Early Exclusive: Civilization VI to release February 9th for Linux with a discount, NVIDIA only for now

Quoting: rkfg... I understand if there are issues with deterministic RTS engines, math could be slightly different on each platform and in the end the butterfly effect causes heavy desyncs. But in a turn-based strategy?.. Weird. ...

It is the same problem, it just (usually) has to do with the randomizer instead of the trigonometry. Wine has also struggled with getting this right in the past. As an example, rand() from MSVC is particularly crappy while the glibc version has a lot more randomness. Wine gets around this by implementing the crappy MSVC algorithm when an app tries to use MSVC's rand(), but this is not the only way to get random numbers. So, unless you capture all the ways that an application does its math differently on different platforms then you're going to have a problem.

No AMD-Support ... that makes me sad ... :(

QuoteRequired specifications
[...]
CPU Speed: 2.93 GHz

I like the precision of that requirement. ;)

Quoting: compholioIt is the same problem, it just (usually) has to do with the randomizer instead of the trigonometry.

Wow, I've never heard of that. I presumed, in TBS the host has authority so it does all the random stuff and then just sends the numbers back to the clients. Deterministic engines are made to fight excessive bandwidth and accompanying lags if you have hundreds of units that are constantly moving. Not the case for TBS I guess... Still, an interesting reason for incompatibility.

Quoting: rkfgUhh, what's up with that non-fully-crossplatform multiplayer? I understand if there are issues with deterministic RTS engines, math could be slightly different on each platform and in the end the butterfly effect causes heavy desyncs. But in a turn-based strategy?.. Weird. The Steam page doesn't have a "Crossplatfom multiplayer" badge, however, in the system requirements there's a footnote "Additional Notes: NOTICE: It is possible for Mac and PC to become out of sync during updates or patches. Within this short time period, Mac users will only be able to play other Mac users." So, the Mac users can play with the Windows users and Linux users can play only with Mac users. What would happen then if a Mac user hosts a game and then both Linux and Windows users connect? Something's wrong here.

Apologies, our wording for the Blog post makes it sound like it will only be compatible with Mac. We fully intend to have cross-platform multiplayer between all platforms, but that didnt make it for launch.

Quoting: rkfg

Quoting: GuestAlthough I do find the reviews still very mixed.

It's the recent ones which are mixed, overall reviews are fine. After DXMD I started to actually read the negative reviews, because it turns out often people rage over quite silly subjects like DLCs or preorders. The game itself could be fine.

This seems to be a total disease of most reviewing systems as of late: People abusing reviews to make a political statement against a studio, instead of judging the GAME. As much as we might hate some DRM systems or DLC practices, but criticism aimed at that has nothing to do with the quality of the game. By all means, people can mention it in the text, but they really should stay away from downgrading a game's score because of it. It's unfair to people who want to get an impression about a game's quality.

I hear the AI sucks. I guess I will find out. It always bothered me that the higher difficulty levels in Civ V gave bonuses because it just meant that I couldn't win earlier.

Quoting: BreezeI hear the AI sucks. I guess I will find out. It always bothered me that the higher difficulty levels in Civ V gave bonuses because it just meant that I couldn't win earlier.

Yeah, the devs said they'd make the AI less stupid, but from what I've heard, it's anything but. The diplomacy engine still seems to make the same completely dumb "We have been friends for 1000 years? No, we're at WAR now!!!" decisions it made back in Civ I.

I hope that the game works well with AMD Phenom II processor.

Quoting: rkfg...
Wow, I've never heard of that. I presumed, in TBS the host has authority so it does all the random stuff and then just sends the numbers back to the clients. Deterministic engines are made to fight excessive bandwidth and accompanying lags if you have hundreds of units that are constantly moving. Not the case for TBS I guess... Still, an interesting reason for incompatibility.

The standard rand() implementation on any system is not actually random, it is very predictable. This is used to great advantage in many games, since by properly seeding the randomizer you can get consistent behavior without transmitting the full state between systems.

So, lets say you have a treasure chest and you want to make sure it always has the same contents no matter who opens it and without transmitting the contents to everyone. All you have to do is seed the randomizer with an ID for the treasure chest, if you do that then you just need to use the output of the randomizer to determine the contents. This means you don't have to transmit "plate mail (with all associated properties), sword (same deal), amulet (having fun yet?)", you just send "seed X" and all clients will run through the same randomization algorithm (calling srand() and then rand() an appropriate number of times).

If, however, your randomizer works differently on different platforms then you run into trouble. The problem you run into is that periodically you want to verify that a client didn't cheat, so every once in a while you send a hash of your state data - if any client mismatches then somebody cheated (or there's a platform difference). In my experience with Wine and RTS games this actually has to do with bugs in MSVC's trigonometry* instead of random number generation, but there are plenty of ways to get a state mismatch.

* This is frequently an interaction between MSVC's calculation technique and a processor bug.


Last edited by compholio on 6 February 2017 at 7:19 pm UTC