hence the spio count drops to 3 and all is well.
Induce large latency and the location of spios gets jittered, hence the game logic sees more than one model as being the "closest" and decides to engage (hence why infantry cover to cover tests on TestMap are generally done by using smoke between them, and not just setting the other unit to "enemy").
But this is just one large theorycraft and I HIGHLY DOUBT that's what happens.
I think that the models not dropping is just piss poor luck where the damage is spread out over all models evenly.
Relic Entertainment
