Hi,

I will post some basic rules in the next days, but I dont want to decide everything on my own.

This thread is suppossed to figure out which Rules and
especially which Settings would be the best.

So you can help me out if you will.

Alright

Scenario Runs
after some testing and reading the manual I think chosing this Settings might be the best to begin with.

Chosing a Player:

Choose any faction you want to and set the other remaing players on Knight(easy)

Game Options

I would recommend these options

In Game Settings (Alt+x):

I also would recommend to turn off everything in the in game settings, but thats your choice as well. Only if you feel confident enough. The less you watch the more time you save.

Winning Conditions:
To win the game you have to conquer at least half the cities on the Map.

Hi,
I have 2 questions regarding rules:
1.) In this thread you show some recommended options and in the rules for the different campaigns there it says "Options = Beginner" -> the only difference I notice being that "View Enemies On/Off" (On for Beginner). I haven't tried it but I guess that is if you hover over an enemy army and rightclick it shows you the army composition. Can you harmonize the rules? Cause I think it is confusing (even if it doesn't change much). I personally would suggest taking beginner-options, since when you record a run you can see that beginner-options are active (blue highlight).

2.) In the rules it states that the timer starts when you press "begin" at the setup screen. I'd suggest moving the timer start to the start of the first turn. The benefit being that loading times are not part of the timer.
Then it is open to debate if the time starts immediately when you see the screen of the first turn or when you press a key to enter the turn. Considering the second option it could be interesting to study the map - especially of random maps - of the positions of starting cities and neutral cities, before the timer starts.

Bringing in some thoughts, probably will be trying some runs in near future.

Have a nice day, happycamper

Hi Hapycamper,

thanks for the suggestions. Yes the rules can of course be changed until it is fair enough. I was not sure about them because there was only me at the beginning running this game.

1) Yes I can change this rule to "Options select Beginners". I am overall not sure about the options maybe we can figure something out what suits running the game best.

2) Not sure about this: The loading times should be the same on every modern computer I assume so I can not imagine that there will be some heavy difference between the runs. Yes I was thinking about this too, but if you dont have time to study the pressure is on you right from the start so you have to make your decisions as fast as possible. I would suggest to stick to the current rule for that.

I will update the rules for Beginners.

Really looking forward for some runs from you, I was just setting up some benchmarks for this game I have no clue if they are really fast or not. (for example JoMama blasted my Isle of Sorcery Run)

Cheers

Hi!
That was a fast response.
I figured the rules are not yet "set in stone", since the game is pretty young on speedrun.com.
ad 1.) The 'difficulty' options could bring some more differentiation in the future, But for now it's fine as it is I guess.
ad 2.) Since most people will use DosBox to run it, it shouldn't be too different with loading times. Unless you change the standard options of DosBox like cycles, then it can behave quite different. But I guess we'll see how it will go.

I'm looking forward to becoming a multiple-game speedrunner
This winter could be my speedrunning-season, we'll see how it goes

See ya

No we have plenty of room to improve the rules
I also did not differ between the different factions and their starting position on the map like in Warlords 1 what also makes a difference how fast you can win. There is much stuff yet to be discussed.

Game difficulty yes it also an additional option in the future but this aligns with the factions mentioned above because unless we are just quite a few runners I did not want to blow up the leaderboard to much in the beginning.
But who knows how things will progess.

What we also maybe consider is to disallow game reload to manipulate the RNG (not sure if this works in W2 but we banned it for W1 so you cant grind a ruin until you find dragons and stuff).

Cya

Hello!

TLDR: Loading times are likely inconsistent on different machines.

After some tests I return to discuss the loading times.
I watched the runs of MilkToast & JoMama and took down your loading times after you press "Begin". Additionally I measured the time it takes for me to load into these maps. I found out that there are huge differences in loading times between our different systems, see the table here:

Campaign--------|--MilkToast---|---happycamper_---|--JoMama
Hadesha----------|-------25s-------|------------11s------------|------------
Erythea------------|-------20s-------|-------------9s-------------|--------24s
Isle of Sorcery--|-------6.5s------|--------------3.5s---------|---------8s

Of course I don't know on what settings you guys ran the game. If you just ran it on the default settings that are given in the dosbox-config from the download, then I question if these settings are good for running Warlords 2...

On the other side I checked if the loading times are the same if I start a campaign multiple times -> there it seems to be stable.

I come back to my suggestion to don't include the loading times into the timer. The other option would be to find settings that make the loading times equally long. I think it could be hard to find the 'ideal' settings.

Cheers, happycamper

PS: about the dosbox-config: I had to adapt 2 settings because my OS is linux and capturing wine-windows via OBS can be a pain. These settings are "output=opengl" (no ddraw in linux) and "fullscreen=false" (I want to see OBS during the run).

Hi happycamper,

yeah I looks like we should change the rue for when the timer starts.
If I recall correctly I did not change anything in my DosBox (OS W10).

I will adjust the rules and retime the already submitted runs soon.

For the new rule starting the timer I think we should go with:
- As soon as player recruits the starting hero (press ok)

Thanks for taking the time and checking it.

Hello all,

I just saw this thread with the discussion about changing the timing rules. I totally agree with that change given the inconsistent loading times across runners. I just submitted a couple runs using the old timing method - MilkToast, would it be easiest if you reject those runs so I can resubmit with the new timing by subtracting the loading time? Or do you feel it is more appropriate for me to create new runs using the new timing scheme?

Regarding the initial game settings, I agree with using the "Beginner" settings moving forward. So far the difference between this and the default settings hasn't played any role, but it is more transparent. This might also allow a natural extension to other categories, just by choosing the "Intermediate" or "Advanced" options.

