Playing big maps, computer performance, how?

[Lemon Merchant]

As long as the code isn't written to use SMP (Symmetric MultiProcessing) it doesn't matter how many cores, or even how many CPUs, you have: the game won't run any faster. You may see the game run more smoothly because hardware interrupts can be handled by an idle core, but that's about as good as it gets.

Yup. That's what I was getting at, only not as concisely as you put it.

 

[Zero-Sum]

What I don't understand is if my graphics are the bottle neck on the PC, then why does the game run perfect in the first half? The graphics are always the same.

If the game can't run correctly on a average PC, then its not ready to ship out.

 

[binhthuy71]

What I don't understand is if my graphics are the bottle neck on the PC, then why does the game run perfect in the first half? The graphics are always the same.

If the game can't run correctly on a average PC, then its not ready to ship out.

Graphics aren't the bottleneck: Civ ran the same on my old machine with a 256MB card as it did with a 1GB card. The problems with the second half are largely caused by the fact that there's just a lot more stuff (Buildings, units, trade, diplo, tile improvements, plus a war or two) in the second half together with the way the game is coded and CPU utilization goes through the ceiling. Add in the fact that Civ seems to grab more and more memory over time - to the extent of requiring that you occasionally reboot - and the second half of a large game slows on the best of machines. You can either play smaller maps or cultivate patience.

 

[Excelsi]

That's not true at all. I started out playing Civ 4 on a Radeon 9550 with 256 meg VRAM. I couldn't play anything beyond Standard maps without the game crashing on me in the late game, even with my settings on Low. It was only after upgrading to a X800 that I was able to play Huge maps, and even then there was quite a bit of lag in the late game. All that is gone with my 8800GT though. Civ 4 doesn't need a high end graphics card, but it does require one that's decent.

I can run Civ4 on a 32mb VRAM laptop, so theres something wrong with this. 256mb is four times the requirement.

If you experience slow loading, slow turn ending or "stuttering" of any kind, you could use more RAM. The major giveaway here is your hard drive thrashing during the lag. An inadequate graphics card result in very obvious symptoms, all it will affect is the frame rate, or in some cases you'll get graphical errors. A vid card will never cause slow turns, if more people understood these things they'd stop wasting money on a Hyper nForce cyberFX76000 where a stick of RAM would do.

CPU is harder to undertand, but as long as it's relatively modern you shouldn't have to worry about it for a few years. Either way, if it develops at a particular point in the game, it isn't graphics, it's memory. If you can run 60fps with a screen full of forest at the early game, you can handle late game exactly the same way since graphics burden doesn't increase with time.

 

[Quintillus]

Don't worry Dave, I'm just funnin' ya. I have no problem with Nvidia, I just use ATI because my motherboard is optimized for those cards. "Dragon" platform and all of that marketing nonsense. My old Nvidia card was an excellent little piece of hardware. I'd still use it if the machine hadn't caught fire.

That's just a slight problem there.

Umm, I don't think CivIV supports multicore CPUs. So unless you have a habit of alt+tabing out of your game a lot, you don't need dual core.

Nope, Civ4 does not support multiple cores.

Neither memory (as long as there is enough to prevent swapping and to have disk cache), nor better graphic card, nor more cores (the game is single threaded for most of the part) can help that. Overclocked single core CPU is probably the best thing to run the game unfortunately.

Well, single-core CPUs are not the fastest per core anymore, so you don't actually want a single-core CPU for the fastest performance. The fastest per-core is probably Intel's top-of-the-line Core i7 975 quad-core, although more economically for Civ would be something like the Core 2 Duo E8500, which is still much faster than any single-core you can buy once it is overclocked.

Well, I'm not the biggest code genius, and I certainly don't understand all of the nuts and bolts. I can only speak to what I know from experience, and some research. Civ is definitely a single core app, and having more cores doesn't really make it faster from that perspective. Vista seems to allocate one core specifically for Civ, and the rest of the cores do the housekeeping chores. This does produce a small performance improvement, notably a reduction in turn lag, and leaderhead load. The machine is also able to handle larger map sizes without bogging down.

True - there's a slight improvement from having a dual-core over a single-core (assuming you could get a single-core as fast as the fastest dual-core). It depends on how many background tasks you have, but generally it's less than 10%.

It's fairly easy to test this sort of thing. My former room mate tested his old single core Athlon against his proud new Phenom purchase last year (Using XP). He set the clocks the same and with the same amount of RAM, and he moved the graphics card to and fro between them. (He's an engineer, so this sort of thing is a big science project for him. Knowing him, he probably still has all of the data hidden away somewhere. ) Then he played some saves. (That's where I learned the test technique.) He did measurements and everything, and there was a definite performance increase in having more cores, though it was admittedly small. Then he played a full game on each machine, using a large map. He thought that a huge map would be too resource consuming and skew the tests in favor of the quad core. Again, a small performance increase in the form of reduced turn lag. The "camera" feature of Civ was a little smoother to both of us, but that is a subjective measurement he discounted.

