Jump to content

Archived

This topic is now archived and is closed to further replies.

Yeatle

So are we going to ignore the horrible tick rate and netcode?

Recommended Posts

It's disgusting that the developers have allowed the game to have such terrible tick rates for so long. Every cam I watch shows my character 5 or more feet away from where i really was. Especially infuriating when I take cover only to be killed because the game shows me out in the open to the other player. 

 

I assume it's due to laziness that it hasn't been fixed yet

Share this post


Link to post
Share on other sites

The replay system is and has never really been a reliable source of information. Don't think it was even designed to be accurate. More of an example. I always watch my death cam and occasionally a replay if needed. They've never been an accurate display.

 

Share this post


Link to post
Share on other sites

I don't think it's because of laziness. More like incompetence. I remember reading that the vast majority of the devs have most of their experience in developing mobile games. That's why it took so long for the game to become playable on average rigs.

Share this post


Link to post
Share on other sites
2 hours ago, Yeatle said:

It's disgusting that the developers have allowed the game to have such terrible tick rates for so long. Every cam I watch shows my character 5 or more feet away from where i really was. Especially infuriating when I take cover only to be killed because the game shows me out in the open to the other player. 

 

I assume it's due to laziness that it hasn't been fixed yet

 

I think you need to read up on how the replay system actually works.

Share this post


Link to post
Share on other sites

@Yeatle @BareNekkedd

 

It's about resources. Mainly CPU, but also RAM. Increasing tick means you have to increase CPU resources.

 

60Hz Tickrate with 100 players equals 6,000 packets a second the server has to process. One per player every 16.7ms. At a 1:1 client update rate that would mean 6,000 updates a second being sent back out. So 12,000 player movement only related processes.

 

Every 16.7ms the server would have to process an incoming packet from every player, calculate historical movement, update historical positions in the sim timeline, calculate current position based on past and new data. Then send each player an updated view of the world.

 

Add projectiles (spawning/pooling, physics etc), Looting, Loot spawning/pooling, inventory management, vehicle interaction/spawning, doors, glass etc on top of those 12,000 movement processes.

 

Now lets increase the base requirements for the player. You'll need more CPU and Ram to bundle, process and send 60 packets plus more cpu and ram to receive and process 60 from the server. That's not counting all the RPC calls that aren't bound to tickrate. Shooting, Looting, Inventory management, Health management, door and window interactions for example. When a player picks up something you need to know immediately so it can be removed from your version of world space. You can't bundle that data with a standard movement update.

 

There's also the fact that when you increase tickrates you have to look at the effects of high latency, loss and variance. Yes a high tick will result in greater precision in movement calculations. But the calculations margin of error increases exponentially with higher player latencies. This increases CPU usage on the server. It has to correct and update past calculations. The higher the player ping, the more resources and time needed by the server to correct their position in world space.

 

Please learn a bit about a topic before you post argument and disgust. It's pretty hard to take an argument on Tickrates seriously when you don't know WTF tickrate is or how it works.

Share this post


Link to post
Share on other sites
23 minutes ago, Rev0verDrive said:

 

Please learn a bit about a topic before you post argument and disgust. It's pretty hard to take an argument on Tickrates seriously when you don't know WTF tickrate is or how it works.

 

Bluehole should maybe then consider smaller maps with fewer players, for instance 60 like Apex, based on your explanation how tickrates works. I mean, 100 players is really just an arbitrary number and doesnt meant a better gameplay experience per se. I personally dont care what "other" players on the map experience during a match of Pubg, so if they manage to balance the map size, phase- and circle speed and loot etc, you could achieve exactly the same individual experience with less players on the map. During a normal Apex match, assuming that you eventually win, I usually experience 4-6 fights, which is almost more than a normal Chicken dinner match on Sanhok etc. 

