Please don't tell me you think your fps is related to your network settings please god don't make me think these thoughts....

Put tin foil around your router, should do the trick.

I don't, I just want to avoid the comments on how you need x fps for y

#5
Most if this is correct, but cl_cmdrate/cl_updaterate > 66 is just bullshit.
It just screws up cl_interp_ratio.

Quote from the valve developer wiki (Basic Networking)

Edit: I just saw that you got that from stabby. stabby stabby's advice concerning the TF2 engine is horrible. He reccomends settings that cancel each other out, launch parameters that don't even exist and a lot of other stuff that is questionable at best.

Edit2: 60000 is fine, higher than 100000 won't change anything, it's unlikely that you even need more than 60000. If your rate is higher than your download/upload it might raise your ping, but honestly TF2 won't ever need more than 100kbits download and even less upload. Unless you're using ISDN and not DSL you'll be fine.

I've been having similar problems to OP.

I copied and pasted the thing for the config earlier and noticed some improvements (still randomly disconnecting the internet and getting frame drops, but they are less bad and less frequent). Is there anything I can do in addition to help? or more specific values I can set?

On speed test I just got
down: 2.09 mbps
up: 0.79 mbps

thanks

chris' fps configs.
For both.
Frame drops are just part of TF2 and have nothing to do with your netsettings
If your internet randomly disconnects (not just when playing tf2) then your connection or your router has a serious problem.

Yeah I made another thread about them, my router is 5 years old and apparently was cheap and bad when we got it.

I already uses chris' fps, but I probably ought to switch to one of the more serious ones. I was getting 200-300 frames with my fibre optic connection at university, and now it's all over the show. Never had frame drops until I moved back home.

thanks for your advice

@5

>set this equal to your cmd rate.
This is not necessary. They can be different and there's no problem with it. In fact, TF2 sets them to different defaults for me on startup.

>rate: unless you have a poor connection, set this to the max value of 128000
lol

cl_interp_ratio: set to 1
lol

1. The defaults are horrible. 66 for both. 40 if you can't get enough fps But less than 60 fps aren't good anyway.
2. Why not? 128kbits is not much. I'm not sure if it's the max though. Most servers restrict it to 100000 anyway.
3. This is a good setting for a good connection, don't know what's so funny
4. dat quote lol

>1. The defaults are horrible. 66 for both. 40 if you can't get enough fps But less than 60 fps aren't good anyway.
You're a daft motherfucker. Whether the defaults are actually good rates as rates or not is irrelevant; the obvious point I was making is that the game wouldn't set them to values that differ from eachother. People who play with differing message rates experience no problems from it. Parroting incorrect information (in this case, the notion that you're doing something wrong if they're set to different values) is wrong.

>2. Why not? 128kbits is not much. I'm not sure if it's the max though. Most servers restrict it to 100000 anyway.
See #6

>3. This is a good setting for a good connection, don't know what's so funny
No, it is not a good setting for a good connection. Setting the interp ratio to 1 causes models to stutter roughly every other frame due to networking jitter, because any two adjacent messages have roughly a 50% chance of having longer than "interp_ratio 1" time between them, causing the second message to be considered missed for a given frame and with extrapolation is disabled in TF2 and an updaterate of 66 on a refresh rate of 60 you'll be skipping six out of sixty six animation frames per second and everything will be in the wrong position most of the time (usually behind where it "really" is for lag compensation). This is why many people experience rockets warping through scouts, and it's the prime suspect of people getting bad hitreg.

>4. dat quote lol
Because the guy I'm responding to didn't have a huge wall of text in a quote or anything either, right???

1. Enlighten me, why should you set cl_cmdrate and cl_updaterate to different values?
2. I am #6. 128000 or 100000 it doesn't make a difference.
3. Network jitter is usually not a sign of a good connection.
4. To me quoting after your own post seems really funny.

>1. Enlighten me, why should you set cl_cmdrate and cl_updaterate to different values?
It's perfectly reasonable to lower cl_updaterate to reduce choke, and perfectly reasonable not to lower cl_cmdrate at the same time.

>2. I am #6. 128000 or 100000 it doesn't make a difference.
You say yourself that more than around 60k is not important.

>3. Network jitter is usually not a sign of a good connection.
Network jitter is complete and utterly unavoidable: it will always be there. Responding like that only shows how little you really know about this topic. How severe it is just affects how high your interp has to be to make up for it, but when you set interp_ratio to 1 and interp to 0 then you absolutely positively will have a roughly 50% extrapolation rate online.

>4. To me quoting after your own post seems really funny.
Quoting my own configs in fact. Wow!

That's interesting, what is interp_ratio supposed to be at in that case?

wait so you're not supposed to have those two rates at 66? Or cl_interp_ratio at 1? Chris has always recommended it, and don't some comp servers force cl_interp_ratio 1?

>what is interp_ratio supposed to be at in that case?
It depends on your connection and what you're doing. My ugly quote from earlier has a bunch of aliases with different "good" ratios and what they're used for. Interp ratio 1 as a universal setting is a nightmare.

..

>wait so you're not supposed to have those two rates at 66?
The only reason to have them that high is to lower latency a little, since it's still not quite the 66.66... that the game runs at. Chris recommends 66 it for good connections, and that's right, but on a chokey one setting updaterate to something lower like 40 is the first thing someone should do (I'm repeating myself).

>cl_interp_ratio at 1?
Bad general advice. Great on LAN, bad for hitscan weapons in general, better than default for projectiles but not as good as something like 1.2 or a bit higher.

