Jump to content

Why Is Hit Registration So...unreliable?

Gameplay

33 replies to this topic

#21 stjobe

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 9,498 posts
  • LocationOn your six, chipping away at your rear armour.

Posted 23 November 2015 - 12:46 PM

View PostAresye Kerensky, on 23 November 2015 - 10:38 AM, said:

MWO servers run around 30 ticks/s

I'd like to see a citation for this. Not that I necessarily disbelieve it, it's just that I'd like to get confirmation; I've never to my recollection seen the tick rate of the MWO servers discussed officially.

#22 Alexandrix

    Member

  • PipPipPipPipPipPipPip
  • Stone Cold
  • Stone Cold
  • 910 posts

Posted 23 November 2015 - 12:57 PM

View Postburns, on 23 November 2015 - 10:09 AM, said:

Trying out the tutorial academy for the 1st time when it was released i was surprised how easily a few ac10 rounds can kill targets (any mech, rly)! It was a night & day difference!


The mechs in training grounds run stock armor,which isnt maxed out and has way to much allocated to the rear.that's one of the reason the fall over so easily.Most of the mechs people actually use in game are max (or nearly max) armor and have it properly allocated between front and rear.

So if the mwo servers really are running 30 tick,then that would definitely explain a lot.In my experience,even in relatively simple shooters like counter strike,60 tick is the bare minimum for a reliable experience.30 tick really isn't even reliable when you just have one big hotbox and a small number of projectiles to be tracked.

Seriously,upgrade your damn servers.it has its place,but 30 tick is just not good enough for a fast(ish) paced fps game.it's not 1996 anymore.

#23 Cabusha

    Member

  • PipPipPipPipPipPipPip
  • Elite Founder
  • Elite Founder
  • 533 posts
  • LocationAK

Posted 23 November 2015 - 01:00 PM

Hit registration is generally pretty good nowadays, but it will taketh and giveth as it sees fit. I've had pps slam me, flash the armor and do 0 damage. OTOH, last night I had a Thud with bright red internals and slammed an ac20 in. Didn't register and I died to his return fire.

Probably 80-90% of the time it's okay.

#24 C E Dwyer

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 9,274 posts
  • LocationHiding in the periphery, from Bounty Hunters

Posted 23 November 2015 - 01:12 PM

View Poststjobe, on 23 November 2015 - 12:46 PM, said:

I'd like to see a citation for this. Not that I necessarily disbelieve it, it's just that I'd like to get confirmation; I've never to my recollection seen the tick rate of the MWO servers discussed officially.


Maybe a question for the next Townhall ?

#25 JigglyMoobs

    Member

  • PipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 1,445 posts

Posted 23 November 2015 - 01:24 PM

you young ones are so spoiled. back in my day quake servers ran on 24 ticks.... :P

having said this there are a few things you can do client side to somewhat improve hsr.

1. 60 fps locks and vsync are bad. if you do this your driver is probably buffering prerendered frames and costing you an extra ~20ms of lag.

2. check the rendering latency of your monitor. on many 60 hz monitors you are getting significant extra lag, like 30 ms or more.

3. check your nvidia or ati drivers to see the number of prerendered frames. if it's more than 1 you are costing yourself about an extra 10 ms.

between the above 3 the world you see on your monitor has about 100 ms of extra lag than what your ping shows and what the server thinks you have.

then if you use lots of postprocessing, high quality tetextures, etc, your adding additional measurable delays.

after you've killed off delays in the graphics chain you can move on to os and nic settings. things like background intelligent file transfer, Windows search indexing, nic jumbo frames and flow control can all add significant delays.

one way you can measure measure some more f these delays is to run a dpc latency checker on your system. you'll see that most of the time its hovering around 100 ms. you have to carefully configure your system to get it below 50.

so, in addition to the Internet, which is the ping you see on your score tab, the server, your system is also introducing significant delays that the server can't predict or measure.

thus the reality you see is offset from the reality the server sees by your ping + 50 to 300 ms, with large unpredictable fluctuations if you are not careful in managing Windows background processes and cpu power saving state, so of course things will seem weird or unreliable at times.

