client to server login time

[Festus1965]

didn't find a similar Thread with search "login time"

beside a crazy difference of server starting time with the mod craftguide (5:03 to 2:20)

I measured in a blink of idea some of my machines from press login until the player is in the world.
"blink" - remember my daughter always so angry that I am faster in that her ...

added:
* Always on my own same server "Asia Thailand" just today, so same mods, ...
* Always with fresh restarted and first login to same own server

Code: Select all

CPU	RAM	GPU		storage	connection	- system -	mtc	loading-time
i7-4770	32 GB	GT 710 / 2 GB	M.2 2800	1 GB	Ubuntu	wine win-64	27-30 sec
i7-4770	32 GB	GTX 960 / 2 GB	M.2 2800	1 GB	Ubuntu	Ubuntu		18 sec
i7-4770	32 GB	GTX 960 / 2 GB	M.2 2800	1 GB	Ubuntu	wine win64	29 sec
i5-3470	4 GB	GT 730 / 2 GB	ssd 550		1 GB	win7-64	win64		1:05
i3-2100	16 GB	GT 710 / 2 GB	hdd		1 GB	Ubuntu	Ubuntu		19 sec
E7500	6 GB	GT 710 / s GB	hdd		1 GB	Ubuntu	Ubuntu		25 sec
J2900	8 BG	intern		ssd 550		1 GB	win7-64	win64		1:36
E1-1200	4 GB	intern		hdd		1 GB	Ubuntu	Ubuntu 5.1.1	1:10
						to internet
						215 / 210 Kb/s			
						Germany	Linux	Linux		0:22
						Pakistan	windows		3:20

I wanna also test 2 of the machines with the 2nd system, but
Windows at base system need much longer then Linux based systems !
Even wine mtclient-64 on Ubuntu still is much better.

added:
This test was made onto the SAME Server (my one), INSIDE my 1 GB Home-Network, same config,
BUT from my different available PCs here.
Even some test from SAME PC booted with Ubuntu then with windows.

So the reason MUST be inside the client or the System at client itself !


Did you know ?

Do you know the reason ?

Ia there a solution ?

[bosapara]

As i discovered, login time mainly depends on 2 things:

1. How many registered nodes. 1000 pcs of nodes takes ~ 15 sec of loading, 10000 pcs ~ 1.5 mins.
2. Resurces (audio, models, pics), less MB - better login time.

Such mods as "moreblocks" or "xdecor" multiplies the number of registered blocks exponentially.

[Festus1965]

I was logging, and took the times (just today) of others only on my own server - so this differences are not included here,

mean here inside my playroom for the kids all PC had the same change as from server, and the GB-net ... and their was nobody else online (we had 24 hours cut of the fiber) and I did only one at same time ...
the only reason for this difference is to search at the clients ... (system, mtc) and as even slow PC with linux are faster than faster PC with windows ...
I have my fact ...

[Sokomine]

I'd exped M2 to be much faster than SSD and that one much faster than HDD, but perhaps that doesn't even come into play here much. It certainly doesn't explain the values you've measured.

Other circumstances may also be a factor: Has the client connected to that particular server before (very likely in this case)? Has the client connected to that particular server very recently so that the data is still cached somewhere (Linux loves doing that; very likely the case here as well, so it's ramdisk we're talking about)? How many other servers has the client connected to (that is, how much other data has accumulated in the cache directory?). It's very likely that those machines with more ram and Linux just kept the data automaticly in memory, making it irrelevant after first start weather they had a ssd or hd.

If the same thing happens at first start of MT after the computer has booted, then it's probably an indication of Linux being better able to deal with a file system with many tiny files. Still, a cd into the cache folder of MT might not be the best idea...not even under Linux.

I'm also intrested in what other aspects influence startup time as I often have to start the game anew when working with one of my mods that modifies the map in register_on_generated to some degree or another. Keeping the amount of mods to the bare minimum needed for the task helps a bit, but speed increases would certainly be welcome with a newer machine.

[Festus1965]

M.2, SSD and hdd
for this loading it doesn't seam to be important
but you feel the dif in using the PC for real - that makes happy, startup so fast ... loading apps

all PC had fresh restarted and then measured at first time of open mtc and connect. That one I thought about before also

.... so this case keeps mystery