>don't some comp servers force cl_interp_ratio 1?
That's their bad configuration problem. My configs set the equivalent cl_interp of whatever the ratio is, so they don't bother me.

cl_interp 0 is the one that makes it do what you said for servers with forced ratio 1?

Also how does this look?

For hitscan classes I use cl_interp 0.033 and cl_interp_ratio 2.2
For projectile classes like Soldier I use cl_interp 0.0152 and cl_interp ratio 1.0032

cmdrate and updaterate are both 66 (my connection is good). I could try inter_ratio 1.2 or 1.4 I suppose, but I'd have to change the cl_interp with since that forum inter_ratio/updaterat=best cl_interp right?

1. There is still no reason to set them to different values by default. Why would you use a config by default that is optimized for servers that get throttled by their bandwidth (especially in comp)?

2. It's not like it's bad setting it to more than 60k so what is your problem with that?

3. Network jitter really shouldn't be that high on a good connection.
You use cl_interp_ratio 1 in your own settings. You even use a lower value at one point

// Harshest timing oriented interp settings -- uncomment only the one appropriate for you
//alias net_harsh "cl_interp 0.0152; cl_interp_ratio 1" // for updaterate 66

0.0152 where it should be 0.0152152

4. Most of your config is identical with chris' settings, the rest are just weird interp settings. You use both cl_interp_ratio and cl_interp which is unecessary and most of the time both counteract each other because you set them to different values.
See valve dev wiki

#18:
Yes, that's right (@ first line). And yeah, those look like perfectly fine settings (@ middle lines). The only (and good) reason to change cl_interp from 0 when you change cl_interp_ratio is to keep zervers from breaking your interp, so that's only something to worry about if you happen to join servers that are badly configured (@ bottom line).

..

#19:

>1. There is still no reason to set them to different values by default.
I never said anything about telling people to set them to different values by default. I said that TF2 happens to set them to different values on startup before my configs execute as a starting point for the argument that having them be different does *not* inherently cause problems and should *not* be instructed to be avoided. There's a big difference between "use differing values by default" and "don't tell other people they can't use differing values".

>Why would you use a config by default that is optimized for servers that get throttled by their bandwidth (especially in comp)?
First, I never said this is something that should be done *by default*. Second, I never said anything about servers getting throttled, rather I've been talking about clients experiencing choke.

>2. It's not like it's bad setting it to more than 60k so what is your problem with that?
If you set it too high for your internet connection then if the server sends you too much you can basically bottleneck your own internet connection and get bandwidth/throttling issues at an internet service level (instead of the game's netcode choking on itself temporarily).

>3. Network really should be that high on a good connection.
"Network really should be that high"? Do you have any idea what you're saying? You CAN NOT get rid of jitter on an online connection. It's ABSOLUTELY IMPOSSIBLE. If you can do it, then for other necessary reasons you're the fucking jesus of the information world. Go get hired by a pro audio company, because you've just solved the impossible perfect conversion analog <-> digital problem. How much or little there is does matter most of the time, but when you set interp_ratio to 1, it *does*; however much jitter there is doesn't matter at that point because then you're identically doing a diceroll about "Did I get this tick's update in time?"

>You use cl_interp_ratio 1 in your own settings. You even use a lower value at one point
Note the following comment:
// Use this instead of the above if you experience rockets "warping" through scouts, microstuttering animations, or other unpleasant jitter when using the above.

>0.0152 where it should be 0.0152152
Nope. Correct value is 0.0[15], I just rounded up on the ten-thousanths.

>4. Most of your config is identical with chris' settings, the rest are just weird interp settings.
Of course it is, he distilled the list of netcode settings down to the minimum relevant ones so it's impossible to end up with anything substantially different.

>You use both cl_interp_ratio and cl_interp which is unecessary and most of the time both counteract each other because you set them to different values.
Okay, I'll break this down for you:
1) The game does use the higher (in realtime) between the two, that's true.
2) Ratio can be restricted by the server to lower (or higher, but not relevant in this case) than what you set it to. Due to so many servers being badly configured and restricting ratio to 1, cl_interp is included as a safeguard against this so that hit registration doesn't suffer. Anecdotal evidence, but I kept getting mad at Loops until I fixed my configs because his pug server restricted ratio to 1, and my hitreg as scout was always absolutely terrible and my rockets were warping through scouts until I set cl_interp to something other than 0.
3) Setting them to different values doesn't make them "counteract" eachother at all. In the cases where cl_interp is higher than cl_interp_ratio/cl_updaterate, that's only because I rounded up and left the values simple/reasonable, and the extremely slight difference from the accurate values this causes is absolutely negligible.

1. I think we're just talking cross purpose now.
It might give you better performance when your performance is not optimal, but for optimal performance they should be the same.

2. 128kbits isn't an issue for most connections plus it's not being used most of the time, so it won't hurt you although 60k would probably be better.

3. I messed up that sentence.
Updates are sorted by server time, not client time.
Unless your fps are capped at 66 and were perfectly in sync with the updates the updates will be used in the NEXT frame that is being calculated anyway.
network jitter > frame calculation time seems really high.

4. I meant that just cl_interp would've been easier and shorter.

if you have a 60Hz monitor are you losing out on those extra 6 frames from updaterate being 66? Is 60 a better choice for 60Hz monitors?

no, because it's something related to client-server connection. client rates should be the most similar possible to server rates, the normal tickrate for most of them is 66, so don't change it.

I don't know why people get so crazy about interp and net settings. 0 for projectiles, and .033 for hitscan. Typically when you think you're missing shots from bad hitreg, its just you not aiming well. If you don't believe me, watch your demo's slowed down.