#26 burns

    Member

  • PipPipPipPipPipPip
  • 272 posts
  • LocationMonerica

Posted 23 November 2015 - 01:36 PM

View PostAlexandrix, on 23 November 2015 - 12:57 PM, said:

The mechs in training grounds run stock armor,which isnt maxed out and has way to much allocated to the rear.that's one of the reason the fall over so easily.Most of the mechs people actually use in game are max (or nearly max) armor and have it properly allocated between front and rear.


hmmm haven´t thought about that. Would be neat to get some actual armor values for the academy targets, then one could experiment a little more - after all the absence of server latency & ticks in academy is still a valuable point in any testing.


Maybe PGI could just revert to 8vs8 as it was in beta (the time when hitreg problems were caused by no HSR and bad ping) - at least for the test server. Someday, they´ll have to get to the bottom of this problem. And with impending steam launch, the strain on their hamsterwheel-powered servers will not lessen in any way on it´s own.

#27 Aresye

    Member

  • PipPipPipPipPipPipPipPipPip
  • Heavy Lifter
  • Heavy Lifter
  • 3,462 posts

Posted 23 November 2015 - 05:59 PM

View PostCathy, on 23 November 2015 - 11:58 AM, said:

I'm quite prepared to believe this as it explains a lot, you have the info/quotes for the doubters ?

View Poststjobe, on 23 November 2015 - 12:46 PM, said:

I'd like to see a citation for this. Not that I necessarily disbelieve it, it's just that I'd like to get confirmation; I've never to my recollection seen the tick rate of the MWO servers discussed officially.


There was a topic titled, "Laser Hit Registration Story," posted by Russ after they relocated to a new data center in August of 2014. There was terrible hitreg (way worse than now) prior to the announcement, and everybody was wondering what the problem was. Russ announced that during the server migration, the servers changed to a default tick rate that was much lower than the 30 MWO is usually run on.

After that announcement, hitreg was improved, pretty much confirming that the tick rate was indeed the culprit.

The post itself was deleted an unknown amount of time ago. Whether or not the post was deleted due to general cleanup or PGI trying to hide the fact they run their servers at an abysmally low tick rate is debatable, but regardless, I was able to scrounge up enough intel through other websites and forums, including a direct quote of the announcement itself. There is enough proof to show that MWO's servers were (and likely still are) run at a tick rate of 30.

Discussions:
https://www.reddit.c...t_registration/
https://www.reddit.c...servers_mwo_is/

Direct Quote:

Quote

Russ Bullock posted:
Here is a brief story for you.

Hit detection a few months ago hits an all new high with the fixing of SRM's along with some bug fixes on laser duration and MG's - everything is great.

We switch data centers at the end of August and forum threads start appearing about poor laser hit registration even though our metrics say everything remains the same and we didn't adjust any of this game code.

Now for this patch we decide to lower the cap time on conquest, during testing we notice inconsistencies in how long it takes to capture. This is when a discovery is made that our servers were no longer running at 30 frames but some windows server default which is much lower.

This was corrected today during the downtime.

Although I have ZERO test data to confirm anything - it seems very logical that this lower server frame rate was effecting our hit registration. Hit detection along with HSR are VERY sensitive to this frame time. ( Neema's words not mine )

So I am hoping that those players who were reporting bad laser hit registration will now report the good news, that things seems all great on the hit detection front again.

Thanks for call it out. Fingers crossed.


Direct Quote Source:
http://forums.someth...0#post436600805

Edited by Aresye Kerensky, 23 November 2015 - 06:08 PM.


#28 Troutmonkey

    Member

  • PipPipPipPipPipPipPipPipPip
  • Moderate Giver
  • Moderate Giver
  • 3,776 posts
  • LocationAdelaide, Australia

Posted 23 November 2015 - 06:24 PM

View PostAresye Kerensky, on 23 November 2015 - 05:59 PM, said:


There was a topic titled, "Laser Hit Registration Story," posted by Russ after they relocated to a new data center in August of 2014. There was terrible hitreg (way worse than now) prior to the announcement, and everybody was wondering what the problem was. Russ announced that during the server migration, the servers changed to a default tick rate that was much lower than the 30 MWO is usually run on.

After that announcement, hitreg was improved, pretty much confirming that the tick rate was indeed the culprit.

The post itself was deleted an unknown amount of time ago. Whether or not the post was deleted due to general cleanup or PGI trying to hide the fact they run their servers at an abysmally low tick rate is debatable, but regardless, I was able to scrounge up enough intel through other websites and forums, including a direct quote of the announcement itself. There is enough proof to show that MWO's servers were (and likely still are) run at a tick rate of 30.

