Preemptive versus Reactive

[Dan]

The helps guides suggest using:

Preemptive if speeds are below 50mbps, and Reactive if speeds are between 50&100mbps.

So at speeds of 60mbps Reactive should be best.

However, when setting CC sliders to 70/70% speeds drops below the threshold so should the router be switched to Preemptive for gaming?

Thanks

[N3CR0]

Its known that Reactive can give you a small amount of packet loss, so Preemptive is better for gaming.

Only use Reactive for downloading.

[Z.Janahi]

Can you assist please in finding where is the setting for it is located on the xr500 ?

[Dan]

19 minutes ago, Z.Janahi said:

Can you assist please in finding where is the setting for it is located on the xr500 ?

It isn't part of DUMAOS as it games it differently. Hence why I am back on original R1

 

[Netduma Admin]

1 hour ago, Dan said:

The helps guides suggest using:

Preemptive if speeds are below 50mbps, and Reactive if speeds are between 50&100mbps.

So at speeds of 60mbps Reactive should be best.

However, when setting CC sliders to 70/70% speeds drops below the threshold so should the router be switched to Preemptive for gaming?

Thanks

That's an interesting question! I'd go for pre-emptive and in that case, as it's optimal at the lower speeds. But there's not much in it.

 

52 minutes ago, Z.Janahi said:

Can you assist please in finding where is the setting for it is located on the xr500 ?

We revamped QoS for DumaOS (made it streamlined and powerful) so there's no longer the need for two options

[Dan]

14 hours ago, Netduma Admin said:

That's an interesting question! I'd go for pre-emptive and in that case, as it's optimal at the lower speeds. But there's not much in it.

 

We revamped QoS for DumaOS (made it streamlined and powerful) so there's no longer the need for two options

Unfortunately it behaves more like Reactive though so introduces some packet loss at lower CC levels

[RL317]

18 hours ago, Dan said:

The helps guides suggest using:

Preemptive if speeds are below 50mbps, and Reactive if speeds are between 50&100mbps.

So at speeds of 60mbps Reactive should be best.

However, when setting CC sliders to 70/70% speeds drops below the threshold so should the router be switched to Preemptive for gaming?

Thanks

This is a great question. I've always wondered if the algorithm should be chosen after you've adjusted your sliders (even though I'm on DumaOS now) especially if 70/70 would take you below 50Mb.

 

I noticed some discrepancies between the two with testing. For example reactive will cause anything lower than 100% CC to show packet loss on DSLReports, while preemptive would always shows 0.0% packet loss without fail. On the other hand a jitter test would show less variance using reactive while your connection is actually under load. However neither algorithm seemed to make a difference to connection quality in game, so I never really knew what to go for. 

[Netduma Admin]

4 hours ago, Dan said:

Unfortunately it behaves more like Reactive though so introduces some packet loss at lower CC levels

By lower CC levels, do you mean when you throttle to <30% (for example), or when you have very low bandwidth in general?

[Dan]

2 hours ago, lllRL said:

This is a great question. I've always wondered if the algorithm should be chosen after you've adjusted your sliders (even though I'm on DumaOS now) especially if 70/70 would take you below 50Mb.

 

I noticed some discrepancies between the two with testing. For example reactive will cause anything lower than 100% CC to show packet loss on DSLReports, while preemptive would always shows 0.0% packet loss without fail. On the other hand a jitter test would show less variance using reactive while your connection is actually under load. However neither algorithm seemed to make a difference to connection quality in game, so I never really knew what to go for. 

All games pre BO4 my gameplay was better on preemptive than reactive. On reactive I felt more vehind other players 

I get higher retransmission rates on reactive as I increase throttling and you can throttle much harder on preemptive before retransmissions start to occur

Ping plotter difference is negligible or non existent IE both offer me a 6-7ms ping and less than 1ms jitter under normal conditions

Under saturation, I think preemptive performed better giving less to no lag spikes.

Given up on BO4 but have been trying to tweak settings to improve Blackout by going back to the basics and rethinking settings that would now be considered natural.

 

[Dan]

45 minutes ago, Netduma Admin said:

By lower CC levels, do you mean when you throttle to <30% (for example), or when you have very low bandwidth in general?