So, ok, big deal, right? The bottom line is that having another, or more, cores takes the load off of a single core machine running Civ. My Phenom is clocked at almost 4 GHz (3.95), so all of this is moot for me, but if I clock my machine at the same speed as the highest you can over clock your mono core, my machine will outperform yours. I don't gamble, but if I did, I would put money on it.

The other variable in that test your roommate did is the architectural differences between the Athlon and the Phenom. The Phenom is probably more efficient per clock cycle, and that could account for a good portion of that difference. The multiple cores probably did cause some of the difference, but I wouldn't attribute all of the difference to that.

A better test would be to disable all but one core in the BIOS, if possible, and to test performance with one core of the Phenom versus all four (or two of the four). I did a similar test on my T7500, testing dual-core 2.2 GHz versus single-core 2.4 GHz on a Civ3 saved game. The single-core variant with 9% extra clock cycles saved 2.6% of the AI turn time - the second core was helping some with background tasks, but not enough to offset a minor increase in clock speed (I do still have the test results saved ).

For the architectural differences reason, give me a Core 2 Duo E8400 to E8600 with one core disabled, and I'd accept the gamble. If you insisted on a true single core, I might accept for a low amount of money with a Celeron 900. That'd be an interesting test - I don't know if that relatively new single-core Celeron would be more efficient per clock than an Phenom or not.

I agree with Lemon Merchant

Also, I read that DirectX 11 has more support for multiple cores.

If Civ uses DirectX, and If that means it automatically uses the latest version of DirectX and doesn't bind to an older version, then you might be seeing benefit from that with a multiple core CPU. I made some assumptions in that statement though...

You can't just consider the game code itself when deciding whether you are getting benefit from multiple cores, you also have to consider all the platforms the game was built on.

If you haven't already, you may want to consider reformatting the hard drive and reinstalling the O/S from scratch. Sometimes starting from scratch like this can give you a nice performance boost.

DirectX 11 won't help games that don't implement it - nor will any version of DirectX. So there's no benefit to be gained from installing DX11 if you have Vista.

@Lemon Merchant
Ok. Win7 might be improved and the kernel could take some benefit of multiple cores. The kernel/API itself might be improved [speed-up], the video/DirectX handling can be improved, it can take benefit of 64bit instructions and the like but the game itself will stay the same.

If you need benchmark, you shall measure the time spent in the game process (I think 'process explorer' could do that; CPU time on task manager).

Running smoothly on win7 should be some very good news, especially to the sale dept of microsoft. I, myself, don't give a damn about it.

The OS could help if it were more efficient in certain system calls that a program used regularly, but the differences are usually so small as to be negligible. I wouldn't spend money on an operating system in hopes of improving an application's performance, with very few exceptions.

Also, Adria is right, turning down graphics options can help with memory in Civ4. They'll definitely help with your onboard video. I have to agree with those who have said that's likely your problem - normally I'd put CPU and RAM as the culprits for Civ4, but you definitely are not lacking in those.

It's disappointing to know this issue persists in BTS 3.19, though. Good that it's not as bad as in vanilla, I suppose, but it would be really nice if it were gone altogether.

 

[Zylornck Davis]

The other variable in that test your roommate did is the architectural differences between the Athlon and the Phenom. The Phenom is probably more efficient per clock cycle, and that could account for a good portion of that difference. The multiple cores probably did cause some of the difference, but I wouldn't attribute all of the difference to that.

A better test would be to disable all but one core in the BIOS, if possible, and to test performance with one core of the Phenom versus all four (or two of the four). I did a similar test on my T7500, testing dual-core 2.2 GHz versus single-core 2.4 GHz on a Civ3 saved game. The single-core variant with 9% extra clock cycles saved 2.6% of the AI turn time - the second core was helping some with background tasks, but not enough to offset a minor increase in clock speed (I do still have the test results saved ).

For the architectural differences reason, give me a Core 2 Duo E8400 to E8600 with one core disabled, and I'd accept the gamble. If you insisted on a true single core, I might accept for a low amount of money with a Celeron 900. That'd be an interesting test - I don't know if that relatively new single-core Celeron would be more efficient per clock than an Phenom or not.

Lemon forgot to point out that the Phenom we compared was an older Phenom series. The 9850 to be exact, while her Phenom is a Phenom II series - big, big difference there. My Phenom is a little closer in architecture to the Athlon we compared it to so the test is a little more valid, but you are right. I gotta say your point of the Celeron to Phenom test might be interesting, bt I can't disable cores with my cheap bios. Time for a new machine anyway. Maybe I can sweet talk Lemon into building one for me. Hers freakin screams.

