iXBT Labs - Computer Hardware in Detail






Multi-Core Processors in 3D Games

<< Previous page

     Next page >>

World in Conflict

It's the second RTS game in our review that may need a multi-core processor. It also features a built-in benchmark. But unlike Company of Heroes, it reflects the true gaming performance, although it's not very cinematic.

From the point of view of technologies, it's one of the most interesting and advanced real-time strategy games. It heavily loads both a CPU and a graphics card. You will see it, when you take a look at the diagram -- the frame rate with maximum graphics quality settings is very low for such a relatively powerful testbed as ours. Unfortunately, the built-in benchmark does not show tenths of the average FPS values. We like performance measurements to be more accurate.

However, the game favors multi-processor systems. Performance gains in this game appear, when we upgrade from a single-core to a dual-core system and from a dual-core to a triple-core CPU in the very least. Our single-core system is outperformed by a dual-core configuration by 1.5-2 times depending on a test resolution. It's a big difference. So performance in this game is apparently limited by a CPU. Let's take a look at the CPU usage:

  Core 1 Core 2 Core 3 Core 4
Average 41.3 35.8 38.1 48.3
Maximum 84.8 56.6 67.7 84.4

We can see that all cores are loaded almost identically, which speaks of good parallel operations. None of the cores was loaded by 100% during our test, so CPU clock rate is not a bottleneck here (even of a single core).

On the whole, the game requires at least 160% of total CPU usage (for a Core 2 processor operating at 2.4GHz). That is it needs at least two cores, or even better -- three cores (taking into account multithreaded nature of this application). A higher performance level of a quad-core system probably has to do with system processes using free resources, not interfering with the game.

S.T.A.L.K.E.R.: Shadow of Chernobyl

This game was present in our previous analysis a year and a half ago. So we decided to leave it be. But this time we did not use the demo recorded by developers, we used our own replay, so our results should be different. However, we know that the game generates a heavier load on a GPU. A CPU may limit performance of this game only in low resolutions. Besides, only one core is very important. This is a multithreaded game, but the main thread that limits performance is only one. Let's see what happens in different configurations:

We can see that the original S.T.A.L.K.E.R. gets a good boost from the second core in 1280x720. The difference becomes almost unnoticeable in heavier modes. Interestingly, the average frame rates for two heavy modes do not differ much. It means that performance is limited by the speed of one core, and multi-core processors will be of no help here.

Performance boost from the second CPU core reaches 10-30%. This game apparently lacks optimizations for multi-processor systems. Or perhaps it's not the game, but the graphics API, which cannot distribute draw calls between processors. Let's see how much the cores of Core 2 Duo Q6600 were loaded by the test:

  Core 1 Core 2 Core 3 Core 4
Average 97.8 1.9 17.3 40.6
Maximum 100 10.4 38.6 72.3

Here is a practical proof of our assumptions. Even though this is apparently not a single-threaded application (it actively uses three cores), the main bottleneck is in the speed of the main core, which is almost always limited by 100%!

Theoretically, S.T.A.L.K.E.R. needs about 160% of a single 2.4GHz core (Core 2 processor). That is a dual-core processor will suffice. But as performance depends more on the speed of a single core, CPU clock rate must be high. A fast dual-core processor can be faster than a slow quad-core or triple-core processor.

S.T.A.L.K.E.R.: Clear Sky

This is a prequel of the original S.T.A.L.K.E.R. The game engine was improved, of course, and its characteristics may have changed. For this very reason we included Clear Sky into our tests. Let's see what have changed as far as multiprocessing is concerned. We want to know whether the engine is still limited by the speed of a single CPU core. Let's take a look at the diagrams:

Judging by the diagram, this is our old S.T.A.L.K.E.R., just twice as slow. The second core demonstrates a similar performance boost. And we have almost the same invisible performance limit in 1680x1050 and 1920x1200.

It's clearly a multithreaded game. And the second core provides up to one third of the boost in average frame rate. But it happens only in the low resolution. The situation changes, when the resolution grows, and the speed of a single CPU core starts to limit the game performance again. We'll see it now:

  Core 1 Core 2 Core 3 Core 4
Average 99.1 1.5 12.8 1.2
Maximum 100 6.3 23.4 9.4

Everything is the same -- the first core is loaded to the brim (99.1% average usage -- that's the bottleneck for sure). The other cores process minor tasks, either system services or secondary game procedures. There is even no sense in calculating the total CPU usage, as the speed depends only on the clock rate of a processor.

Anyway, a singe core does not suffice (over 115% of its resource are used even if performance is limited by the speed of a CPU core), but the game cannot use more than two cores effectively either. So the best choice is a fast dual-core processor here.

Write a comment below. No registration needed!

<< Previous page

Next page >>

blog comments powered by Disqus

  Most Popular Reviews More    RSS  

AMD Phenom II X4 955, Phenom II X4 960T, Phenom II X6 1075T, and Intel Pentium G2120, Core i3-3220, Core i5-3330 Processors

Comparing old, cheap solutions from AMD with new, budget offerings from Intel.
February 1, 2013 · Processor Roundups

Inno3D GeForce GTX 670 iChill, Inno3D GeForce GTX 660 Ti Graphics Cards

A couple of mid-range adapters with original cooling systems.
January 30, 2013 · Video cards: NVIDIA GPUs

Creative Sound Blaster X-Fi Surround 5.1

An external X-Fi solution in tests.
September 9, 2008 · Sound Cards

AMD FX-8350 Processor

The first worthwhile Piledriver CPU.
September 11, 2012 · Processors: AMD

Consumed Power, Energy Consumption: Ivy Bridge vs. Sandy Bridge

Trying out the new method.
September 18, 2012 · Processors: Intel
  Latest Reviews More    RSS  

i3DSpeed, September 2013

Retested all graphics cards with the new drivers.
Oct 18, 2013 · 3Digests

i3DSpeed, August 2013

Added new benchmarks: BioShock Infinite and Metro: Last Light.
Sep 06, 2013 · 3Digests

i3DSpeed, July 2013

Added the test results of NVIDIA GeForce GTX 760 and AMD Radeon HD 7730.
Aug 05, 2013 · 3Digests

Gainward GeForce GTX 650 Ti BOOST 2GB Golden Sample Graphics Card

An excellent hybrid of GeForce GTX 650 Ti and GeForce GTX 660.
Jun 24, 2013 · Video cards: NVIDIA GPUs

i3DSpeed, May 2013

Added the test results of NVIDIA GeForce GTX 770/780.
Jun 03, 2013 · 3Digests
  Latest News More    RSS  

Platform  ·  Video  ·  Multimedia  ·  Mobile  ·  Other  ||  About us & Privacy policy  ·  Twitter  ·  Facebook

Copyright © Byrds Research & Publishing, Ltd., 1997–2011. All rights reserved.