Exactly the former, under harder throttling I can go very low on preemptive and maintain A+ across the board with zero retransmission where's DumasOS for the exact same speeds will have retransmission levels above 5 to 12%

This is why I am currently back on original firmware for now. 

[RL317]

1 hour ago, Dan said:

All games pre BO4 my gameplay was better on preemptive than reactive. On reactive I felt more vehind other players 

I get higher retransmission rates on reactive as I increase throttling and you can throttle much harder on preemptive before retransmissions start to occur

Ping plotter difference is negligible or non existent IE both offer me a 6-7ms ping and less than 1ms jitter under normal conditions

Under saturation, I think preemptive performed better giving less to no lag spikes.

Given up on BO4 but have been trying to tweak settings to improve Blackout by going back to the basics and rethinking settings that would now be considered natural.

 

Yeah for the most part on local PingPlotter tests I'd see 6 or 7ms, jitter usually averaging around 0.25ms, and sometimes as low as under 0.1ms. For some reason on TalkTalk if I ping my DNS directly rather than the default gateway (79.79.79.79 and 79.79.79.80) I get between 2 and 5ms even though they're based in Scotland and I'm nearer to the south England coast. Preemptive and reactive testing on PingPlotter didn't really seem to show a difference. 

1 hour ago, Dan said:

Exactly the former, under harder throttling I can go very low on preemptive and maintain A+ across the board with zero retransmission where's DumasOS for the exact same speeds will have retransmission levels above 5 to 12%

This is why I am currently back on original firmware for now. 

I guess my case is a strange one then. On reactive I could use 99% and see re-xmits. On preemptive I'd need to drop into a few hundred kilobits, whatever that'd be on the CC sliders, for the grades to start to take a dump. Unticking share excess shows re-xmits too 🤔

 

I never saw a difference in gameplay using preemptive or reactive though. I was going mad trying to figure out why some people swear by certain combos of algorithms and CC sliders give them a magic sweet spot and amazing connections lol

[Dan]

Agree unticking excess can cause issues hence I do everything off of the CC sliders

Have long given up looking for amazing connections; leprechaun has those at the end of the rainbow.

Geo filter does not help on blackout so R1 is only being used to throttle hence the preference for original firmware and and preemptive. 

Have seen quite a few positive comments about not throttling at all which is about the last thing to try so the R1 might come out of the setup

[RL317]

1 hour ago, Dan said:

Agree unticking excess can cause issues hence I do everything off of the CC sliders

Have long given up looking for amazing connections; leprechaun has those at the end of the rainbow.

Geo filter does not help on blackout so R1 is only being used to throttle hence the preference for original firmware and and preemptive. 

Have seen quite a few positive comments about not throttling at all which is about the last thing to try so the R1 might come out of the setup

I think I've officially retired. Nothing I try stops me from UDP lagging on low pings. I couldn't even run a custom game without teleporting so I knew it would be a bad idea to go online... 

 

but after a local offline game warm up vs veteran bots which gave me hitmarkers as good as I used to get in rare good games on 7ms to the UK server, I said fuck it and went online. Bad idea...

 

Broken aim assist, no time to react against idiots who supposedly snap like godly keyboarders, and either delayed hitmarkers or none at all in all 5 games I played on 7 and 14ms respectively. I played 20 minutes vs veteran bots, went straight online and averaged worse stats than I did vs veteran bots in every game with absolute dream lobbies... including those with not one player averaging a kill per life. I'm the only person online at home, using 70% QoS, prioritising CoD ports with traffic prioritisation, matchmake with 30ms ping assist only, run brand new VDSL and ethernet cables from wall to modem to R1 to PS4, have an HG612 setup optimally for TalkTalk in bridge mode, everything setup optimally on the R1 and PS4, and use a monitor with under a tenth of a frame (1.6ms) input lag at 60Hz. My CoD tracker stats would show me averaging 35-40% accuracy in those rare smooth games online, or 45-55% offline, with fully autos using 20 sensitivity on a controller... and now I get slapped by bots in every lobby 🙃 I must just be cursed. I've never seen anyone experience this so consistently, never mind this severely. I watch killcams and look like the noobs in K3/Swagg's videos, just running blind in a different position to what's really happening, looking elsewhere, supposedly not firing. 

