iXBT Labs - Computer Hardware in Detail






Multi-Core Processors in 3D Games

<< Previous page

     Next page >>

Call of Duty 4

It's a multiplatform game, which engine evolved from Quake 3 Engine. But there is probably nothing much left of it. In return, we have good support for multiprocessor systems, which is a must for game consoles. In modern terms, the game does not have very high requirements to a graphics card, so it may be limited by CPU speed. Even though the next part of the game is already out, there are no significant changes in its engine because of its multiplatform nature. Let's take a look at the results:

It's almost the same situation as in Crysis. Only the difference between the single-core and dual-core configurations is bigger here. A strong performance boost is demonstrated only in our comparison of single-core and dual-core processors. In the Low mode (AF and AA are enabled anyway) the difference is close to twofold, but it's noticeable even in High modes -- over 20%. There is a small performance difference for three versus two cores as well, but it's below 3%. A single core is clearly not sufficient in this game, but it hardly needs more than two cores. Let's take a look at the CPU usage:

  Core 1 Core 2 Core 3 Core 4
Average 53.4 57.4 25.7 59.4
Maximum 87.5 87.5 57.2 100

The game, drivers, and the operating system apparently distribute the load between cores better in this case -- three cores out of four are loaded evenly, and even the fourth is working. However, this game does not spend as much resources on processing Direct3D calls as Crysis, that's probably why performance is not limited by the speed of a single processor core.

Let's recalculate the require CPU power per single core. The average CPU usage in this game is about 195% in our case. It's close to the power of a dual-core processor. Together with the 100% usage of one of the cores, it explains a little gain from the third core in 1280x720. On the whole, Call of Duty 4 needs more than just a single-core processor (as fast as a single core in Core 2 Quad Q6600), but a corresponding dual-core CPU will cope with this game just fine.

Enemy Territory: Quake Wars

This multiplayer game is based on the DOOM 3 Engine from id Software, which has always been notable for good multi-CPU support. This engine is rather processor-intensive, as it uses a CPU to compute and apply shadows. Besides, a CPU is used to compute physical interactions.

What's important, unlike DOOM 3, Quake 4, and Prey, Enemy Territory: Quake Wars is all about battles in open ground. It may also have a serious effect on the result, because open scene rendering traditionally depends on a CPU more than on a graphics card. Let's test it:

But in this case a CPU has even weaker influence on performance than in the previous games. Results of multi-core processors are not as peachy as in the previous applications. Even though we get 45% of performance gain from the second core in the easy mode, it disappears in heavier modes. There is some performance difference in the medium resolution, but it decreases to the level of a measurement error in the highest resolution. Let's take a look at the core usage:

  Core 1 Core 2 Core 3 Core 4
Average 19.7 29.8 31.8 44.4
Maximum 37.5 59.4 54.7 67.2

The operating system distributed the load between all cores almost evenly, although three cores got more work than the fourth. Anyway, there is apparently not much work for a CPU to get performance gains from more than two processor cores. Let's calculate the required processor power -- about 125%. That is, the game needs a tad more than one core of a 2.4GHz Core 2 processor in 1280x720 with anisotropic filtering and antialiasing. It's a perfect explanation for our results -- only the second CPU core is actually useful.

Race Driver: GRID

It's another game based on the multiplatform engine. It evolved from Colin McRae Rally: DiRT. This engine is very good at multiprocessing. Depending on the number of processors in a system, it creates several threads for 3D rendering, physics computing, AI, audio data, loading data from a storage drive, Force Feedback, etc. Configuration files include options to distribute threads among up to eight processors.

Of course, the game cannot use all capacities of eight processors, but developers chose the right approach. The game does not have high requirements to graphics cards, so we can expect its performance to be limited by a CPU. So we may finally see benefits from the third and/or fourth core. Unfortunately, this game does not offer normal benchmarks, so we had to use FRAPS:

So, there is indeed a small performance difference between three/four- and dual-core configurations, to say nothing of a single-core system. But it's small. The game apparently runs well with just two cores in our test resolutions. But single-core CPUs definitely won't do for this game, as they are 1.5-2 times as slow as the dual-core configuration. That's a true multithreaded application, developed that way from the very beginning. Besides, it does not have high CPU requirements. Even the third core makes itself useful in low resolutions. Let's take a look at processor core usage:

  Core 1 Core 2 Core 3 Core 4
Average 66.5 87.2 53.0 58.6
Maximum 85.9 96.9 82.8 75.0

All four cores are more than half-loaded, peak load being 80-95%. It's a clear sign that game performance is limited by weak processors in 1280x720 and lower (maximum graphics quality settings, MSAA 4x and AF 16x). And what do we have in terms of a single core? 265%! That's it -- the first game in our tests that apparently needs more than two 2.4GHz cores (of a Core 2 processor) in 1280x720. So performance gains from the third core are not accidental, and have nothing to do with a measurement error (it's FRAPS, anything can happen).

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.