Hi,

I will verify the runs today and substract the time during this process, so there is no need to resubmit them.
I already coreccted older runs and started the timer by hitting the ok button accepting your first hero.

Cool thanks guy for your suggestions.

Currently thinking about other ways to run the game:

-Tutorial map - annoying because of the instruction popups, but you can finish the game by capturing 4 out of 6 cities with a congratulations screen - and it's really short. why not?

-completionist style: all scenarios back to back? including / not including the tutorial? could include some timing splits there which would be something new.

-other difficulties / settings - thinking about meaningful settings

-terminate all opponents (with or without having more than 50% of cities?)

-no ruins / temples etc. Finish without inspecting with hero

any other ideas? Some of them seem interesting at least in theory I think

Hi,

sure we can come up with new categories if you like to run them we can create them.

Tutorial: Nobody will miss it, but it can be added if desired.
Completionist: This would be the Full Game run then
Other difficulty settings: Would blow up the LB to much, but we can talk about some adjustments when selecting different difficulties
Terminate all opponents: Its the same as 100% category for Warlords, so I think we can also add this
No ruins/ temples: Should be added later, if there is still need for more categories.

So there seems to be some need for new categories so I wll create a few

Cheers

I did more testing regarding CPU-capping: more concretely if AI-turn-time and the time it takes for units to move is affected by the performance of the machine. I tested on a netbook and my gaming laptop and compared them to each other.
So what I noticed is that:
1.) Unit movement time seems independent of CPU as they were the same on both machines
2.) AI-turn-time seems to be dependent on CPU-performance though... a short explaination of what I did for testing may help for understanding:
So I started a game with 1 player and 7 AI and clicked away observing them. I took the time it takes after I press "end turn" and when my round starts again when the banner apears again. In the first turn the AI usually doesn't move, that is why it is good for comparison. On the gaming laptop it takes ~1 second until the next turn and on the netbook it took me ~ 6 seconds to get to turn 2.

While this an "extreme case", it might also be relevant for our setups, since the runs are getting faster and we're closer to a point where we're just seconds or soon maybe even milliseconds from each others runs.
Therefore I propose that we change the our dosbox settings to a fixed cycle number, so that the times we have are truely compareable. I'd run some tests on my machines to check if I can find consistent cycle settings.

sometimes they move on turn one. did you check back after the fact if the AI actually didnt move? in some cases they go for runes or even if they just get a hero to join them that affects time I found. I can run turn one in the same scenario on the same machine 10 times and 9 out of 10 times it takes 1 second, one time it takes like 5 seconds.

But I agree, if a fixed cycle speed makes things fair, that's good. i dont really have a too strong machine myself so thats definitely a factor.

And I guess this needs a lot of testing too.

Yeah, I checked, they didn't move on turn 1 or turn 2.
My experience is that they never move in turn 1, sometimes in turn 2.

For fixed cycles in dosbox you can read:
https:/​/​www.​dosbox.​com/​wiki/​Dosbox.​conf#cycles_.​3D_fixed_nnnn_.​7C_max_.​5Bdefault.​25.​5D_.​5Blimit_cycle_limit.​5D_.​7C_auto_.​5Brealmode_default.​5D_.​5Bprotected_mode_default.​25.​5D_.​5Blimit_.​22cycle_limit.​22.​5D
I'm looking into it right now. Maybe I can get an idea of the range of cycle we can consider.

First look into times@different cycle numbers:
|..Cycles..|.loading in time.|.turn 1 time.|
|.------------.|.------------------------.|.-----------------.|
|.50000...|...........15s..........|.....1.4s.......|
|.100000.|...........9s............|.....0.8s.......|
|...max....|..........9.5s..........|.....0.8s.......|
|.------------.|.------------------------.|.-----------------.|

Seems my laptop is capable to get to ~100k cycles. I don't know why I had less time for loading in at 100k fixed than for the max setting. I had the same applications running and no extra load to my knowledge. I think we have to gather some more data from your hardware to figure out what a good cycle number is.

on another note:
It seems to me that my netbook can offer a maximum of ~20k cycles as a test also confirmed. Since it takes a long time for loading in ~40-50s on the netbook I consider this not suitable for speedrunning and will not consider it anymore for tests. I say this because if we would consider all the very old hardware we'd have to go down to like 3k cycles but loading times would be awful - probably several minutes - and nobody would want to run the game anymore...

Thanks for the information.
On fixed it says:

How high you can go depends on the power of your host CPU and on the selected core (above). If the value is too high for your CPU the emulation will slow down and the sound starts to skip.

This could potentially be a problem and needs to be verified on a range of multiple machines I guess?

We recently made some rule-adjustments regarding the dos-emulation-setting about cycle times: From now on runs should be submitted with a setting of 50.000 (50k) cycles fixed to ensure that the runs can be properly compared. The previous setting was max which just lets you run the game as fast as possible on your hardware but the competition between runs would be flawed then (as the fastest PC has an unfair advantage).

The old runs will remain on the page with the field "cycles" set to "Max". For new runs we kindly suggest to run it with 50k cycles and set the variable cycles to 50.000 -> you can also read the section in the rules for the setup:

Quote


To change the Cycles in DosBox 0.74 you have to do the following:
After installing the latest (Windows) version of DOSBox (version 0.74-3), there is a batch file called 'DOSBox 0.74-3 Options.bat'. Running this opens the configuration file 'dosbox-0.74-3.conf'. In this file there is a setting called 'cycles' that should be set to 'cycles=50000'.