Share this post


Link to post
Share on other sites
28 minutes ago, Baboopa said:

 

Bluehole should maybe then consider smaller maps with fewer players, for instance 60 like Apex, based on your explanation how tickrates works. I mean, 100 players is really just an arbitrary number and doesnt meant a better gameplay experience per se. I personally dont care what "other" players on the map experience during a match of Pubg, so if they manage to balance the map size, phase- and circle speed and loot etc, you could achieve exactly the same individual experience with less players on the map. During a normal Apex match, assuming that you eventually win, I usually experience 4-6 fights, which is almost more than a normal Chicken dinner match on Sanhok etc. 

 

You'll never see BH drop the player count. It's one of those carved in stone game selling/marketing features that sets it apart from others.

 

They should've started at lower numbers and worked up. Like 50/60Hz tick, 1:1 client update rate and then profiled map size, player count, loot numbers etc. It's what I'm doing. Current goal is 64p, 60Hz tick with 1:1 update rates (client replication) out to 150m or so (replication interleave), then a soft to coarse degrade. I don't need high update rates for players at 600m or more. 10-15Hz is fine.

Share this post


Link to post
Share on other sites
1 hour ago, Rev0verDrive said:

 

You'll never see BH drop the player count. It's one of those carved in stone game selling/marketing features that sets it apart from others.

 

They should've started at lower numbers and worked up. Like 50/60Hz tick, 1:1 client update rate and then profiled map size, player count, loot numbers etc. It's what I'm doing. Current goal is 64p, 60Hz tick with 1:1 update rates (client replication) out to 150m or so (replication interleave), then a soft to coarse degrade. I don't need high update rates for players at 600m or more. 10-15Hz is fine.

Wouldn’t lower tick rates at distance affect long range weapon viability? Or would the reduction in tick rate only drop off outside of what would be considered viable weapons range?

Share this post


Link to post
Share on other sites

I think that there are classes so people always get drawn at the best possible tic, but other things might have much lower updates depending on how far away they are.

Share this post


Link to post
Share on other sites
5 hours ago, Yeatle said:

It's disgusting that the developers have allowed the game to have such terrible tick rates for so long. Every cam I watch shows my character 5 or more feet away from where i really was. Especially infuriating when I take cover only to be killed because the game shows me out in the open to the other player. 

 

I assume it's due to laziness that it hasn't been fixed yet

 

You clearly have no clue how tick rate works.

Share this post


Link to post
Share on other sites
9 minutes ago, umpem said:

 

You clearly have no clue how tick rate works.

or in fact, what it is

Share this post


Link to post
Share on other sites
18 minutes ago, DeadlyEggXoP said:

I think that there are classes so people always get drawn at the best possible tic, but other things might have much lower updates depending on how far away they are.

Yep, but @Rev0verDrive noted the drop off for players themselves... meh. Educate me. Lol. I have no idea...

Share this post


Link to post
Share on other sites
10 minutes ago, theexecutioner said:

Wouldn’t lower tick rates at distance affect long range weapon viability? Or would the reduction in tick rate only drop off outside of what would be considered viable weapons range?

 

Replication interleave has it's pros and cons. The pro side is you can increase update rates the closer you get to another player. The con side is long range interactions "can" take a hit. The only crap part is rapid peak engagements. Yet those can avoided if you use a frustum approach in the replication interleave.

 

Have your normal radial segments, but also have a focal cone ... aka player FOV. When you scope up the FOV narrows. you could tie into this to give high update rates further out based on ADS and a Focal cone.

 

This is what DICE LA did for Battlefield 4. They used a multi-hemisphere approach coupled with and FOV cone.

 

Here's BF1's version.

 

ghayrqZ.jpg

 

Any players at 200m or more are only updated at 3Hz... That's once every 333.333ms. Yet traditionally the vast majority of interactions (gun fights) happen within 30m or closer. But that's Battlefield. Not sure what the numbers are for PubG. Would depend on the player and their style.

Share this post


Link to post
Share on other sites
Just now, Rev0verDrive said:

 

Replication interleave has it's pros and cons. The pro side is you can increase update rates the closer you get to another player. The con side is long range interactions "can" take a hit. The only crap part is rapid peak engagements. Yet those can avoided if you use a frustum approach in the replication interleave.

