72 replies to this topic

[Metus regem]

Deathlike, on 21 February 2018 - 07:29 PM, said:

The worst part is that you hear little to nothing about netcode updates at all. It may not even got unnoticed, but at this point there are some things that are "built into the game" that isn't even being vetted (like how shooting projectiles like PPCs seem to "miss" its targets even while you see your projectiles hit them).

But hey, it's easier to sell mechpacks and not improved netcode.

For me Death, the worst is when I watch my PPC's pass through my target... I've had that happen a few too many times... hit what should've been rear LT/CT/RT, saw no damage go up on the readout, watched as the PPC's continued on to hit terrain.... I've lost matches because of that... I learned how to reduce the chances of that happening, setting PPC's to stagger fire seems to reduce the odds of it happening, but it still does... by the way, my average pin in MWO is around 50 for NA servers.

[Jackal Noble]

Relishcakes, on 21 February 2018 - 08:41 PM, said:

I am incredibly confused by all of the issues people keep having with hit reg, convergence, and connection issues. Maybe I am one of the blessed few but I don’t notice issues with convergence(to be fair I am not certain what this even is). I have had 1 disconnect in 2 years and it was an issue on my end. How the hell is everyone else running this that’s having so many problems?

Not even sure what planet you live on RelishCakes, but it sure as hell isn't this one.
It always amazes me when I see the longest standing T4 player in the game comment about something.
485 games total from season 4 to season 18. just let that sink in. Like no idea, at all.

[Imperius]

If only more people backed me when I tried pushing for a new engine update. I was told community warfare updates > keeping this game current.

Oh and the guy I met at MechCon that said “eff that dude on the forums asking for engine updates” it was a pleasure meeting you.

[Metus regem]

Imperius, on 22 February 2018 - 01:30 PM, said:

If only more people backed me when I tried pushing for a new engine update. I was told community warfare updates > keeping this game current.

Oh and the guy I met at MechCon that said “eff that dude on the forums asking for engine updates” it was a pleasure meeting you.

Well Imperius, it was a pleasure to meet you at MechCon, but I know it wasn't me against the engine update, I know MWO needs it, too much frankin code was done to CryEngine to make it do things it wasn't meant to do for MWO... But I understand why they did what they did back then... hopefully after MW5 we'll see a new version of MWO on Unreal.

[Bulletsponge0]

Hobbles v, on 22 February 2018 - 07:19 AM, said:

Its not hit reg. Its people not understanding ping. You people are missing your targets because they arent there. If you have over 80 ping the delay is extremly noticable. Remember 100 ping is about 1 tenth of a second in an fps that is an eternity. Mechs can move a full body length in that time span.

I get 18-35 ping on average in NA. I have no hit reg problems there. I end up on euro server with 120 ping i notice i have to lead more and instant hit weapons like lasers become harder to pinpoint with. There is nothing pgi can do to fix this other than split the pop into even more regional servers they dont have ths resources or the population to support that.

HSR is suppose to be PGI's solution to ping (latency) disparities. Weapons simply not registering damage (happens a lot) is a separate issue altogether.

[Johnny Z]

Some are running around invulnerable and others dropping mechs with a touch of their weapons... If I said more I know I would get trouble.

[Mawai]

In general over the years I haven't personally experienced many hit reg issues. Probably two to three cases where I could definitively say something odd was going on.

However, I do have some ideas about why it happens and why PGI has a real challenge fixing it.

1) MWO is server authoritative. This means that the actual location of all mechs and the aim of all shots and resulting hits are calculated ON THE SERVER. Everything displayed on the client is only an approximation of the battlefield and the location of opposing mechs. This is true for your client and every other client in the game.

2) The roundtrip ping time between the server and any client is twice the ping time to the server. The round trip time from one client to another is the total of the ping times for both clients doubled. So if you have a ping of 150ms and an opponent has a ping of 200ms then when one client takes an action it will take at least 0.7s (almost 1s) for that information to get to the other client and back again. In addition, the server has to approximate where one player is aiming and where the target is (which may not match either of the clients display at the time the shot is made).

This is the key point. When you fire at a target on the client ... YOU are not firing at the actual target, you are firing at where the client thinks it should be at the moment you fire. Its actual position is determined on the server. If MWO allowed the clients to determine hits then there would be no problem with hit registration but there would be a HUGE problem with cheating and hacks. (as it is MWO is mostly limited to aimbots that have the same hit registration as players since the aim is based on the client display).

