Im taking a little informal survey here to determine what the average net speed while playing is. Id would like to know:
Your ISP’s advertised speed
Your actual speed while the game is running
PG on or Off
brand of cable modem (optional)
I dont really care to much about your machine specs but you can post those if you want.
To get the speed while your running, open task manager, click on performance tab, click on the box your drawing your connection from ie Ethernet Wifi etc. for this test dont pause the game you need to flying while you check the numbers sitting on the ground or being paused will report 0 speed.
Also I would be interested in knowing what your bufferbloat is when you run the speedtest from dslreports.com .
Advertised speed: 100mbps
Game speed: this varies but usually rides around the 1.5 - 2 mbps with occasional bursts up to 70 mbps on occasion.
PG = off ( with PG on my network speed will run avg about 2-3 mbps with spikes upto 70 )
VPN - I setup a vpn server/client setup over the weekend and for 6 hours avg 30mbps with PG on, left it run over night and that was back to 1or 2 mbps.
Bufferbloat Ive seen reported as high as 3600ms (this needs to be measured with nothing else running) with out QoS running.
If I setup Qos on the router and trim my netspeeds to 70/10 mbps bufferbloat goes away and my networks speeds while the game are running are alot higher 20-35 mbps, and I noticed alot of the ‘problems’ of the game disappeared. But this doenst seem to always work this way.
So what Im trying to determine is wether its the servers that cant cope with the bandwidth usage or the providers are throttling the connection because their load is to high…
Thankyou.
1 Like
As a Network Engineer for an ISP I suspect that you will find that there is no clear answer. The cause of peoples issues will vary from congested links out on the internet, to local hardware issues, to a specific server cluster at a specific data center having issues that day. Finding a universal cause is highly unlikely.
That said the data could be interesting in itself I will try and remember to post mine when I am at home this evening. I would be good to include if the user is wired or wireless on their local LAN, the brand/model of the chipset for their net adaptor, network adaptor driver versions, OS version, number of devices active on the local network.
2 Likes
The game can’t utilize your speed. That would be crazy.
While playing it shouldn’t be stable, it spikes when loading more scenery.
I have 1gbps, have no idea about cables or anything else. Router is arris something, don’t know.
Alot of what Ive been reading about this seems to indicate that modems based on the Puma chipset are bad and cause a ton of buffer bloat, and if you can run from a broadcom based modem the problem goes away.
But I have noticed if I use QoS on the router to shape the traffic avg speed goes up (most of the time) and the sim looks and acts differently.
But if the people that are having the stutters and alot of the other issues are also exhibiting high buffer bloat, it could be as simple as switching out for a non-puma based modem.
And your right there is alot more involved with traffic flow as well. But this is just a start, hell it might get 100% better on the 27th then I can go back out of troubleshoot mode and go back to user mode lol ;p
Not all - problems only with Intel PUMA 6/7 chipset.
So puma 5 or 8 are good? I read somewhere that the puma chipsets seem to work in certain areas of the country without the bufferbloat issue.
Are ISP’s intentionally using them to limit bandwidth to get my connection to report clean I have to trim about 20% off my advertised speed to get a ‘clean’ connection.
ISP’s are not worrying about the chipset at all. Cost, familiarity and compatibility with their backend equipment are the driving forces behind which brand of modem/router an ISP uses.
8 wait what am I sure of what. What Ive been reading indicating that there are issues with the puma 6 and puma 7 chipsets. I dont know if there’s and 8 I havent read anything indicating this. It also says that in some parts of the country the puma chipsets work correctly with no extreme blufferbloat.
Now this reminded me of a situation years ago, during the birth of the public internet where our network engineer suggested and got installing the USR terminal server / router units. The problem that we discovered after installing them in the new POP’s were the the the telco systems in some of those areas were unable to support the 56k tech. (wow that was along time ago)…
I run benchmark flights here, always the same point a to point b route same plane, and most of the time same weather. Some flights the network will be pulling avg 1-2 mbps, and other times 20-30 upto 80 mbps.
So what Im trying to figure out, is whether the issues lies with the provider (here use our speed test on our inside network…because we only have a 128k connection to the outside world), or the Azure network (and there have been tons of articles out there about their problems).
I mean why go spend $200 on a cable modem if it isnt going to change anything because the ISP happens to be throttling the games connection because they cant support the traffic or because the game cant supply the bandwidth.
I called tech support the other day when I first started to suspect they were just throttling the connection, and the tech assured me that they werent. Later on in the conversation she said 'I dont know what throttling is but I assure you we dont do it. I read other reports indicating otherwise.
So there you go Im just curious at this point in what the general population is getting for speeds.
Both last night and this morning I did a few flights well monitoring how much bandwidth MSFS was using. A total of around 6 hours of flight time, with flights of various lengths and starting and ending at both rural and population centers like LAX.
At no point did the sim use more than 3Mbps and averaged less than 1Mbps. Overall bandwidth usage was higher at times, however that can all be attributed to the wife streaming her crime shows, and kids downloading/playing games during times of higher usage.
Note that the Puma chipset defect has more to do with latency rather than bandwidth consumption.