 

[Zero-Sum]

Well my graphic settings, graphic details? Were on high, I don't know what is the default ,but knowing me I probabbly set it to high when I first installed the game. I dropped it down a bit to the medium setting, the game runs more smooth now during game play. Although after midway through the game it will still take a few seconds between turns, but at least now it runs smoothly, and not laggy and choppy like before.

 

[Merkill]

My tip is alt tab to windows and then back late game when the game starts to lagg , this usually makes the screen go black for a lil bit but then the game runs fine again so no need to reload or restart.

There seems to be a cache in the game that gets full after a while and alt tab to windows and back clears it , atleast thats my guess of what is happening.

I am running an intel E8400 8gig ram ati 4870 512mb 2x640gig caviar black vista ultimate 64bit vlited.
so the late game lagg is not from a lack of memory

 

[warpstorm]

I am running an intel E8400 8gig ram ati 4870 512mb 2x640gig caviar black vista ultimate 64bit vlited.
so the late game lagg is not from a lack of memory

It very well could be lack of addressable memory. Since Civ 4 was compiled for a 32 bit operating system, it cannot use all of your memory. Getting much over 1.5 gigs out of one process is not gonna happen.

 

[Merkill]

warpstorm yupp thats true enough forgot all about that.

But then civ never use that much ram so i still think there is some cache or limitation put in the game.
This Cache seems to fill up faster now that i use bat2.0 mod i just have to do it maybe one more time in total for a10h game day.


but the alt tab to win and back in late game works fine for me and 2 friends that play the game so im a happy gamer .

 

[pesgores]

My computer is NINE years old, I've never upgraded it (only installed a DVD player) and I have the same problems as you. With a 9 year old computer. Vs a 1 year old. I only have 512 Mb of RAM, 80 Gb HD and a video card so good that doesn't let me play Europa Universalis III. Fortunately I have this one, but it is constantly occupied by my brother.

 

[StargazingDog]

Save your game and reboot once in awhile. There's some sort of memory/CPU leak that starts showing up at around the Renaissance era. The game will start to lag and become sluggish. If you reboot the game though, it will clear up and function much better for awhile. I usually need to reboot after 2-3 hours of playing in the later game because of this problem.

I haven't come across this. Maybe it is related strictly to Beyond the Sword, or a mod. I play Vanilla and it runs the same whether I have been playing for hours or just booted Windows and loaded my save game. I haven't played anything bigger than "standard" size yet. The game does slow down in the later stages, but the end of turn "spinning wheel" still only lasts for a few seconds, even in a war. The game clearly runs slower when moving around the map, but this seems to just be the coding of the game, since the performance doesn't improve at low graphic settings.

I am playing on a circa 2004 system comprised of a 3.2Ghz P4 and Radeon X850, everything set to high, 4x anti-aliasing, 1600x1200.

 

[the343danny]

Just got a new system with 2.66Ghz i5 and Radeon HD 5770. I heard in this thread that graphics doesnt matter too much, so how much would my Radeon make a difference compared to my old 8600GT? Also, since it doesnt use dual or higher cores, does that mean that having 4 physical cores does no good for this game (my old had 1.86 dual). I can run everything on high atm, but is that mainly because I got 4 gigs of ram?

 

[StargazingDog]

My X850 XT is weaker than an 8600GT, and as I said, I run max settings (high, detailed textures, anti-aliasing) at 1600x1200 and its smooth.

For another reference, a 2.0Ghz Core2Duo MacBook running the older Intel GMA950 video chipset chugs on high settings while an equivalent 2.0Ghz Core2Duo MacBook running the newer Nvidia 9400M video chipset plays it completely smooth on high settings (both running 1280x800).

Yes, your multiple cores gives you no benefit for this game, other than allowing you to run other programs in the background with little performance penalty.

Also, the game runs perfectly fine with a standard size map and 1GB of memory.

 

[WuphonsReach]

For huge maps, Civ4 is going to get up around 1-1.5GB of RAM used, so I'd put 2GB of installed RAM as the minimum. I don't know about video RAM, I've always used 256MB or better cards with Civ4.

As others have said, restarting Civ (or going back to the main menu and then loading a save) seem to speed it up again. I generally find that after 2-3 hours, past 1500 AD, things start to get a bit pokey where it takes half a second to select a unit in a stack. Restarting fixes that for another 2-3 hours and is a good reason to go pet the cat.

(My old PC was an Athlon64 X2 4600+ w/ 2GB RAM, now I'm using a Phenom X4 2.5GHz w/ 4GB RAM. Still using my GeForce 8800GT 512MB PCIe video card.)

For older games that aren't multi-core aware, you'll want the fastest per-core speed that you can get. Which would probably be a dual-core in the 2.8-3.2GHz range right now. Quad-cores offer more overall performance, but generally lower per-core speeds.

The biggest advantage of multi-core for gaming is that it lets all of the other background tasks (including the sound card and video card drivers) do their work on the 2nd core, which lets Civ4 hog one core all to itself. Depending on what all you have running in the background, that could be a noticeable improvement.