44 replies to this topic

[chewie]

Me: Ping 90, wolvie 6K, 3 large lasers.
Him: ping 248, Cicada.

I hit him across the torso (from the rear) tracked and saw the shots hit on my screen.... doll never flashed, he just ran off into the distance behind some rocks (and into his team mates support)

I back up around outcropping of rock.

he comes flying back around, hits my ct with his medium lasers.

Only my ct flashes right before I die. He's still flying along with the same damage he had (before I hit him, yet HSR says i didnt....)


F.M.L.



Its frustrating to say the least.

[RedlineHunter]

I have noticed a lot of problems with shots not hitting where they are suppose to. *I get shot in the LA from front and I take LA and rear RT damage*

[Murphy7]

My computer is a lap top that has trouble maintaining 30 FPS in the game, but I do tend to "enjoy" a ping in the high 20's and low 30's most games. Pre-patch, this meant being behind cover and getting mystery damage to appear on my mech - and the joy of coming around a corner to be obliterated by mechs that were just then drawn for me.

Now at least when my computer tells me I am behind cover, other people's home states aren't offering me up on a platter all the time.

[GroovYChickeN]

HSR is a funny thing. It really is a fine line between working and not.

First: This is by no means a complete overview. It's more of a rough generalization for the sake of discussion. And no matter how much tuning they do to HSR, there will never be a perfect fix.

Remember that the server has to keep track of every players position and orientation as viewed from the server side. It will only keep track of a few seconds of historical data to use in HSR calculations at best (a 1 second desync would be 1000 milliseconds. I would guess that average ping is around 200 - 500ms). If you are too far out of sync from the server then unregistered hits can/will still happen.

To counter act this we can just keep more historical data right? Sure! Although, if the server keeps too much HSR data it can have a negative impact on performance causing server side lag and thus unregistered hits and other funny business can occur. (we could get into things like how the data is store and called and cache misses and blah blah blah but we are going to skip the super technical)

There are also the issue of HSR keeping data for too long and handling it perfectly. This can cause funny issues for some low ping players. This is what is usually going on when someone complains about laggy hitboxes in FPS games, and EVERYONE hates it when this happens. So in order to keep the LPB's happy PGI has to keep this in mind.

There are even more issues with HSR that can pop up depending on the operating cycle of the server vs the players client. If your game is running at 130fps on a 144hz monitor that doesn't necessarily mean that the info is being sent to the server at the same rate or even that the server is processing it at the same rate. If what you are seeing is running at a refresh rate different than how often the data is being sent back and forth between your client and the server, then you can again see some really odd behavior (One shot kills, dying with out seeing the other player shoot, laggy hitbox syndrome, ect...). Now I don't know how PGI is doing it but DICE had this problem with BF4 and only recently "fixed" it by tying your clients update cycle to your fps.

Add on top of this any client side tricks (extrapolation or interpolation being the common ones) that PGI is using to smooth our your gameplay.

I have a very strong feeling that the problems with HSR and hit detection in MWO are not a single issue but rather a combination of may different issue (some of them I talked about, many I did not).

Anyway as stated above, there is not perfect solution for online games. There will always be problems no matter how they choose to deal with client/server latency.

TL:DR - Dealing with lag is hard and there are no easy answers.

Edited by GroovYChickeN, 13 July 2015 - 01:21 PM.

[stjobe]

Shadow Magnet, on 13 July 2015 - 11:20 AM, said:

I would gladly do that I would know that some dev would actually look at the video.

But I think PGI has all the information they need, the server logs of every match. I have a hard time to believe that they can't write a program that evaluates the data automatically. Just scan for fired alpha strikes in the data and then check how much of the damage actually was applied to the target.
Then compare a period of tuesday after the patch against some period of the same length on the weekend. There must be a large difference in the average number of applied damage vs fired damage, I will bet a Wave 3 Executioner package

I'd like to draw your attention to this thread.

In it, a player argues that missiles are doing too much damage. He had numbers, test results, but it wasn't before people started posting videos that the devs perked up - and, lo and behold, he had discovered a bona fide bug in the way splash damage worked. Not only for the streaks he was testing, but for all missiles. SRM's did up to 19 points of damage each, due to the way the splash damage sphere interacted with the hitboxes.

The result was a hot fix, then another, then some rebalancing, and a year later or so we finally got not only pin-point damage for all missiles, but improved hit-registration as well.

So yeah, video does make all the difference.