BitMinery Forums

Guest


Author Topic: Encouraging Less Laggy Salvager Builds  (Read 238 times)

ErthParadine

Encouraging Less Laggy Salvager Builds
« on: October 11, 2016, 06:51:43 PM »
As much as we all love whole-server salvage/mining related lag, I've been chatting with Schine regarding a few ways to mitigate issues, improve performance, and NOT severely impact build creativity.

WHAT:
In the game's default configuration, you can add as many stacks (groups) of salvage modules as you want, to a single computer. As per Schine's advice, that is changing effectively immediately. A maximum of 1000 groups of salvage modules can be connected to a single computer, there is no adjustments being applied to the maximum number of blocks within a single group.

WHY:
When you start salvaging with 1000's of groups, and they're all connected to a single computer, this causes an issue with the game: 1000's of groups are now trying to update just one or two storage "totals" for each salvaged block type, in a single storage repository. Due to the single-threaded nature of how StarMade handles updates to this data, by trying to brute-force an integer update (instead of queueing updates in a non-blocking way, and then updating the integer via a separate thread), salvaging with such a large array results in blocking all other I/O...which everyone else perceives as lag.

WHINE:
If you really do need such a huge salvager/miner, then add additional computers: this adjustment does not hinder how many salvage modules you can add to a single ship, it just requires that you divide the groups of modules amongst multiple computers (and their respective storage).


Now, everyone say it with me (in a Klingon accent):
     For The Server!
« Last Edit: October 11, 2016, 07:10:41 PM by ErthParadine »

Nacon

Re: Encouraging Less Laggy Salvager Builds
« Reply #1 on: October 11, 2016, 07:51:16 PM »
Sounds like Schine really need to come up with a new way for the beam/cannon to have much wider width pathways instead of single "ray".   Like having explosion effect module as slave but much wider than 1 block width.

Sachys

Re: Encouraging Less Laggy Salvager Builds
« Reply #2 on: October 11, 2016, 08:28:10 PM »

Might be wise to note that beyond 100 blocks length, an individual array gets no benefit / boost / perk / buff - its just over 100m long, no better than a 100m long array.

ErthParadine

Re: Encouraging Less Laggy Salvager Builds
« Reply #3 on: October 11, 2016, 09:08:20 PM »
Might be wise to note that beyond 100 blocks length, an individual array gets no benefit / boost / perk / buff - its just over 100m long, no better than a 100m long array.

Yep, in this case it means your "highest" yield setup is now 1000 groups of 100 block long arrays all connected to 1 computer.

Sachys

Re: Encouraging Less Laggy Salvager Builds
« Reply #4 on: October 11, 2016, 09:51:31 PM »
Might be wise to note that beyond 100 blocks length, an individual array gets no benefit / boost / perk / buff - its just over 100m long, no better than a 100m long array.

Yep, in this case it means your "highest" yield setup is now 1000 groups of 100 block long arrays all connected to 1 computer.

then that makes all my miners perfect. :D

spocklin

Re: Encouraging Less Laggy Salvager Builds
« Reply #5 on: October 11, 2016, 11:35:51 PM »
My miner is in a grouping of like 700 at 100m long...so I should be good then?

ErthParadine

Re: Encouraging Less Laggy Salvager Builds
« Reply #6 on: October 12, 2016, 02:11:32 AM »
My miner is in a grouping of like 700 at 100m long...so I should be good then?

Yep! Haven't noticed any lag from your mining anyways.

They suggested 1000, and after some quick tests on staging, we went with that.

spocklin

Re: Encouraging Less Laggy Salvager Builds
« Reply #7 on: October 12, 2016, 04:02:15 AM »
Yeah, sounds like this will be an ok update...just not the one I wanted them to get done right now.