 

What a big fat question mark ¯\_(ツ)_/¯ 

[Dan]

Dang!

My PS4 pro can get a bit spassy in offline mode if too many boys are added but nothing like that. Perhaps it's a combination of your boiler firing up, TV changing channel, door bell ringing and microwave being turned on. Or perhaps the neighbours are secretly fracking.....

Other than that surely that must be packet loss or a hardware issue? 

[Netduma Fraser]

Maybe I'm missing something here IIIRL but I would always expect online to play worse than local. Also the second online clip looks fine to me. Can you explain? If you're getting a lag spike on a local game then it would indicate there could be a non network issue somewhere or perhaps an ethernet issue.

[RL317]

1 hour ago, Netduma Fraser said:

Maybe I'm missing something here IIIRL but I would always expect online to play worse than local. Also the second online clip looks fine to me. Can you explain? If you're getting a lag spike on a local game then it would indicate there could be a non network issue somewhere or perhaps an ethernet issue.

Well of course, but that doesn't really explain why there have been games online on 7ms pings that play like offline (rarely), and the rest are like one bars. The second clip shows aim assist totally disappearing around someone before stuttering around them, and then me getting hitmarkers with delay in line with what someone on a literal one bar should expect. 

Custom games are accessible only while online, so if they play fine one day and they're lagging the next there's a network issue... it just doesn't line up with what the R1 tells me at the time. Local games are offline and flawless, as well as giving me aim assist, hit detection and smooth in sync player animations like I'd see when an online game actually runs well (a massive 1% of the time, if that). Those were 200ms+ hitmarker delays - nothing close to normal. That's about the best I've seen for weeks. 

A few days after installing DumaOS I had around 10 flawless games before the one bar-like low ping games reappeared. I had four games one day (all good), then the first six games of the next day were good until suddenly on the same server in the same lobby my hitmarkers start appearing 200-500ms late again, if at all. It's disgusting. 

[RL317]

I almost forgot to link this. This was the last time on the old R1 firmware that I got decent connections for more than just a single game in 100 or more. In fact it was 2 hours of flawless matches, and that was back in April. This is on 32ms or so to the Spanish server and I got instant hitmarkers in comparison. Watch in slow mo:

As soon as you fire one target on shot, the ELO sight ammo counter dropping down should coincide with a hitmarker:

If it does this on 30ms, why would a 15 frame delay be normal on 7 or 14ms?

The problem is it's not laggy players causing the issue. The problem is lag on my end with my UDP traffic. Pings are low and flawless, and it means nothing apparently. The connections I get now are infinitely worse than playing on a phone hotspot, which is just outrageous. 

[Netduma Fraser]

It won't have been the same server in that case, same location yes but likely a different server, the ID probably wouldn't have been the same, maybe slightly different. Have you tried on DumaOS in that case to force higher pinged servers then to see if they're playing any better?

[RL317]

11 hours ago, Netduma Fraser said:

It won't have been the same server in that case, same location yes but likely a different server, the ID probably wouldn't have been the same, maybe slightly different. Have you tried on DumaOS in that case to force higher pinged servers then to see if they're playing any better?

I used to try to force Danish and Polish servers (30 and 40ms respectively) because whenever the game would naturally put me on them with my geofilter off, they played astonishingly well. I even played on the New Jersey server at times (74ms) and could find faster hit registration. Now no matter what server I get on to, it's like a guaranteed one bar. 

 

It actually seems to have gotten worse since moving to TalkTalk - whatever they're doing on their end in terms of gaming traffic is bad, while pings are actually more stable and slightly lower in some cases. I mean I got this on the R1 diagnostic after switching:

 

--- 8.8.8.8 ping statistics ---

20 packets transmitted, 20 packets received, 0% packet loss

round-trip min/avg/max = 6.645/6.823/6.962 ms

 

I was getting numbers like this everyday, so when I did this test above I calculated the jitter between pings and it was 0.077ms. 77 MICROSECONDS. It's flawless on the surface with no packet loss. 

