Jump to content


how to implement 20vs20 or 30vs30 battles, devs please read


  • Please log in to reply
9 replies to this topic

cheapbooks #1 Posted Mar 06 2016 - 12:11

    First lieutenant

  • Players
  • 37195 battles
  • 888
  • [CBKS] CBKS
  • Member since:
    08-21-2013

This is how I think 20 vs 20 or 30 vs 30 matches should be played, taking into account limited server resources.

 

 

 

 

I once saw a clan battle in the Lakeville valley with 15 tanks vs 15 tanks and you could see the lag in the gfx display due to the 30 tanks on the screen at once.

 


 

if server resources are limited, than larger games could be limited only to people who play a premium account, or to people who are playing a premium tank that they paid for. this way, people who pay for services are paying for the additional resources required for larger maps or larger teams.


 

every map uses a certain amount of server resources. therefor, the larger team battles can be limited to maps that use less server resources. the resources used by a map will be a factor of its size, the number of objects, and the line of sight. more objects and larger maps use more resources. however, maps that have lots of sloped land, or objects to block the line of sight, will use less resources because the tanks will not be able to see as far, so the server will not calculate beyond objects or landscaping that blocks the line of sight, which means it uses less resources. so the ideal map for using the least amount of resources will have no objects (no tress, bushes, or houses), and be sloped with hills repeatedly on the map. also, each map should be tested to see how many server resources they use, and the maps that use the least resources can be used in battles with more tanks.


 

a smaller map will use less resources than a larger map, however it increases the chance that many tanks are in the same location which might reduce performance. a larger map, with tanks spread out on multiple spawns, will probably have better performance than a smaller map.


 

every tank uses a different amount of server resources. tanks that reload faster, or drive faster, or have a farther viewing distance, will use more resources. they need to have calculations done more frequently. slow tanks like SPGs and heavy tanks will use fewer server resources because they fire less often due to long reload times. tanks that don't move much will have fewer updates made with the server in regards to their position. so determine which tanks use the least amount of server resources, and use them in the battles with more than 15 players per team.


 

most programmers use poor programming style in these types of games. here is how most programmers calculate per cycle, and how WOT probably does, and why it is wrong:

1. step A (highest priority calculations)

2. step B

3. step C (lowest priority calculations)

4. repeat, starting at 1


 

Here is how it should be done:

1. step A (highest priority calculations)

2. check elapsed time, return to 1 if too much time elapsed

3. step B

4. check elapsed time, return to 1 if too much time elapsed

5. step C (lowest priority calculations)

6. repeat, starting at 1


 

the second is a better method because if the server load gets to high, then the game will still play, however with diminished features. the highest priority calculations are always done first, and the lowest priority calculations are done last.


 

for example, collisions use a lot of server resources in a 3d environment. a smart programmer will turn collisions on and off to save resource. most 3d programmers would check collision every cycle, but it is not necessary. they can be checked every other cycle for example. or it can dynamically change based on server load. or you can check collision for ammo fired more often than you check collisions for tanks hitting other tanks, or for tanks hitting houses. this improves performance.


 

every cycle, the server has to calculate spotting, shots fired, collisions, and object movement. performance can be increased by limiting spotting ranges, or by performing calculations less often. for example, if you are checking spotting every cycle, you could do it 4 out of every 5 cycles instead, or 1 out of every 2 cycles instead and increase performance. as I explained earlier, it should change dynamically based on server load or other needs.


 

before, there were no dynamic gfx changes in the game. now there are. this is an example of dynamic programming to save resources. but it needs to be done throughout the game, not just the client side display. most games sadly do not do any dynamic adjustment whatsoever.


 

so using my method, if your tank can see 500 meters, and the server goes under heavy load for a few seconds, the view range might be dynamically reduced to 450 meters until the load returns to normal, which might be only a split second, and the user would never know


 

15vs15 battles will use less resources than 30vs30. if the server does 10,000 battles per day, you could allot 90% to 15vs15 and 10% to 30vs30. or, you could only run the 30vs30 games when there are fewer people on the server, which happens overnight when most people are sleeping. I see at most 30,000 players on the US servers and as low as 5,000. run the 30vs30 games when there are only 5,000 people on the servers, and the servers should be able to handle it no problem.


 

we all know what 15 vs 15 is like. 30 vs 30 is a 2x increase in the number of tanks. however 20vs20 is only a 1.3 times increase. if 30vs30 is too much, why not 20vs20, and add extra arty to keep resources low? I am sure that the current servers can handle games that are a bit larger than they currently are.


 

more spotted tanks increases server resources. less spotted tanks uses less server resources. make tanks become unspotted more quickly when server usage is high.


 

passmark can help make the decision as to which server to buy. however, the numbers are misleading. if 2 processors have the same passmark score (lets say 10,000), but different number of cores, the one with the least number of cores will be faster. a cpu with passmark 10,000 and 1 core has a core that can achieve speeds of 10,000. a cpu with passmark of 10,000 and 2 cores has 2 cores that can only achieve 5,000 per core. the first would operate faster because the core is faster, and applications tend to dominate on 1 core.


 

also, all motherboards are not created equal. back in 2000 there was a web site that measured the speed of motherboards using the same CPU, and around 90% of the top 100 were ASUS. even the floppy drives were faster than other motherboards. that's why I was a loyal ASUS customer for many years.


 

people who pay for the game should have the right to pay for better services. so you could limit the larger team battles to people who pay for premium accounts, or for people who are playing in a premium tank for the battle, or charge a small amount of gold (such as 50) to join the battle. this will also give the incentive to more players to pay for premium accounts or to buy premium tanks in order to get the opportunity to play the larger team battles.


