So I did some data gathering on another type of farm: large pumpkin farms. With the large amount of redstone being activated, the immense numbers of pistons to be processed, and the item entities from the pumpkins, these can have the potential to cause lag if one is not careful.
The farm I selected to be tested is the Mountain Dwelling pumpkin farm. In order to measure the impact of a farm without having Large amounts of terrain/10+ players to simulate activity, I selected a different metric for my tests: MSPT.
MSPT, or Milliseconds Per Tick, is a measure of how long the server took to process a given tick. if it exceeds 50 mspt, the server will miss the time window for the tick to execute, and the tps will decrease.
In order to accurately measure the farm without spending large amounts of time testing, I elected to install a toolkit known as the carpet mod. This mod adds numerous functionality specialized toward in-depth farm development, one of which is the "tick warp" function. The Tick warp removes the 20 tick maximum restriction and permits the game to run as fast as possible to simulate much more gameplay with significantly less time.
To Isolate specifically the farm, The farm was WE'd into a void world, meaning that there was quite literally nothing else around the farm to incur mspt penalties. Everything not pertaining to the farm itself was also removed, and items collected were directed into lava. As such, this test does not measure the following: The impact of hoppers from the item storage, The impact of surrounding terrain, the impact of the other farms near mountain dwelling. The test focused on the largest farm in the complex.
The Tests were performed using my i5-8400 CPU, with 4 GB of allocated RAM. GPU does not affect internal server performance.
The Tests, and results, are as follows:
Test 0: Void
For this test, I moved to a location without the farm, with quite literally nothing for at least 1k blocks in any direction. This test establishes a baseline mspt for the void world.
Test Duration: Simulated 72000 ticks (1hr)
Results: Server ran with average mspt of .2, speeding through the test at a blazing fast 5122 tps.
Test 1: Farm on rapid clock speed
For this test, I configured the farm with an Ethonian Hopper clock, with a period of 51.15 seconds (1 stack of items). I did run this and subsequent tests for a lesser amount of time such that I was not waiting for the data all day.
Clock: Ethonian Hopper Clock (Period = 51.15 seconds)
Test Duration: Simulated 12000 ticks (20 mins)
Result: Server ran with average mspt of 12.01, considering baseline mspt, we can deduce that the farm on a high clock speed "costs" approximately 12.01 mspt on average.
Test 2: Farm on slower clock speed
For this test, I configured the clock to run with a period of 4min16s, the max an ethonian hopper clock can run for.
Clock: Ethonian Hopper Clock (Period = 256seconds (4m16s))
Test Duration: Simulated 12000 ticks (20 mins)
Result: Server ran with average mspt of 6.22, with baseline mspt, we can deduce that the farm on a high clockspeed "costs" approximately 6.02 mspt on average.
Test 4: Farm off
For this test I disconnected the clock completely and investigated how the server handled the farm on idle
Clock: None
Test Duration: Simulated 12000 ticks (20 mins)
Result: Server ran with average mspt of 2.44 mspt "cost": 2.24 mspt
From these results, it is evident that the clock frequency has a large effect on the mspt "cost" of the farm, incurring heavy penalties over when the farm simply idles, and heavier when the clock cycles more frequently. This is indicative of the fact that the largest mspt cost occur when the farm actually fires. Hence, I performed Test 4.
Test 4: When Firing
For This test, I started the warp after pressing a button to fire the farm once. The duration was calibrated as best as I could to only include the segment of time in which the pistons were extending on the farm.
Clock: Manual Activation
Test Duration: 400 ticks (20 seconds)
Result: Server ran with an average of 24.84 mspt, indicating that the firing sequence of the farm costs 24.64 mspt, for its short duration.
between tests 3 and 4, I think its evident that while a pumpkin farm does not cause increases in mspt most of the time, while the farm fires there is a sharp, sustained, increase in mspt, which if the server mspt is already abnormally high, can cause great lag for players on server. If the server is already running borderline, at 50 mspt, an increase of 20mspt can already decrease tps below 15. this decreased tps would also increase the amount of time until the farm completes its firing cycle, as they are usually redstone activated, which is tied to the tps.
Error Sources: This data is not nearly of the same quality of the data provided by luis in former posts. I don't have the ability to easily pull instantaneous mspt data into a spreadsheet/graph. Furthermore, the server has differing hardware than I do, hence this data is not indicative of actual server impact.
Conclusion: As I have neither any way of simulating the actual VC server, both in back end, and specific hardware, the absolute numbers in this data should not be compared to the actual server. Instead, This data demonstrates the relative impact that a pumpkin farm like mountain dwelling can have on mspt. The relative differences between the mspt data demonstrates that farms like Mountain Dwelling have large impacts on server performance specifically when they are firing and harvesting the pumpkins, and significantly reduced impact when idle.