“Quad Core vs Dual Core” Shootout: Intel’s Q9550s vs. E8600, Part Two
3DMark06
3DMark06 still remains the number one utility used for a system benchmark. The numbers it produces aren’t indicative of real-world gameplay – or any gameplay in general – and for that reason we really dislike using it to compare different systems. However, as long as the rest of the tech world uses it to evaluate gaming performance, we will too. We find it is mostly useful for tracking changes in a single system, what we are mostly doing now. There are four “mini-games” that it uses for benchmarking graphics, as well as two CPU tests. The scores are “weighed” and added together to give an overall number and there is a further breakdown possible with these mini games that we are charting for you.
Here is a scene from one of the four benchmark “mini games” used to benchmark GPU performance. It will give your PC a real workout even though the default resolution is only 12×10.
Here are the results of our 3DMark06 benchmark comparison using the benchmark at its default settings:
Here we see almost linear scaling with CPU frequency. If all games are like this, we would all overclock and use quad-core CPUs as it is consistently “outperforms” the Core 2 Duo – even when it is clocked +260 Mhz higher than the Core 2 Quad. Well, now let’s look at the mini-game frame rates:
Here is the identical situation exemplified in the mini-game’s frame rates. ‘Playing’ the 3DMark06 Mini-games would require an overclocked quad core CPU to get the best performance out of our HD4870 or GTX280. So, lets move on to the second benchmark, Vantage.
GTAIV and L4D would have been good games to add to this suite. Looks like there still isn’t a huge difference in dual vs. quad as long as you have enough speed, even in games that are multithreaded, although I think you’ll see bigger differences when you move to your Xfire tests.
Thanks for the comments. L4D is Source Engine and will scale just like HL2 which I will test next. It takes quite a bit of time to create reliable custom timedemos. Look for CoD4 to be added in the CrossFire tests which will be in a more detailed review in a couple of weeks.
In the meantime, I have a nice comparison of Phenom II vs. Athlon X2-6000+ – both with SLI’d GTX280s; probably next weekend. I will present an extreme example of CPU scaling – in some extreme cases, the frame rates doubled with the faster CPU.
Ah, I thought L4D was multi-threaded unlike a lot of Source games, but I could be mistaken.
And I’m looking forward to your AMD shootout, because I essentially have an Athlon 6000+ and I feel I’m CPU-limited in a few games, specifically TF2.
It is true that the newer Source games are enhanced and more demanding than the earlier ones. Multi-threading is probably improved. I probably will pick L4D up as a benchmark, although Source is really not one of the more demanding engines.
Here is the Source Engine Wiki:
http://developer.valvesoftware.com/wiki/Source_Engine_Features
“Multi-core. Source engine games utilize multi-core processors in both the PC and XBox 360 to deliver high-performance gaming experiences.”
The benchmarking is complete on the AMD shootout and I believe you will find the results moving from 6000+ to Phenom II much more extreme than in this current one.
We start out with a 6000+ in an AM2 board – nforce 590 based (Abit AN932x SLI); then we upgraded the board to an AM2+ ASUS Crosshair Formula II board (nforce 780A based), for Phenom II. Benchmarks were run on the performance difference from nforce 590 to 780a with the 6000+. Then with Phenom II. Everything else in the system stayed the same with SLI’d GTX280s. Of course, the OS was reinstalled but stayed as Vista64 throughout the benching.
The source engine shows very little benefit from multi-core. The touted improvements Valve promised (scaling to n-cores) have only been demonstrated in their map building tools, but not in actual games.
It’s also capped at ~300 FPS, so it skews maximum framerates above that.
Nice, makes me feel fine with a dual core still. I had thought that by now quad cores would be implemented better, guess that isn’t the case.
Thanks for the info. I’m just speaking from experience with the Source engine. When I enable multicore support in TF2 (using mat_queue_mode), my framerates (subjectively speaking) jump up by at least 50%. In some spots on maps, my framerate will drop below 30 without mat_queue_mode set to 2, but with it my framerate usually stays above 40. The bad part is that the game likes to crash itself and my system.
Also L4D, even during heavy action, runs (and looks, IMO) much better than TF2, although I haven’t done any testing with the game’s setting itself.