And/or how many hops, and to where, as each person that will be different.
I could have 10ms to one location, but over 200 on the next hop, and 5 finally at MS servers. It’s all over the place
1 Like
Correct people’s routes to the servers will be different dependent on ISP and location. This will cause different latencies.
The Puma chipset defect (as admitted to and outlined by Intel) causes high latency at the users modem/router. Essentially the periodic maintenance tasks overload the chipset causing a delay in frame forwarding. This will show up to the end user as a latency spike and jitter.
What you are describing is a completely different scenario. If you have good latency at the end node 5ms in your example and a node is showing 200ms someplace in the middle your actual traffic is not being subjected to that 200ms latency. ICMP traffic is deprioritized, actual traffic is forwarded first before the node will respond to the ICMP traffic. This would be a node that is seeing high traffic levels but not yet enough to cause an issue.
Yeah I know that puma=latency issues but when you get a bufferbloat in excess of 2000 at times, it seems to be a contributing factor.
What you wifes doing on the net shouldnt be an issue if you used taskman to watch network speeds, unless of course your both on the same machine. ;p
Now this morning on my wireless during a 22 min flight, and PG on I avg a pretty constant 30mbps up until I got to a certain point then we dropped to the 1-3mbps range not much scenenery after that point, though after landing and taking off again, I was getting 30mbps with 12, 10, 8, … fps thing going on, different bug I would ascertain card wasnt overheating running at the same temp it always does.
Guess I’ll have to wait until the 27th and see what surprises that brings … fingers crossed.
I am monitoring traffic off my switch that lies between my modem and internal LAN, not off my PC. 
What location are seeing MSFS use 30Mb? Curious if I can recreate that.
I fly from KCLE to KERI along the Lakeshore. This is my benchmark area. There are times that PG makes Cleveland look like it was incinerated in a nuclear blast and other times it almost looks good.
The higher my bandwidth seems to be running the better Cleveland will look…
There seems to be no rhyme or reason to it at the moment. Lighting is another issue (in the caravan) on flight even with all the internal lights on you can barely read areas on the cockpit dash, and other time that its so bright inside and out that you almost have to put on sunglasses…
ALL I WANT IS SOME CONSISTENCY! 
I doubt you will get consistency like you are looking for. More than likely the game will use as much bandwidth as it needs up to some maximum. The server will be setup to limit bandwidth per connection to limit the likely hood of only a few users bringing it down. I am going to fly that route a couple times will report back what I see.
Okay I’ll turn on multiplayer…maybe I can harass you some in my Caravan! Bwhahahahah …
Oh wait the servers are down at the moment ;p
This is an example of what Im talking about…it will do this or better in a cached area in the benchmark flight. It will do it for about 20-??? mins depending on the time of day as well. Sorry I thought I got the whole display in it for you.
I was able to catch the spikes you are talking about. I can only get them to happen near cities with photogrammetry. I see smaller spikes if I fly in areas I do not normally fly. This is going to be expected behavior, when their is more data to download from the servers the data usage will go up until the needed data is downloaded.
If you go into the general options under data you can limit bandwidth usage if you want, although I would expect doing so would negatively affect graphic quality especially in cities with PG data. Not 100% sure but perhaps increasing rolling Cache size may help limit bandwidth consumption in areas you fly in a lot.
For your data points I have a 250Mbps cable service that is very solid, we rarely have issues with it. dslreports shows buffer bloat never higher than 43ms. I am not doing any QoS, although I do have the ports connecting to my kids rooms and our Wireless network rate limited to 50Mbps, going to our upstairs where me and my wife are plugged in are unrestricted.
Yeah I get all that but I dont really have ‘spikes’ it will remain relatively steady at the 30mbps. Regardless of whether Im looking at Cleveland or not. I dont know next week I get paid gonna buy a cheap broadcom based modem and see what that accomplishes if anything, and hey after that we have the update that will fix all this anyways Right. RIGHT? lol
I was just flying the area for 50 min was getting a steady 30mbps all the way to KBUF had live traffic on and I wanted to land on 6 because as you make your approach there’s a perfectly good looking bldg complex that redraws itself right before touchdown… so Im making my approach watching to see what the bldg does, when all of a sudden what should appear but an A320 leaving on 24. I think I heard his wheels scrape the top of my plane as we passed. Wish I would of got a pic of it but I was to busy laughing as it happend …
Location - Malmö Sweden.
250/50 cable service. Pretty old Netgear cable modem.
Dslreports data - pc idle.
Bufferbloat = 68ms down / 40ms up
In game flight (rural Denmark, daylight)
Up - 56>72 kB/s (rough range)
Down - 1>9mb/s (rough range)
Naturally outbound traffic increases when running simconnect applications. I ran LittleNavMap, Volanta and SkyElite all at the same time and saw probably around 1.8mb/s upload consistently enough.
Download traffic remained in the previous range.
Notes.
All flights performed in same area. Around same game time (18.00>20.00) around EKYT
Real time flying was between 19.00 and 23.00 local time.
Photogrammetry off.
Live weather (not live time)
Live traffic
(In fact all data services are on except photogrammetry)