 

Have your normal radial segments, but also have a focal cone ... aka player FOV. When you scope up the FOV narrows. you could tie into this to give high update rates further out based on ADS and a Focal cone.

 

This is what DICE LA did for Battlefield 4. They used a multi-hemisphere approach coupled with and FOV cone.

 

Here's BF1's version.

 

ghayrqZ.jpg

 

Any players at 200m or more are only updated at 3Hz... That's once every 333.333ms. Yet traditionally the vast majority of interactions (gun fights) happen within 30m or closer. But that's Battlefield. Not sure what the numbers are for PubG. Would depend on the player and their style.

Yeh I see. Thx that makes sense. Must be a bitch to implement tho... you’d have to dynamically update tick rate as the player moves the scope in ADS, tracking scope multipliers and magnification... but I see the advantages. Neat sln.

Share this post


Link to post
Share on other sites
11 minutes ago, theexecutioner said:

Yeh I see. Thx that makes sense. Must be a bitch to implement tho... you’d have to dynamically update tick rate as the player moves the scope in ADS, tracking scope multipliers and magnification... but I see the advantages. Neat sln.

 

The engine does a lot of this already...natively. The Replication Graph adds a hell of a lot to it. This was added in version 4.20 ... I'm running 4.21 currently and 4.22 just released.

 

Here's a feature highlight and overview of the replication graph ... Epic Games Live Stream. If you're interested.

 

Very heavy focus on BR and High player count game modes.

 

 

 

If the above vid doesn't jump to the right time, the main part starts at the 7:18 point.

 

Relevancy and priority ... Determines when and what actors get replicated to you.

https://docs.unrealengine.com/en-us/Gameplay/Networking/Actors/Relevancy

 

Share this post


Link to post
Share on other sites

I believe PUBG doesn't use the latest version though do they?

 

@Baboopa keep in mind that player count is not everything. Apex legends like you noticed has 60 players and smaller maps, yet their tickrate and netcode is actually worse than PUBG. With the current infrastructure PUBG has they probably could drop down to 60-64 players, which is something that has been done for along time. 100 players was something that was strived for in such games for a long time and apart from games like Planetside you really were limited to ~64 players in a single match. I remember Project Reality, a mod for Battlefield 2, tried to go 100 players. Servers didn't manage to keep up...

Share this post


Link to post
Share on other sites
33 minutes ago, HellDuke said:

I believe PUBG doesn't use the latest version though do they?

 

@Baboopa keep in mind that player count is not everything. Apex legends like you noticed has 60 players and smaller maps, yet their tickrate and netcode is actually worse than PUBG. With the current infrastructure PUBG has they probably could drop down to 60-64 players, which is something that has been done for along time. 100 players was something that was strived for in such games for a long time and apart from games like Planetside you really were limited to ~64 players in a single match. I remember Project Reality, a mod for Battlefield 2, tried to go 100 players. Servers didn't manage to keep up...

 

I think they're still around v4.16.3. To get to 4.22 or even v4.20 they'd have to rewrite and retest (Refactor) a mega shit load of code. It's easy enough to simply clone a project up to the latest version (a few clicks actually). But the problem is a lot of things get deprecated and or heavily changed version to version. Like entire classes being removed and replaced by new classes with different names and required inputs.

 

...

 

I ran a server with project reality. I also ran a BFV PoE server. Pushed PoE up to 80 once, but I had to massively increase the cpu and ram on the box. Was going through Leet Servers and Branzone at the time. Shit wasn't cheap. Around $120 a month.

Share this post


Link to post
Share on other sites
3 hours ago, Rev0verDrive said:

 

The engine does a lot of this already...natively. The Replication Graph adds a hell of a lot to it. This was added in version 4.20 ... I'm running 4.21 currently and 4.22 just released.

 

Here's a feature highlight and overview of the replication graph ... Epic Games Live Stream. If you're interested.

 

Very heavy focus on BR and High player count game modes.

