Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Random_Dragon]

So why all the extra bits then?

My assumption was for the sake of idiotproofing it, since the last version's thread was several pages of Toady and others trying to figure out why it abruptly borked.

[Greenbane]

Is there any advantage to 64bit support beyond being able to allocate greater amounts of RAM?

I suppose multi-core CPU support is still a ways off.

[Gwolfski]

Is there any advantage to 64bit support beyond being able to allocate greater amounts of RAM?

I suppose multi-core CPU support is still a ways off.

Bigger, older worlds . Marginal speed upgrade.

[Quietust]

I seriously doubt it, since i use vs2015 currently and i don't need all that crap with any of my executables.

When you install Visual Studio 2015, it installs the runtime libraries for you - if you were to try and run those programs on a system where VC2015 wasn't installed, you would need those DLLs.

Also , statically linking can often give a nice boost to perf as it allows more optimizations.

It is also an extremely large obstacle to getting DFHack to work.

[ORCACommander]

anyone else long for the good old days where what ever dll the ends user needed was shipped with the final products bin folder instead of having to install dozens of run times?

[Cexp]

I guess this is "expected" or at least known, but different versions (Windows/Linux) generate different worlds with the same seeds (I think the land might actually match, but the histories diverge).

I noticed when testing for speed. Disregarding the differences (since the populations and event numbers are similar), the Windows versions seems about 5% faster (for worldgen).

[jecowa]

Even on the same system I didn't think the same history would generate twice on two worlds with the same world gen settings and seeds.

[Toady One]

Is that still happening regularly?  I fixed two problems with it for this time and had a world regenerating properly, but it's possible these issues still exist into 43.05.

[jecowa]

Sorry, the last time I messed with it was 0.42.06.

[They Got Leader]

If we were using the 64 bit beta release of 43.04, can we migrate saves over with mild to no issues?

I haven't noticed any problems yet at least, so I believe so.

I just tried it myself, and it seems to be working perfectly fine. My assumption is that because none of the RAWS were edited, the game world does not have to be remade.

[ChaotikDwarf]

I seriously doubt it, since i use vs2015 currently and i don't need all that crap with any of my executables.

When you install Visual Studio 2015, it installs the runtime libraries for you - if you were to try and run those programs on a system where VC2015 wasn't installed, you would need those DLLs.

Also , statically linking can often give a nice boost to perf as it allows more optimizations.

It is also an extremely large obstacle to getting DFHack to work.

Well, i have to admit that since i switched to vs2015, i didn't deploy anything  to the outer world.
I'll give it a try.

[Shonai_Dweller]

If we were using the 64 bit beta release of 43.04, can we migrate saves over with mild to no issues?

I haven't noticed any problems yet at least, so I believe so.

I just tried it myself, and it seems to be working perfectly fine. My assumption is that because none of the RAWS were edited, the game world does not have to be remade.

Except those door opening Gorlaks. I guess that's a raw edit.

[Chase]

Animal people aren't scattered through out character race selection any longer.

[Quietust]

Is that still happening regularly?  I fixed two problems with it for this time and had a world regenerating properly, but it's possible these issues still exist into 43.05.

World generation appears to be consistent within a single version - both 0.43.05-win32 and 0.43.05-win64 were able to generate the same world 3 times in a row, but they could not generate the same worlds as each other (i.e. 64-bit history was consistently different from 32-bit history), which suggests that there might be other problems lurking around somewhere (which seems a bit odd, given that you have to specifically ask for 64-bit variables on Windows).

[Toady One]

Animal people aren't scattered through out character race selection any longer.

That might just be that world -- that's supposed to happen on worlds where certain animal people get integrated into a society.

World generation appears to be consistent within a single version - both 0.43.05-win32 and 0.43.05-win64 were able to generate the same world 3 times in a row, but they could not generate the same worlds as each other (i.e. 64-bit history was consistently different from 32-bit history), which suggests that there might be other problems lurking around somewhere (which seems a bit odd, given that you have to specifically ask for 64-bit variables on Windows).

Yeah, since I was incorrectly writing that off to different compiler optimizations before, maybe I should investigate.  There aren't many places where explicit 64 bit variables are used, especially in world gen, but there could be something.  Vector sizes return as 64 bit (at least, according to the thousands of warnings I got, size_t wasn't a 32 anymore), so maybe it's a conversion in some calculation with those, since those are very common.