Civ 5 interview. First 'native DX11 game'

[RussianSensation]

The performance difference between i5 750 and Q6600 at stock speeds, and Core i7 frequency scaling are unbelievable

 

[BFG10K]

The performance difference between i5 750 and Q6600 at stock speeds, and Core i7 frequency scaling are unbelievable

Not as unbelievable as the 2560x1600 scores, where the GTX480 is over ten times faster than the GT240.

 

[tviceman]

I honestly think that with a full suite of DX11 benchmarks done with up-to-date drivers, the gtx480 would prove to be considerably more powerful than both an hd5870 and what (many) people give it credit for.

 

[bryanW1995]

hopefully they'll do an update with barts and cayman.

 

[Sylvanas]

Tessellation = massive performance hit for previous gen architectures.

They set Tessellation to low, not that the performance hit was low.

By using low Tessellation, the GeForce GTX 285 rendered an extra 93%, GeForce GTX 260 received a 100% performance bump. The Radeon HD 4850 was also 82% faster and the Radeon HD 4890 achieved 58% frames per second.

Most shocking is both how poorly NV's GT200 architecture performs at with tessellation and how much NV has improved its tessellation with Fermi:

Well, considering these older gen cards do not have a Tesselation unit (and I doubt Firaxis uses 'special' code for AMD's older and less programmable Tesselator present since the R600) I'd assume theres some sort of CPU overhead that attempts to do the same as the dedicated hardware in DX11 cards. That may account for the drastic differences in performance on older gen cards.

 

[dguy6789]

Q6600 is almost 2x slower than Core i5 750 (same as SC2). i3 540 is 20% faster than E8500 despite the higher clock speed of the latter, while i5 750 is still 24% faster than the i3 540. Looks like Civ5 loves everything: modern architecture, cache, clock speed, and a quad-core. So much for claims that C2D/Q is almost as fast i5/i7s in games. Granted, this type of game is an extreme example of CPU dependence.

They didn't test a real C2Q, and by that, I mean a Yorkfield with 12MB of L2 cache and at least 3.2Ghz clock speed. I want to see how one of those performs.

 

[bryanW1995]

how badly do you want to see that? my computer is an i7, my daughter's computer (on the other side of my office) is an x3350 currently running @ 3.36. only difference is i'm on an ssd and win 7 x64 while she's on xp pro x86 and a wd6400aaks.

and if I really wanted to get crazy then my mom has a q6600, though she's only at 3.06 currently.

of course, all of this would actually require that, you know, I actually stop playing civ5 for a while and start benching it instead... plus I just replaced the ram in my x58 sli and I'm having to learn how to oc it finally.

 

[dguy6789]

I'm not dying to know but it would be nice. The website only benchmarked a 2.4Ghz Q6600 to represent the entire Core 2 Quad line. Lots of enthusiasts have Q9550s or Q9650s in the 3-4Ghz range, which is substantially faster than a 2.4Ghz Q6600.

 

[RussianSensation]

Lots of enthusiasts have Q9550s or Q9650s in the 3-4Ghz range, which is substantially faster than a 2.4Ghz Q6600.

The Penryns will probably do a lot better, but Q6600 only has 8mb of cache. i7 is about 20% faster clock for clock versus a Penryn in the benches, so Q6600 is going to be even slower (i3 540 was 21% faster than E8500, despite lower clock speed in the bench). This would put a Q6600 @ 3.4ghz = i7 @ 2.8ghz level (best case scenario assuming same 20% disadvantage).

i7 @ 2.8ghz = 37 fps
i7 @ 4.0ghz = 52 fps

It will be an improvement though. At the same time, Q6600 is almost 4 years old. For such an old processor, its performance is still very impressive for most apps.