Discussions:
https://www.reddit.c...t_registration/
https://www.reddit.c...servers_mwo_is/

Direct Quote:


Direct Quote Source:
http://forums.someth...0#post436600805

I was around for that. Can confirm hit reg was terrible and then suddenly fixed

#29 Roadbuster

    Member

  • PipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 1,437 posts
  • LocationAustria

Posted 24 November 2015 - 12:59 AM

View PostMister Blastman, on 23 November 2015 - 10:26 AM, said:

Use pulse lasers. GG you win.

AC/20 and PPCs are just asking for stuff to register. Don't even get me started on SRMs. I might as well be picking daisies and throwing them into the wind, hoping they might hurt the target.

Pulse lasers are easy mode. I don't like easy mode.
I have more fun playing uncommon mixed loadouts. ;)

#30 Roadbuster

    Member

  • PipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 1,437 posts
  • LocationAustria

Posted 24 November 2015 - 01:09 AM

View PostJigglyMoobs, on 23 November 2015 - 01:24 PM, said:

you young ones are so spoiled. back in my day quake servers ran on 24 ticks.... :P

having said this there are a few things you can do client side to somewhat improve hsr.

1. 60 fps locks and vsync are bad. if you do this your driver is probably buffering prerendered frames and costing you an extra ~20ms of lag.

2. check the rendering latency of your monitor. on many 60 hz monitors you are getting significant extra lag, like 30 ms or more.

3. check your nvidia or ati drivers to see the number of prerendered frames. if it's more than 1 you are costing yourself about an extra 10 ms.

between the above 3 the world you see on your monitor has about 100 ms of extra lag than what your ping shows and what the server thinks you have.

then if you use lots of postprocessing, high quality tetextures, etc, your adding additional measurable delays.

after you've killed off delays in the graphics chain you can move on to os and nic settings. things like background intelligent file transfer, Windows search indexing, nic jumbo frames and flow control can all add significant delays.

one way you can measure measure some more f these delays is to run a dpc latency checker on your system. you'll see that most of the time its hovering around 100 ms. you have to carefully configure your system to get it below 50.

so, in addition to the Internet, which is the ping you see on your score tab, the server, your system is also introducing significant delays that the server can't predict or measure.

thus the reality you see is offset from the reality the server sees by your ping + 50 to 300 ms, with large unpredictable fluctuations if you are not careful in managing Windows background processes and cpu power saving state, so of course things will seem weird or unreliable at times.

Very informative post, thx! :D
Going to try this.

#31 Cyborne Elemental

    Member

  • PipPipPipPipPipPipPipPipPip
  • 4,000 posts
  • LocationUSA

Posted 24 November 2015 - 01:45 AM

You'll never get PGI to unveil what their Tickrate is, I know I never could get it from Russ or Neema, its a closely guarded secret.

I know BattleField-4 used to run at 10 Ticks/s, which I was fascinated with, because it was absurd for a high tier comp game of that quality.. thats even less Tp/S than Minecraft :o

Which means is probably less 30.

Edited by Mister D, 24 November 2015 - 02:14 AM.


#32 Rhaythe

    Member

  • PipPipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 4,203 posts

Posted 24 November 2015 - 06:31 AM

View PostMister D, on 24 November 2015 - 01:45 AM, said:

I know BattleField-4 used to run at 10 Ticks/s, which I was fascinated with, because it was absurd for a high tier comp game of that quality.. thats even less Tp/S than Minecraft :o

which is probably why simu-kills are so common.

#33 Ghogiel

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • CS 2021 Gold Champ
  • CS 2021 Gold Champ
  • 6,852 posts

Posted 24 November 2015 - 06:43 AM

View Poststjobe, on 23 November 2015 - 12:46 PM, said:

I'd like to see a citation for this. Not that I necessarily disbelieve it, it's just that I'd like to get confirmation; I've never to my recollection seen the tick rate of the MWO servers discussed officially.

Aresye is 100% correct.
They posted about this ages ago when they did a server upgrade and that had reset the server tick rate to default windows setting.

#34 Mister Blastman

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Survivor
  • Survivor
  • 8,444 posts
  • LocationIn my Mech (Atlanta, GA)

Posted 24 November 2015 - 07:46 AM

View PostRoadbuster, on 24 November 2015 - 12:59 AM, said:

Pulse lasers are easy mode. I don't like easy mode.
I have more fun playing uncommon mixed loadouts. ;)


Yup, they are easy mode. But, the only way to get them to fix things is to break the game.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users