3) In the early days I played a lot of light mechs. You literally had to aim about a mech length in front of the target to obtain a hit if you were in a melee with both lights moving at full speed. Similarly, with the old collision code, the server and both clients calculated different outcomes from the collision so the targets were never where the would appear on the client. It would take a round trip ping time or two (since they probably coordinated both clients) for the fallen mechs to be displayed properly.

4) Host State Rewind was introduced in order to improve hit registration. Host state rewind records where the mechs are on both clients and back up their positions to resolve fire taken at a particular point in time. This helps with lag issues and higher ping times. If ping is constant then the hit registration gets much better since the odds of the target being at the location on the client when you fire at it go up.

My hypothesis is that HSR and resulting hit registration are not resilient to variable ping times and packet losses.

HSR works pretty well for constant ping times. However, what happens if ping goes from 50 to 150ms between data updates to the server? Or what happens when some packets are lost? The server and connected clients have to make some assumptions about where the targets are. However, when the target information gets updated between when you fire on the client and the firing data is received on the server then the server says your shot is a miss because your target wasn't where your client displayed it. When your client receives the position update it corrects the target position but your fire is reported as missing the target.

This could account for the majority of hit registration issues but it is only a suggestion of what might be happening. Only PGI knows what actually happens in the code.

In either case, if this is a fundamental cause of hit registration issues, then I don't see any easy fix. If you assume that the client positioning of the target is correct then the game is open to hacking. Any other assumption will lead to some shots missing due to effectively a desync between where your client draws the target and where that target is deemed to be at the time you fired on the server.

PGI could easily have addressed the issue in the community by being upfront about limitations but, in my experience, they have always chosen the path of least communication.

[Brain Cancer]

Imperius, on 22 February 2018 - 01:30 PM, said:

If only more people backed me when I tried pushing for a new engine update. I was told community warfare updates > keeping this game current.

Oh and the guy I met at MechCon that said “eff that dude on the forums asking for engine updates” it was a pleasure meeting you.

Much like a certain failed crowdfunding, the resources for such an effort are in another giant robot castle.

[Imperius]

Metus regem, on 22 February 2018 - 01:39 PM, said:

Well Imperius, it was a pleasure to meet you at MechCon, but I know it wasn't me against the engine update, I know MWO needs it, too much frankin code was done to CryEngine to make it do things it wasn't meant to do for MWO... But I understand why they did what they did back then... hopefully after MW5 we'll see a new version of MWO on Unreal.

No, it wasn’t you and I didn’t bother to remember his name. He came and interrupted my conversation with George Ledoux. You and I have had our differences in opinion, but it never went to petty insults. I almost was about to tell him oh that’s me nice to meet you. Then again I was there to enjoy myself. Not to be confrontational.

Edited by Imperius, 22 February 2018 - 03:01 PM.

[Bulletsponge0]

Hit reg is as bad today as I've ever seen it

[Relishcakes]

JackalBeast, on 22 February 2018 - 11:36 AM, said:

Not even sure what planet you live on RelishCakes, but it sure as hell isn't this one.
It always amazes me when I see the longest standing T4 player in the game comment about something.
485 games total from season 4 to season 18. just let that sink in. Like no idea, at all.

Strong doubt that I am the longest standing T4 in game. However I am glad you can find my statistics. In defense of these statistics I have only to say this: For a long time I gave 0 ***** about the greatest of statistics and way more about having fun and running whatever build I wanted. I have settled into a build I like and is actually effective as shown by my statistics recently(the last few seasons). I do have to ask though why do you think that a person with ****** stats immediately had no value when their input is brought to the table. I might even go so far as to say you invalidated your own opinion(if ever you decide to weigh in) in this post simply because your first choice was to immediately jump on stats.
TLDR I know my stats suck, I don’t care, don’t be a douche.

