The next developer to release an SMP patch was Infinity Ward, when they released their 1.01 patch for Call of Duty 2. CoD 2 is considered by many, including myself, to be the most visually impressive game currently available. Everything from the map detail to the volumetric haze created by smoke grenades, the game is by all accounts very taxing on your system. If a developer could create a patch to let a computer’s processor be used more effectively and offload certain physics and netcode calculations to another core, the resulting increase in performance for such a demanding game would probably be tremendous. The joint statement from Infinity Ward, Activision, and Intel is as follows:

Notice that this statement does not make any forward-looking claims about the quantitative benefits of the patch. The only way to get an idea for the performance increases from the SMP patch is to run some tests.
{mospagebreak title=Test and Se}
Test
Before starting the actual test, I would like to point out another performance variable with Quake 4 and OpenGL games in general. Back in late October/early November of 2005, stories were totally flooding the internet about some driver “hack” that ATI had that would improve their cards’ performance in games that utilize OpenGL. ATI cards have pretty much always been inferior to NVIDIA in OpenGL games, so the notion that this situation might be reversed was big news. It turns out that the “hack” was nothing more than the Catalyst version 5.11 drivers. The new drivers were supposedly able to deliver another 15%-20% performance increase in OpenGL games for the Radeon X1K series. Eventually, users discovered the update had a positive effect on X800 series parts as well. This claim is also too significant to ignore, so it will also be tested.

Setup

• Shuttle RS300+IX150 based motherboard<
• Intel Pentium 4 3.0GHz with Hyper-Threading
• ATI X800XT All-In-Wonder
• 2×512MB Mushkin Enhanced PC3200 RAM
• Creative SoundBlaster Live! 24-bit
• 80GB Western Digital Caviar Ultra ATA 133
• Video Drivers: Catalyst versions 5.10, 5.11

(All parts running at stock speeds for this test. Audio was not disabled.)
(Note: in order to enable SMP once the Quake 4 patch is installed, the console command “r_useSMP 1” must be entered.)

Every test was run with the following programs running in the background to mimic a real world gaming setting as closely as possible: mIRC, Xfire, AIM. Speedfan, ZoneAlarm firewall, and AVG virus scanner were also running for the tests.
{mospagebreak title=Quake 4 Patch}
Quake 4
To get a good idea of the performance gains that are generated, it is necessary that we first test the game with the variables that will give us the “lowest” performance: Catalyst 5.10 and SMP disabled. Our testing of the Quake 4 patch consisted of the creation of a NETDEMO in a fast-paced multiplayer setting, then using the console command “playNetTimeDemo” to obtain scores. So starting off we have a direct performance comparison between having the patch and not having the patch on the lowest in-game performance settings while using Catalyst 5.10 drivers

You can see from the test that the 1.0.5.0 patch does in fact increase performance. The question is: By how much? Well, according to the test, the SMP patch increased the systems frame rates in Quake 4 at low quality on Catalyst 5.10 drivers by an average of 15.8%. In the world of competitive FPS gaming, 15.8% is HUGE.

The next thing we want to find out is how performance is affected by the SMP patch on high quality settings. So, for this test we again used the Catalyst 5.10 drivers and turned the Quake 4 in-game quality settings all the way up.

This test also shows a significant performance increase when using the SMP patch. You can see that the effects of the SMP patch trickle off as the resolution increases, but this is more than likely the result of the video card having a harder time pumping out frames at high resolution and the highest detail. Overall, at the resolutions that gamers use, the average performance increase from the SMP patch was 14.0%. Not quite as high as on low settings, but impressive nonetheless.
{mospagebreak title=Quake 4 Drivers}
Having tested the performance on Catalyst 5.10 drivers with and without the SMP patch on all notable combinations, let’s move on to the “OpenGL hack” Catalyst 5.11 drivers.

The first test will be just like before, showing the performance increase of the SMP patch on low quality settings.

The first thing you’ll notice is the slightly more consistent readouts, a testament to both the stability/performance of the Catalyst 5.11 drivers, and CPU limiting that takes place in Quake 4, but more on that later. The results this time indicate that, again, the SMP patch has generated a substantial performance increase. This time the average increase through all four resolutions was 12.63%, a bit less than before. However, notice if you will that the general frame rates themselves are all quite a bit higher than they were before. It seems as though the effectiveness of the SMP patch is dampened slightly when used in conjunction with high performing drivers like the Catalyst 5.11.

