RSS Feed

HCW Tech Blog

For the latest info on computer hardware, tech, news, video games, software tips, and Linux, check out our new improved front page: HCW Tech Blog

Reviewed by: Carl Nelson [12.01.07]
Manufactured by: AMD, Intel



I first wrote about the importance of CPUs in today's games - or lack thereof - in our Core 2 Quad review. The fact remains that it's the video card that has far more influence on how well a game plays - and this is more than ever with today's cards.

That is pretty much the purpose of this review - if you are building a system primarily of gaming and are within a certain budget, it's a much better idea to take $100 from your CPU budget and buy one of these dual core processors, and put that extra cash into your video card budget.

With graphics cards getting faster, more CPU horsepower is available to do other crucial computations in games. Things such as AI, particle effects such as smoke and fire, pathfinding, and real-time physics effects have just as much impact on the immersion of a game as the graphics. We chose to use Crysis and Call of Duty 4 in our CPU reviews, and test them in a way that attempts to remove the video card from the equation. Crysis represents the most technologically advanced gaming engine to date, and COD4, well, kicks ass. We also include 3DMark 06's CPU Test, as it specifically benchmarks the processor in terms of gaming. Even though it's not exactly a real world test, the benchmark scores should be even more pronounced, allowing us to see exactly how a CPU can can allow games to become more immersive than ever. Here's some info from the 3DMark 06 Whitepaper:

Both CPU tests use our new game engine, and rely on AI, physics and game logic to generate a multi-threaded workload that can be distributed on multiple processors, cores or even on a single processor. Ageia PhysX library and D* Lite path finding AI algorithm are produce demanding CPU loads. Tests are run in a fixed frame rate of 2FPS for more equal CPU loading. Resolution is locked to 640x480 to decrease GFX influence of performance. The shader profile is locked to 2_0 and no dynamic shadows are used. The D* Lite AI algorithm generates unit path requests in a dynamic path finding grid, where each unit represents a moving obstacle, and paths sync back at 200ms-600ms intervals. The complexity of the path request fulfillment varies; the dynamic re-planning algorithm can re-use the state of previous searches. The Ageia physics uses 87 units and their rigid bodies at 20ms physics steps. Professional reviewers can disable a second CPU or other core of either a virtual or physical dual core system for comparable results regardless of the number of cores or CPUs.

Here are the scores from the 3DMark 06 CPU test; a GeForce 8800GT was used in all our gaming tests.

3DMark 06

This result tells us that dual-core processors perform relatively the same, depending on clock speed. Because the test is heavily threaded, the quad-core pulls way ahead. Let's see if this occurs in actual games though...


Chunks of stuff are flyin' everywhere!

Crysis comes with built-in CPU and GPU timedemo benchmarks. The benchmarks are run via a script, similar to those seen in Quake engine games. Rather than being a recorded demo simply played back though, the graphics and physics effects are all done in real-time. This makes Crysis a great benchmarking tool across the board, since the CPU has to calculate all the effects as if you were playing the game. Best of all, it uses portions of the actual game, rather than a 'worst case scenario' custom level. What you see here is what you get when you play this part of the game. The demo is run 4 times (with real-time physics, no two runs will be the same) and the average framerate of the 4 runs are given below.

The game was run in DX9 mode with a resolution of 1024x768 and all graphics effects set to Medium. The Physics Quality was set to Very High.

Crysis, the most advanced game around, doesn't even utilize four CPU cores in a purely CPU-bound scenario. To be honest, I was expecting this graph to look more like the 3DMark 06 graph. Instead, we see a direct correlation to CPU speed, not necessarily CPU cores, and the AMD chips are once again holding their own in this gaming test.

Call of Duty 4: Modern Warfare

SAS: Ka-boom

Since COD4 doesn't include a benchmarking utility, manual benchmarking was required. For this, I ran the same 30 second portion of the game five times on each CPU. An average of those 5 runs were given. I chose the beginning of the "Heat" level because it started off with a lot of explosions, and it has to calculate a lot of ragdoll effects with a ton of enemies being thrown about, and more firing your way. Also, the same 30 seconds can be spent sitting in the exact same spot, so runs are easily repeatable and benchmarks aren't affected by gameplay as much.

The graphics were set to Low, running in 1024x768 with no AA or AF. Most of the effects are disabled - no shadows, no specular map, no depth of field effects, etc. Ragdoll effects and Bullet Impacts were on though, and Number of Corpses was set to Insane.

Finally, we see a real game being direct affected by a CPU having four cores. However at this point, it takes turning off graphics effects to the point of making a game look like crap for us to see a 12% performance increase from "twice the CPU power". Still, it may be a sign of things to come... As it is now though, Dual Core is the way to go for gaming.

Next Page: (System Performance: PCMark Vantage)