 

I guess I'm kinda stuffed because I just signed up to a new 18 month contract with them, and it's one with a "price freeze guarantee" so I won't be able to switch free of charge at any point until that's over. Funny enough I was trying to choose between TalkTalk and Plusnet before making the switch, and it was only after I committed to TT that I learned Plusnet prioritise their gaming traffic rather than just ICMP. So it looks like I'm stuck with a connection that plays worse than when I tried the Australian server in BO3 at 250ms+ LOL

 

Obviously the R1 is doing its job here. That's what makes it so ridiculously frustrating. My line looks perfect on the surface and then I run a UDP traffic capture on my PS4 and see silly delays. If I get low pings, no jitter and no packet loss it can't be a line/cable/socket issue can it? ☹️

[Dan]

16 hours ago, Netduma Fraser said:

It won't have been the same server in that case, same location yes but likely a different server, the ID probably wouldn't have been the same, maybe slightly different. Have you tried on DumaOS in that case to force higher pinged servers then to see if they're playing any better?

I have been doing this with the original R1 firmware and had a few good games at higher pings.

It is a bit trickier though with the server issues; even within the USA their are quite a few misplaced servers which the game naturally grabs. 

You can temp ban , restart game but when the next one happens, you restart and the old one gets through again

It's like filling a colander with water and trying to stop it leaking with one pair of hands

[Dan]

5 hours ago, lllRL said:

I used to try to force Danish and Polish servers (30 and 40ms respectively) because whenever the game would naturally put me on them with my geofilter off, they played astonishingly well. I even played on the New Jersey server at times (74ms) and could find faster hit registration. Now no matter what server I get on to, it's like a guaranteed one bar. 

 

It actually seems to have gotten worse since moving to TalkTalk - whatever they're doing on their end in terms of gaming traffic is bad, while pings are actually more stable and slightly lower in some cases. I mean I got this on the R1 diagnostic after switching:

 

--- 8.8.8.8 ping statistics ---

20 packets transmitted, 20 packets received, 0% packet loss

round-trip min/avg/max = 6.645/6.823/6.962 ms

 

I was getting numbers like this everyday, so when I did this test above I calculated the jitter between pings and it was 0.077ms. 77 MICROSECONDS. It's flawless on the surface with no packet loss. 

 

I guess I'm kinda stuffed because I just signed up to a new 18 month contract with them, and it's one with a "price freeze guarantee" so I won't be able to switch free of charge at any point until that's over. Funny enough I was trying to choose between TalkTalk and Plusnet before making the switch, and it was only after I committed to TT that I learned Plusnet prioritise their gaming traffic rather than just ICMP. So it looks like I'm stuck with a connection that plays worse than when I tried the Australian server in BO3 at 250ms+ LOL

 

Obviously the R1 is doing its job here. That's what makes it so ridiculously frustrating. My line looks perfect on the surface and then I run a UDP traffic capture on my PS4 and see silly delays. If I get low pings, no jitter and no packet loss it can't be a line/cable/socket issue can it? ☹️

I'm Plusnet so yes they prioritise gaming within their packages. Originally you had to pay for game pro.

Are you within your cooling off period.  You can challenge talk talk even after then. You have a good set of data to show them (although it may feel like banging your head against a wall). 

Even without gaming prioritisation you should only struggle at peak times. Does it happen all of the time

[RL317]

1 hour ago, Dan said:

I'm Plusnet so yes they prioritise gaming within their packages. Originally you had to pay for game pro.

Are you within your cooling off period.  You can challenge talk talk even after then. You have a good set of data to show them (although it may feel like banging your head against a wall). 

Even without gaming prioritisation you should only struggle at peak times. Does it happen all of the time

I signed up two months ago so no, I just missed it. Typical

 

It's a good 99% of the time, no exaggeration. It'll take me 100 or even more games just to find one on the same server (or even one with a higher ping) that randomly gives me a great connection, and if I'm lucky it'll last two or three games before going back to weeks of nonsense. For the most part I get on expecting the same and just play 15 or 20 minutes before I get off again. If I'm feeling really stubborn (like "why the fuck shouldn't I be allowed to play? I'm gonna sit here for a while") I'll just be testing connections for an hour or more. 