 

server resources are probably different for different tiers. a tier 10 battle might use different resources than a tier 4 battle. limit the larger teams to tiers that use less server resources.


 

so my recommendations to the devs for implementing battles with more than 15 tanks per team:

- run the larger team battles during off-peak times, such as overnight

- use maps that use less resources, or design new ones

- use tanks that use the least resources (slow, long reload)

- use smaller incremently increase in team sizes (20vs20 or 25vs25)

- make tanks become unspotted more quickly

- limit larger team battles to tiers that use less resources per battle

- change the code to be dynamic such that it has a certain amount of time to complete each task per cycle, therefor not every task is calculated per cycle

- limit it to players playing on a premium account or in battle with a paid premium tank

- use the fastest SSD drives on a RAID, the most RAM, and fastest multicore servers with best passmark score per core (not overall)

- use the fastest internet connection (1 gigabit instead of 100mbit for example)

- use CPUs that have fast benchmarks per core, not per CPU

- use motherboards and hard drives that have the fastest benchmark scores

- upgrade the server hardware annually to fastest technology


 

please forward this to your developers.


Edited by cheapbooks, Mar 06 2016 - 12:13.


SqU4D3Nn #2 Posted Mar 06 2016 - 12:26

    Staff sergeant

  • Players
  • 18565 battles
  • 383
  • [T-GBU] T-GBU
  • Member since:
    07-26-2013
It's not only problems with server resources, it also puts A LOT of strain on the BigWorld Engine, not only that, people who have potato PCs will not be able to play those battles..... because of potato PCs..

zed2204 #3 Posted Mar 06 2016 - 12:43

    Major

  • -Players-
  • 42035 battles
  • 2,183
  • [LIO] LIO
  • Member since:
    11-09-2014
Easiest solution is respawn, 30 tanks at any given moment but with 1 respawn total number in battle goes to 60 without the need any changes or more people in queue
And larger maps since there will be more tanks to cover it
I don't think FPS is that much of an issue, seeing all 15 tanks at spawn don't bring FPS down at least for me, and with the draw limit you wont see all 30-60 even if they are all spoted, besides distance make models decrease in quality so less load


t8z5h3 #4 Posted Mar 06 2016 - 12:49

    Captain

  • -Players-
  • 39467 battles
  • 1,122
  • [FUSON] FUSON
  • Member since:
    04-23-2013

Everything your saying "seems" solid but your though process is completely wrong 

1. Every match uses the same amount of resources 

2. The map size is to small for larger amount of players

3. You can't jest toss resources at a problem and expect it to be fixed (let's get dull 16 core processors to run a single threaded program that will be better then a single processor with the same amount ghz) (it not that simple)

 

 



Jr_Jr #5 Posted Mar 06 2016 - 13:40

    Staff sergeant

  • Players
  • 22695 battles
  • 452
  • Member since:
    09-17-2013

Or just keep 15v 15. Why would you want 30v 30?

A single player would have far far less sway. So carrying is almost impossible.

They would have to raise arty cap.

Cap to win during a match were you 100% don't need to cap would be annoyingly more common ,and harder to stop.

Not a single map could handle that many people, and remain fun to play.

 



5ixty5ix #6 Posted Mar 06 2016 - 14:14

    Staff sergeant

  • Players
  • 25754 battles
  • 304
  • [-UT-] -UT-
  • Member since:
    08-10-2011

Yeah...can't wait to start getting blown out 30-1. Those old 15-1 games are getting boring.....

 

Sorry man. I just can't see it happening. Teams can barely co-ordinate 15 players per side, the entire award system would have to be reworked, and with the NA server size you'd play against the same people every 3rd battle. 


Edited by 5ixty5ix, Mar 06 2016 - 14:18.


SanguTik #7 Posted Mar 06 2016 - 14:36

    First lieutenant

  • Players
  • 23200 battles
  • 525
  • Member since:
    11-28-2013

I'd actually prefer to see smaller teams of 12v12 or 10v10 where a player had more impact on the game and could carry easier. It would make the game more dynamic due to more space to maneuver and less overcrowding on certain maps. It also would let MM put matches together faster and let the matches be more balanced. If you want teams that large go back to Battlefield.

 



Blue_Prints #8 Posted Mar 06 2016 - 14:44

    Sergeant

  • -Players-
  • 7883 battles
  • 106
  • Member since:
    11-20-2014

View Postzed2204, on Mar 06 2016 - 05:43, said:

Easiest solution is respawn, 30 tanks at any given moment but with 1 respawn total number in battle goes to 60 without the need any changes or more people in queue
And larger maps since there will be more tanks to cover it
I don't think FPS is that much of an issue, seeing all 15 tanks at spawn don't bring FPS down at least for me, and with the draw limit you wont see all 30-60 even if they are all spoted, besides distance make models decrease in quality so less load

 

i like this idea because if you are in a giant heavy on a open map or a long range td in full city you can switch 

IanSanJr #9 Posted Mar 06 2016 - 15:12

    First lieutenant

  • Players
  • 63546 battles
  • 792
  • [OTTER] OTTER
  • Member since:
    09-20-2011

15 v 15 

each team get 3-5 arties.

 

30 vs 30

each team get 5-8arties.

 

I will kill my self or quit play game

 

f*** arties. I feel like 1 tank drive toward town to enhance cover then get focus fire by arties and died.



lordsheen #10 Posted Mar 06 2016 - 15:22

    Corporal

  • -Players-
  • 25194 battles
  • 64
  • [SOYUZ] SOYUZ
  • Member since:
    07-22-2014
i think 300 vs 300 with everyone having 5 lives(respawns) would be more enjoyable than 30 vs 30




1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users