 

 

 

If the above vid doesn't jump to the right time, the main part starts at the 7:18 point.

 

Relevancy and priority ... Determines when and what actors get replicated to you.

https://docs.unrealengine.com/en-us/Gameplay/Networking/Actors/Relevancy

 

Woooah. I gotta look at that again when my brain ain’t so full. Great to see the devs talking to it tho.

Share this post


Link to post
Share on other sites
18 hours ago, Rev0verDrive said:

You'll never see BH drop the player count. It's one of those carved in stone game selling/marketing features that sets it apart from others.

 

I don't necessarily disagree with your statement, but they have budged on that number...at every tournament. They're vision is carved into stone until it isn't.

 

Are you developing a game?

Share this post


Link to post
Share on other sites
5 minutes ago, tacsandstrats said:

 

I don't necessarily disagree with your statement, but they have budged on that number...at every tournament. They're vision is carved into stone until it isn't.

 

Are you developing a game?

 

No other BR, "That I'm aware of" has 100 players on a 8x8KM map. It's a huge selling point. Lowering player counts would require massive reworks to the BR maps. 64 square kilometers is a lot of space for 100 players.

.....

 

Yes, I'm developing a game in UE4. I was literally waiting on the release of v4.21 for the "production ready" Replication Graph. 👍

Share this post


Link to post
Share on other sites
On 2/19/2019 at 3:21 PM, Rev0verDrive said:

 

No other BR, "That I'm aware of" has 100 players on a 8x8KM map. It's a huge selling point. Lowering player counts would require massive reworks to the BR maps. 64 square kilometers is a lot of space for 100 players.

.....

 

Yes, I'm developing a game in UE4. I was literally waiting on the release of v4.21 for the "production ready" Replication Graph. 👍

 

Pubg is unique in that it's 100 players on large 8x8 km maps, and I think this gets overlooked when comparing BR games. Pubg is literally doing something no one else is, so their shortfalls are unique. BUT their tournaments are 64 players on the 8x8km maps, so they have budged on their selling point, and they haven't reworked the maps to make it work.

 

My point is only that any argument from them (internally) to draw a hard line at 100 players is weak considering they do it already. Eventually the game player base will naturally decrease, and they will have to make hard decisions, such as dropping modes, maps, or player count. (A map doesn't necessarily have to be redesigned. Have a circle started at the plane drop phase.)

Share this post


Link to post
Share on other sites

The problem is other games do have better server code and do easily handle large numbers of players.

Just ensuring nearby players are actually displayed correctly would greatly stop me ALT-F4ing every couple of hours.

The number of times I've been 360'd by somebody when they walk through the door on my screen is really getting on my nerves.

Share this post


Link to post
Share on other sites
On 2/18/2019 at 11:25 PM, HellDuke said:

With the current infrastructure PUBG has they probably could drop down to 60-64 players

 

Just noting, that by 7-10 minutes into most matches the player count is already down to about 60 people.

 

As for 64 players in a tournament, I don't think they do this for performance (pretty sure they are LAN based) as for giving teams space/time to gear up.  The pro players almost never go for early drop engagements, about the only time you see that is when two teams decide to drop in the exact same spot. Otherwise, everyone tries to avoid each other because alot of the pro scene is about placement. I don't think you can use PUBG tournaments to support a reduction in player count.

Share this post


Link to post
Share on other sites
5 hours ago, Armastadt said:

 

Just noting, that by 7-10 minutes into most matches the player count is already down to about 60 people.

 

As for 64 players in a tournament, I don't think they do this for performance (pretty sure they are LAN based) as for giving teams space/time to gear up.  The pro players almost never go for early drop engagements, about the only time you see that is when two teams decide to drop in the exact same spot. Otherwise, everyone tries to avoid each other because alot of the pro scene is about placement. I don't think you can use PUBG tournaments to support a reduction in player count.

 

An excellent point, there are many games you can play smoothly at 60 players. So there clearly are issues with how the game handles stuff.

Share this post


Link to post
Share on other sites

×
×
  • Create New...