Now that's very strange. Three quite different CPUs all getting 145 seconds? Seems odd to me. First though,
ThERat said:
By the way, do you take the timing from pressing enter until the first production window pops up?
I did it from pressing enter until all the production windows were gone and I could move a unit on the next turn.
That you didn't get twice the performance doesn't surprise me at all, but I would have expected at least some improvement with one or both of your CPUs.
Then again, the i7 870 does have 256 KB of L2 cache per core, half of what the Phenom II does, and 33% less L3 cache than the Phenom. Generally it's regarded as faster per-clock, but maybe the lesser cache is canceling that out. I could see it potentially not being as fast per-MHz as the Core 2 if L2 cache really is super-important for Civ3, but this three processors, all different, getting the exact same time is really weird. Looks like we need someone else with a similar model to one of ours. If I hadn't gotten a slightly better score when I overclocked, I'd find this really weird. Or maybe a different save, although a different save could be more or less sensitive to the amount of cache.
If no one else replies in a year, I might upgrade my CPU to the same line but with 50% more cache, and be able to test at the same clock speed and see if it makes a difference...
Brainiac18 said:
It's not what I meant. When running VM at max load (during a Civ3 play inside it's winXP), I see in win7 Task Manager-processes-CPU that overall load of this VM is ~60%. At the same time, inside VM's winXP, in it's Task Manager I see 100% XPU load. So why my win7 is not giving all CPU power I got to VM? That' my question. Because in that case I would see, for example, that winXP (inside VM) CPU load is not at it's maximum, or we would have better results.
Ahhh, okay. How many Processors did you give the virtual machine in VirtualBox's settings? In Virtual Box 3 that's under the System options, I don't know if they moved it in VirtualBox 4 or not. If you gave it one processor, it won't be able to use more than about half of your host system's power, which is why you see Windows 7 reporting about 60% usage. If you give it two processors instead (or three or four, although this would probably decrease performance more than help), it should be able to use 100% from the host's point of view as well, although if all you're running is Civ3, you'll probably see 50% usage in the guest and 50% in the host (approximately). I don't know how XP will behave if you suddenly give it two processors, either. I've only really messed around with giving Linux VMs more than one processor.
I mean, this could mean that Civ3 is about Mhz of CPU, cause you don't increase L2 cache, but the performance uped. On the other hand, my result was identical, so I hope we will have more people involved here
The MHz will definitely matter, but it's looking like L2 may indeed be a very important factor that can outweigh megahertz as well as L3 cache. Kris' Celeron test is a good example - when he increased the clock speed by 55%, the time it took only went down by 18%, rather than the 35% you'd expect if the megahertz were all that mattered. Perhaps at some point I'll see what happens if I underclock my CPU to 1.1 GHz (half as many MHz) - my guess is is would take noticeably longer, but probably not 290 seconds (twice as long).
And I'm not satisfied :/ Today I got AMD again, Phenom II X2 560, L2 cache is the same, but this CPU has shared 6MB L3 cache so I really hope there will be sense it that 