 

I can only think of a few days since February of LAST YEAR where I got more than one smooth responsive game during a session. There was the 8th of March this year (5 solid games, shite the next day), the 21st of April (2 hours of constant good connections, shite the next day), and then a few weeks ago on DumaOS. It was awful to start with, then I tried this new setup messing around with bandwidth allocation to different devices rather than global bandwidth control and I got four smooth games one day on 7ms to the UK server, and the first six games of the next day before it went back to one bar jokes. It's gotten even worse since then.

 

The only other time I've had decent connections in that time was on the 17th and 19th of May this year, where I set my filter in the north of Scotland when connecting to online services and avoided servers, then disabled the geofilter and got player hosted lobbies. In two games I pulled host and it was incredible. In the rest I connected to random hosts around Europe and on anything up to the 60ms I got to a host in Denmark, it played better than 99% of my games on servers, including the UK one at 7ms. I distinctly remember one game where there was a host migration and suddenly there were hexagon (packet loss) icons on screen for a full five minutes, and the connection was still better than most I've played on 7ms. Unfortunately it would take me up to 30 minutes just to get a lobby that was populated enough for me to start a game, so it's not exactly a reliable method of forcing decent games; my attempts to do this in the week following those games all failed to find a game. 

 

This was from 8th of March. My jaw hit the floor when I started playing, and most of my time was spent laughing because it was so unbelievable

Naturally the next day it was back to desync dogshit 🤣 what's really weird is that I'd been using the "typical" R1 setup that day - 70/70 preemptive, CoD Ultimate profile, PSN in hyper lane and auto everything else. That setup had never given me these results before. I've tried literally hundreds of different settings - optimal or otherwise, and tweaked one at a time or as a combo of new settings - and never attributed any brief moments of success to them because those good rare games are truly random. 

 

Actually thinking back to February 2017... I think that's when I got my R1. November 2016 was awful on IW, but December through February were comparatively great (say around a third of games were on great connections) while I was using just my BT Smart Hub. Most would come to the conclusion that the R1 is the problem, yet I've had great games with the R1 too. Clearly my network is capable of giving me these great connections because they do sometimes (well, seldom) appear, and I've spent so long on bad ones I can detect them straight away, in my first gunfight of the game. The problem is there's nothing in my settings that would logically affect my usual connection performance, so I just get on confused and get off confused - whether the game plays well or not. 

[Netduma Fraser]

You mentioned you've done UDP captures on the PS4 which are showing delays. 2 points, I assume you have been using Hyper Traffic/Traffic Prioritization. If so have you tried using advanced rules on this for the ports you've seen being used in your captures?

Also if that doesn't work it may be worth approaching talk talk with these captures and showing them that theres an issue

[RL317]

12 minutes ago, Netduma Fraser said:

You mentioned you've done UDP captures on the PS4 which are showing delays. 2 points, I assume you have been using Hyper Traffic/Traffic Prioritization. If so have you tried using advanced rules on this for the ports you've seen being used in your captures?

Also if that doesn't work it may be worth approaching talk talk with these captures and showing them that theres an issue

When I did that test I was on the old firmware so I was using PSN in hyper lane. It always seemed pointless trying to set up individual custom rules since you couldn't add a range of ports with hyper lane, and there are up to 20000 ports used on the server end when receiving traffic lol. I guess I could try again with DumaOS, but the PSN hyper lane preset covered all of those ports anyway. It included UDP 1024-65535 right? The server port is randomly picked from 30000-50000 (it was 40040 when I did my test) so that should have been sufficient. And of course if the issues are on the ISP end, surely the latency on those packets will be there before reaching the R1 anyway? I've considered trying the Huawei Openreach modem's QoS but I'd have to boot it into router mode and I don't know if it'd have an impact if the issue occurs before the modem. Also it appears to use 802.1p with the 0-7 priority range thing which I can't get my head around, as well as other settings that I'm unfamiliar with. 

 

It is nice to see you can now mention a single port in more than one traffic prio rule without the "streams overlap" error message. Maybe that'll help with testing, but probably not until (if) I find myself with a working connection again lol

[Netduma Fraser]

Yeah it does cover that range but I do remember some users adamant that prioritizing port 3074 for example manually gave them a better experience. If its on the ISP end then yeah it won\t help too much. They probably have a user guide online to help with that. New modem or confront ISP with your captures are probably your best bets for next steps