Mod load time

You should have told me this before I merged my 40-some Culture Units into one set of files.:rolleyes:

I still think for the benefit of the modding, its much easier to have as many merges as is possible. The main reason to maintain modular form is if players might want to not play with a particular thing and have it easily removed. But then again, I don't think too many actually utilize this ability either. I wonder how many do.
 
I still think for the benefit of the modding, its much easier to have as many merges as is possible. The main reason to maintain modular form is if players might want to not play with a particular thing and have it easily removed. But then again, I don't think too many actually utilize this ability either. I wonder how many do.

We get complaints if we do something which relates to the extra religions and don't make it modular. Not sure which religions they are removing but perhaps the latest tech tree changes will draw them out since if there is no religion to found why is there a tech to found it. ;)
 
For me:
Jainism
Mormon
Ngaiism
Scientology
Sikh
Tengriism
Voodoo
Yoruba.

Which reminds me, they keep XML erroring on me for the female missionaries.
Yes, I can see the technologies to research them even when they are not activated.

Cheers
 
For me:
Jainism
Mormon
Ngaiism
Scientology
Sikh
Tengriism
Voodoo
Yoruba.

Which reminds me, they keep XML erroring on me for the female missionaries.
Yes, I can see the technologies to research them even when they are not activated.

Cheers

How can I fix stuff if people don't tell me it is broken. :p
 
I've just forgotten to tell you. Doesn't take long to click those errors away anyway.

Cheers
 
I have pushed the caching to the SVN.
For now it is deactivated by default as bugs are to be expected. You can activate it with the global define DO_CACHING in A_New_Dawn_GlobalDefines.xml.
First run will create the caches and they are then used afterwards unless one of the XML files has been changed. The loading time is reduced from 85s on the first run to 40s on the second run for me (fast test run, not repeated).

I have written some additional serialization methods but not all info classes for which it would be useful have them yet so further speed increases are possible.

The important thing now is for you to test it and especially check for missing stuff on buildings, units and the like when the game is loaded from the caches. The serialization functions have not been tested for quite some time and it is not unlikely that some modders added stuff to the XML but not to the serialization functions (although most seem to have updated them fine).
 
I have pushed the caching to the SVN.
For now it is deactivated by default as bugs are to be expected. You can activate it with the global define DO_CACHING in A_New_Dawn_GlobalDefines.xml.
First run will create the caches and they are then used afterwards unless one of the XML files has been changed. The loading time is reduced from 85s on the first run to 40s on the second run for me (fast test run, not repeated).

I have written some additional serialization methods but not all info classes for which it would be useful have them yet so further speed increases are possible.

The important thing now is for you to test it and especially check for missing stuff on buildings, units and the like when the game is loaded from the caches. The serialization functions have not been tested for quite some time and it is not unlikely that some modders added stuff to the XML but not to the serialization functions (although most seem to have updated them fine).

Cool. Thanks for picking up on this, where I was too lazy to do more than whinge about it ;)
 
Cool. Thanks for picking up on this, where I was too lazy to do more than whinge about it ;)

I don't think any of us are 'lazy'... quite the opposite. We just have our skill sets and priorities that vary is all. You guys are making this mod overcome ALL the problems that AND has been plagued by and we would have continued to have been plagued by to such an extent that you're paving the way for this mod to be not only one of the best mods on the site, but one of the best games ever made. We can all dream and some of us can begin to implement those dreams. But you guys make it possible for all those implementations to run correctly and smoothly and keep the game playable. I'm impressed with EVERYTHING you two do!

Well done AIAndy! Keep us posted as to whether you find and work out any bugs. My god, between you and Koshling, my imminent merge is looking more and more intimidating by the day! lol! I'll just have to be VERY careful I think.
 
Keep us posted as to whether you find and work out any bugs.
I have to rely on people that play a lot to find the bugs as I would not recognize missing stuff on buildings given how rarely I actually play C2C.
So if there are no reports in the next days I will just activate caching by default which then should hopefully get feedback soon.
 
Is it recommended to have this in the CivilizationIV.ini

; Disable caching of file system (may slow initialization)
DisableFileCaching = 1
; Disable caching of xml and file system (may slow initialization)
DisableCaching = 1

While enabling DO_CACHING in A_New_Dawn_GlobalDefines.xml ?
_________________

Disabling caching in the .ini speeds up gameplay generally doesn't it?
 
Is it recommended to have this in the CivilizationIV.ini

; Disable caching of file system (may slow initialization)
DisableFileCaching = 1
; Disable caching of xml and file system (may slow initialization)
DisableCaching = 1

While enabling DO_CACHING in A_New_Dawn_GlobalDefines.xml ?
_________________

Disabling caching in the .ini speeds up gameplay generally doesn't it?
No to both questions.
Especially file caching has a serious effect on the start time of C2C that is independent of the new caching I added.

I recommend having both those options at 0 (meaning that the caching the exe does is active as well) and then setting DO_CACHING to 1.
 
I recommend having both those options at 0 (meaning that the caching the exe does is active as well) and then setting DO_CACHING to 1.

But will it "hurt" if the first two are set to 1, and the third is set to 1 as well? Because I would like to leave the first two at 1 for other purposes. If so, then I will just change the first two, as it would be easier to leave them at 0 then to change the third to 0 each time the SVN is updated (when you actually change it in future SVNs).
 
@AIAndy:

I did a test of load times on my crappy rig with caching on, and the results were;

  • Pre-load: 119 seconds (not much we can do about that)
  • Init XML: 40 seconds (Yay!)
  • Other functions: 10 seconds

So 170 seconds total. Not bad, considering how big of a mod we have.
 
But will it "hurt" if the first two are set to 1, and the third is set to 1 as well? Because I would like to leave the first two at 1 for other purposes. If so, then I will just change the first two, as it would be easier to leave them at 0 then to change the third to 0 each time the SVN is updated (when you actually change it in future SVNs).
They are completely independent so any combination works.
Btw, if you make any changes in your local SVN copy and update then, it should merge in your changes and only rarely cause conflicts.

@AIAndy:

I did a test of load times on my crappy rig with caching on, and the results were;

  • Pre-load: 119 seconds (not much we can do about that)
  • Init XML: 40 seconds (Yay!)
  • Other functions: 10 seconds

So 170 seconds total. Not bad, considering how big of a mod we have.
Since I have made sure that DisableFileCaching is set to 0, I do not get a lot of Pre-load time.
 
They are completely independent so any combination works.
Btw, if you make any changes in your local SVN copy and update then, it should merge in your changes and only rarely cause conflicts.

Sounds good. BTW, I was speaking of my "working" copy of the SVN, not the one I update daily. I update delete and replace my working copy every week or so with the latest SVN, and that is the one I would change. I may forget changes that I made previously when I do so.
 
They are completely independent so any combination works.
Btw, if you make any changes in your local SVN copy and update then, it should merge in your changes and only rarely cause conflicts.


Since I have made sure that DisableFileCaching is set to 0, I do not get a lot of Pre-load time.

I have both caching options set to zero already, my computer is just old and crappy, so that probably contributes to the issue.
 
We'd spoken a bit earlier about MAFs taking place, particularly on weaker systems, if file caching is set to 1. Has anything you've done given us any cause to suspect this would no longer take place?
 
Back
Top Bottom