I've said it before, and I'll say it again. In terms of gaming performance, the CPU is not nearly as significant as the video card being used. Not even close. In order to get anything more than a marginal performance difference in these tests, we have to crank down the quality to the point where you might as well not be playing the games at all. I mean who will ever want to play Unreal Tournament 3 in 800x600? Or even 1024x768?
That being said, games are beginning to rely more and more on the processor for things like AI, physics, and game logic, so multi-threaded performance is still important. Therefore, we can't just dismiss gaming performance entirely.
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 AI 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 6 times (with real-time physics, no two runs will be the same) and the average framerate of the 6 runs are given below.
The game was run in DX9 mode with a resolution of 800x600 and all graphics effects set to Low. The Physics Quality was set to Very High.
Amazingly, we are back to the insane performance scaling we saw in the SiSoft Sandra floating-point tests. The i7 920 starts off where the QX9770 left off, and it just ramps up from there.
Unreal Tournament 3
So many games use the Unreal 3 engine that we had to start using it as a benchmark. There are two methods of benchmarking UT3 - one plays a botmatch in real time, while running in spectator mode. The other simply does a 'flyby' through a level. I would have loved to use the botmatch benchmark, but it was simply impossible to get repeatable results. I was seeing variations of up to 5 FPS from one test to another, so that idea had to be scrapped. Instead, we just ran a flyby demo:
This time results are all pretty close, but the 965 is able to overcome whatever limits all the other processors have hit.
ET: Quake Wars
Just another game I thought I would throw in, based on the id Tech 4 engine (Doom 3) which is far less popular than U3E.
With ET:QW, we're back to the insane scaling again, although the leaps aren't quite as significant this time.
Call of Duty 4
We had to include CoD 4, considering it is one of the most popular multiplayer games around right now. This is a manual test, done using FRAPS in one of the more stationary levels. The settings are 1024x768 at Medium, but the amount of bodies is set to "Insane". And believe me, there are a lot of bodies flying around in this level!
The same scaling takes place here, in tiny increments this time. I have no idea why the Phenom does so well in this test; I tested it over and over to make sure. There is no good reason for it to perform like that, and under normal usage would be well below the Q9550, like it is in every single other test performed here.
HL2 Episode Two
Finally we have the latest Source Engine game - Half Life 2: Episode Two.
There's that nice ramp again!
I don't think this gaming benchmark needs much of an introduction. It's not exactly the best real-world benchmark for CPUs, so we won't spend too much time dwelling on it. But since games are becoming more threaded, and are making more use of the CPU for things like physics and AI, the individual CPU score should be considered:
3DMark Vantage allows the Core i7 to shine, since it probably uses every advantage the CPU has over Penryn. From insanely high memory bandwidth to Hyper-Threading, Core i7 is able to excel in this benchmark. However, it's still just a benchmark, and not an actual game, so we can't get too excited about this score.
Copyright © 1999-2017 HCW Network | All rights reserved | Our Privacy Statement