Turn Speed: A Game Breaker?

[darthx86]

Originally posted by battists
Good idea on the benchmark. Here's my take:

My reference specs:
CPU = 700 mhz Pentium
RAM = 256MB
Video card = ATI Rage Mobility
OS = Windows 2000 SP1

It took about 32 seconds for me.

Ok, I just got it installed and tested, and it took about 29-30 seconds.

As a result of my upgrade, my FSB went from 200MHz (DDR) to 266MHz (DDR). The video card was upgraded, as was its driver. But most importantly, the CPUs were upgraded from Durons to Athlon MPs (Athlon 4s).

Since you have a Pentium 3 700 (or is it a 700e?), your bus speed is 100MHz, yet had a better time than my 1GHz Duron also at 100(200MHz) FSB. Since I improved dramatically, my guess is that it is CPU cache related - your P3 and my Athlons both have 256K of L2, while my Durons had 128k cache.

Does anyone out there have a Duron or Celeron system with Windows 2000 that could give this a test? My curiosity has been piqued!

[Polonius]

I haven’t tried Darthx86's save game yet, but recently did some tests with one of my own. 2 different computers and before and after the patch.

So for those of you who are interested in figures, here are the results of putting the stop-watch on a test turn. The save used was an 8 civ game with all the map showing and close to 300 cities in total.

Machine 1:
(Pentium 733mhz. 128 megs RAM, TNT2 Ultra 32 meg video card)

Before the patch:
All movements on....1min 55 secs
All movements off.... 1min 25 secs

After the patch:
All movements on..............1min 41 secs
Animations off/ Show on.....1min 16 secs
All movement options off.....48 secs


So if you want a big improvement - put the patch on and turn all the movements off (the patch also allows more movements to be turned off than the unpatched game).


Machine 2:
(Athlon 1.2ghz. 256 megs RAM, Geforce 2 32 meg video card)

After the patch:
All movements on...............1min 13 secs
Animations off/Show on......50 secs
All movement options off.....36 secs

A further point of interest was that on this machine it made no difference to the speed if Nortons antivirus checker was on or off. It also made no different if another 128 megs of RAM was added to make it 384 megs, or (in the other direction) if 128 megs was removed to leave only 128 megs RAM.

[darthx86]

The performance seems to be affected mostly by the CPU itself, whether it's L1, L2, FPU, or flat out MHz remains to be seen.

(Hmm, maybe they can release SMP support in the next update?)

[Polonius]

Hi Darthx86, I tried your saved game and got 29/30 secs (On the Athlon 1.2ghz - Win98 old edition (not SE)).

Interestingly, your save had plenty of units but not much happening. Most of the troops were "asleep". The movement options were all turned on except for two - "Show friend moves" and "Show enemy moves". Turning those two on showed a few ships from other civs and increased the time to 45/46 seconds.

Turning all the movement options off made no real difference from 29/30 secs as there did not appear to have been much happening in those areas to turn off.

In contrast, the save that I'd used had other civ's troops moving through my territory, and a lot more happening generally. So turning the options on and off made a much more noticeable difference.

It looks as if the game has a huge amount of calculating to do no matter what, so the faster the CPU the better - probably. RAM would help up to a point and then make no further difference, it seems. Turning the movement off can help a lot - but it looks like the main thing we need to upgrade is our patience!

[darthx86]

Originally posted by Polonius
Hi Darthx86, I tried your saved game and got 29/30 secs (On the Athlon 1.2ghz - Win98 old edition (not SE)).

Interestingly, your save had plenty of units but not much happening. Most of the troops were "asleep". The movement options were all turned on except for two - "Show friend moves" and "Show enemy moves". Turning those two on showed a few ships from other civs and increased the time to 45/46 seconds.

Yeah, I just grabbed the latest one I had available. It doesn't matter too much what options are enabled/disabled, so long as they're standardized for the test.

It looks as if the game has a huge amount of calculating to do no matter what, so the faster the CPU the better - probably. RAM would help up to a point and then make no further difference, it seems. Turning the movement off can help a lot - but it looks like the main thing we need to upgrade is our patience!

Patience is always good, but the RAM doesn't seem to be much of a factor at all for increasing speed... I'm still curious as to whether it's MHz, FPU, or cache that has the most impact. (I'm guessing it's cache, since my 1GHz Duron was about 10-11 seconds slower.)

My request goes out again - anyone out there with a celeron or duron system (preferably 800MHz+) that can test this for us?