[Jackal Noble]

Relishcakes, on 22 February 2018 - 03:09 PM, said:

Strong doubt that I am the longest standing T4 in game. However I am glad you can find my statistics. In defense of these statistics I have only to say this: For a long time I gave 0 ***** about the greatest of statistics and way more about having fun and running whatever build I wanted. I have settled into a build I like and is actually effective as shown by my statistics recently(the last few seasons). I do have to ask though why do you think that a person with ****** stats immediately had no value when their input is brought to the table. I might even go so far as to say you invalidated your own opinion(if ever you decide to weigh in) in this post simply because your first choice was to immediately jump on stats.
TLDR I know my stats suck, I don’t care, don’t be a douche.

I don't care about your stats, either. What I stated had nothing to do about stats.
It was about legitimate experience.

[Jackal Noble]

Why you might ask?
It was directly related to the topic at hand. The more you play the more experience you have, and the more likely you are going to see the HSR issues and discos.
You were like "everything is fine guys, HSR is great."
I know that is not and have seen on multiple occasions as others that is not. That led to wondering how experienced you were with actually seeing what was/is happening in game. That's it Relish, nothing else.

TLDR - Your game stats are fine, you just don't play the game a lot.

Edited by JackalBeast, 22 February 2018 - 03:23 PM.

[Relishcakes]

JackalBeast, on 22 February 2018 - 03:21 PM, said:

Why you might ask?
It was directly related to the topic at hand. The more you play the more experience you have, and the more likely you are going to see the HSR issues and discos.
You were like "everything is fine guys, HSR is great."
I know that is not and have seen on multiple occasions as others that is not. That led to wondering how experienced you were with actually seeing what was/is happening in game. That's it Relish, nothing else.

TLDR - Your game stats are fine, you just don't play the game a lot.

Then I misunderstood the nature of your comment and I apologize. However I DID say i might be wrong but i’ve not seen any issues apart from the occasional double PPC-pass through.

[Jackal Noble]

Relishcakes, on 22 February 2018 - 03:26 PM, said:

Then I misunderstood the nature of your comment and I apologize. However I DID say i might be wrong but i’ve not seen any issues apart from the occasional double PPC-pass through.

No problem. It is just a lot more frustrating at higher levels, when you don't have a lot of room for mistakes only to watch your shots not register and effectively cost you your mech.

[Jackal Noble]

Also, really loving having plus 10 discoes since the last patch. Nothing makes me want to support this game more than getting to send error reports to the new streamlined error reporter, that just closes fast enough for you to not see that it is actually going anywhere.

[MechaBattler]

Imperius, on 22 February 2018 - 01:30 PM, said:

If only more people backed me when I tried pushing for a new engine update. I was told community warfare updates > keeping this game current.

Oh and the guy I met at MechCon that said “eff that dude on the forums asking for engine updates” it was a pleasure meeting you.

"Engine update" would require rebuilding the whole thing. They've modified the original engine so much they can no longer simply update. So really they would have to do a complete rebuild.

I think maybe the long shot we could hope for is that they essentially convert MW5 into MWO2.0. They've already built much of the core mechanics there. And they're going to work on netcode for coop. It doesn't sound like too much of a stretch for them to go the full monty and recreate it as MWO2.0. Plus it'll be easier to use assets from one to the other and swap people working on them around.

[Jackal Noble]

ie just got discoed. again. walk 10 steps and surprise surprise, there are like 3 guys already on my six.

[Diablobo]

I've never really seen anything not register, but I must admit I am doing well just to hit a fast moving small mech, much less track where they hit and do battle damage assessments on the fly. The only real issue I see is how shooting a light mech at full speed with a laser to the torso is a waste of time. Even shooting the legs is problematic because the moving legs tend to spread the damage between the two, especially when shooting from the flank.

I also think that PPCs suck. They move so slow and unless you have a really good connection and a fast puter, you are going to have issues. I also notice that the stream trails that they give off make it look like they should have hit, but because of ping and the optical illusion they seem to pass through. That's pretty much why I don't use them. I don't have any problem hitting with ballistics, but PPCs and I just don't get along.

Edited by Diablobo, 23 February 2018 - 12:38 AM.

[Ghogiel]

Cara Carcass, on 22 February 2018 - 07:32 AM, said:

jup!


notice, that upon reconnectiong of the pilot the hitreg works again.

I just want to point out that some time ago PGI made it so head hitboxes are purposefully disabled on DC'd mechs. In this instance this is not a bug as you can see that the leg shot also works correctly as intended as well as the disabled cockpit.