Following the trend, the next test will have the same settings as the previous one, but this time the in-game settings in Quake 4 will all be set to the highest possible quality. Again, this will give us an idea of the effectiveness of the SMP patch when playing Quake 4 on high settings.

Here we have some interesting results. With the settings cranked, the SMP patch on the Catalyst 5.11 drivers rendered an average performance increase of 12.8%, slightly higher than the 12.63% received on low settings. This is probably the result of the native Anisotropic Filtering and Anti-Aliasing that the X800 series cards afford. We assume this also has something to do with the tremendous ability of the Catalyst 5.11 drivers.

SMP Results
The average performance increase from the SMP patch throughout all 4 tests was 13.8%, which is not the 25% claimed by id’s initial statement. However, 13.8% increase on a game like Quake 4 can mean up to about an 8.7 FPS boost. I don’t know about you, but I think I speak for most gamers when I say that I will take an extra 8.7 FPS any day.

Catalyst 5.11 Results
According to the results of the tests, the Catalyst 5.11 drivers provided a bigger increase in performance than the SMP patch. For tests run on high settings, the Catalyst 5.11 drivers gave a performance increase of 22.8%. This is amazingly high, and in fact is higher than what we even expected. A 22.8% increase in Quake 4 can mean a boost of well over 14 FPS – big number there. For tests run with low quality settings, the performance increase was a bit smaller, at 15.6%. This is more around the number we expected, and can mean an increase of almost 10 FPS.
{mospagebreak title=Call of Duty 2}
Call of Duty 2
The first thing to note about Call of Duty 2 is that it is a DirectX game. It does not use OpenGL, and therefore there should be no performance difference whether we use Catalyst 5.10 or 5.11. These tests were run using the Catalyst 5.11 drivers. First off we have the effect of the SMP patch on the lowest quality settings.

You can see right away that the trends are obviously not reminiscent of the Quake 4 results, meaning that this game is not nearly as CPU limited. However, the potential performance increases can still be high considering it is such a demanding game, and indeed we noticed an average boost of 25.1%. Now, 25.1% might seem like a small number in a hardware world where numbers can be in the trillions, but to put this into perspective: if I was 25.1% taller, I would be taller than Yao Ming. That’s right NBA, watch out. In a game like CoD 2, a 25.1% performance increase is absolutely staggering – but that is on low settings.

This next test will be for CoD 2 on the highest possible settings. Playing CoD 2 on the highest settings at the highest resolution is pretty much the most demanding thing you can tell your computer to do these days. So with this SMP patch, we don’t expect too high of a performance increase due to the more prominent limitation coming from the video card.

Indeed, the average increase from the SMP patch in CoD2 at the highest possible settings and at resolutions that gamers use was just 10%. This is the lowest increase we have seen from any of the tests, but 10% in CoD2 on high detail is certainly nothing to scoff at. A 10% increase can easily result in frame rates going from just playable to pretty enjoyable.
{mospagebreak title=Conclusion}
Conclusions
Going into this test we weren’t really expecting that much. Our aging Pentium 4 system would surely not yield any results that were truly impressive, perhaps a 5 FPS gain here and there at the most. Boy, were we wrong.

id Software, Infinity Ward, and Intel have produced pieces of software that are absolutely incredible additions to any dual-core gaming system. Every test that we ran showed off the tremendous advantages of enabling SMP for demanding games like CoD 2 and Quake 4. The patches produced an average performance increase of 15.2%; 17.5% in CoD 2 and 13.8% in Quake 4.

In addition to the results of the SMP patch, we also saw the effect of using Catalyst 5.11 (and later) drivers in OpenGL Quake 4. In our tests, we achieved a performance increase of 19.2% in Quake 4. This is an incredible number considering all we had to do to acheive it was update the video drivers.

Both patches were created in a collaboration with Intel, and as such they are only said to work on Intel dual core processors (Pentium-D, Extreme Edition 840 and 955) and Intel Hyper-Threading processors (Penitum 4, Exreme Edition). However, tests around the internet have shown that the patch also works on dual core AMD systems - and sometimes to a greater extent.

It should also be noted that these SMP patches will most likely work better with a real dual core processor than they do with Hyper-Threading. The Quake 4 1.0.5 patch could easily live up to its 25% increase on a similarly configured system running dual core, we are not saying that the claim is innacurate in all cases, just ours.

In light of these tests, we VERY STRONGLY recommend that anybody who has a modern dual core system should update their games to utilize the SMP ability of their computers. In addition, we see no reason for ATI users not to upgrade to Catalyst 5.11 or later. The performance increases in all cases are too good to be true, but they are.