Editors Challenge Sponsored by Inte...
Video Challenge Sponsored by Intel ...
FiringSquad Video Challenge - Preli...
FiringSquad Editors Challenge Round...
||3 entry(ies) in this category
| Let the Games Begin: Vista,XP and Your Hardware (9 comments )|
by: rados (15) | Posted in cluster Editors Challenge Sponsored by Intel Round 2
Posted 75 months ago ( edited 75 months ago ) in category DEFAULT
LET THE GAMES BEGIN?
|» MEDIA (5)|
Figure 1:Let the games begin!
Figure 2: System Hardware
Figure 3: The Games
Figure 4: Average Framerates on the GeForce 7900GT
Figure 5:Average Framerates on the ATI X1900XT
So you’ve bought yourself a pretty sweet gaming rig. It has everything you could need to run today’s top of the line gaming titles. With all the research and money that you’ve invested in your system, you might think that you’re set. However, a damper on your hardware’s performance and on your gaming experience can hit you from an unexpected corner: your operating system.
For the past six years serious PC gamers have predominantly been using Windows XP as their operating system of choice. However, with the launching of Windows Vista earlier this year, gamers suddenly find themselves with an interesting dilemma. Stick with Windows XP or upgrade to the newer, sleeker, Vista?
Microsoft has recently been trying to promote both its Windows XP and Vista systems as gaming platforms through their “Games for Windows” marketing campaign. According to their website, “Games for Windows branded titles work on Windows XP and Windows Vista based PCs, including both 32 and 64 bit editions.” One might assume therefore that both Windows Vista and Windows XP offer equal game performance on the same high-end hardware, or that Vista being the newer system, might give you an edge. Benchmark testing under both operating systems reveals something different.
The hardware performance of a high end gaming rig was tested under fresh installs of both Windows Vista (Ultimate Edition) and Windows XP Pro (Service Pack 2). In order to get the most objective test results possible, the system was also tested using two different graphics cards.
XFX GeForce 7900GT
(See Fig. 1)
A homebuilt machine with the following specifications was used:
2.4 gigahertz AMD Athlon X2 4600+ dual core CPU
ASUS A8R32 DELUXE motherboard
2 gigabytes of Patriot DDR400 ram
Two 150GB Western Digital 10,000 rpm Raptor Hard drives
(For further details see Fig. 2)
Testing began with an install of Windows Vista and Windows XP on their own separate drives, followed by the latest ATI and Creative hardware drivers. To benchmark the hardware, four of the most popular gaming titles to come out in the last three years were installed. The games included as follows,
Company of Heroes (v1.4)
Doom 3 (v3.31)
Unreal Tournament 2004 (build 3369)
(For further details on the games see Fig. 3)
All of the games were run with their graphics options set to the maximum, with the exception of Doom which was run one step below the maximum on high. In addition, all of the tests were performed at a resolution of 1024x768.
To obtain an average FPS the in-game benchmarking features of F.E.A.R and Company of Heroes were used. whil slightly different methods for Doom and Unreal Tournament were adopted. Benchmarks for Doom were obtained through a third party benchmarking program from hocbench.com called “Doom 3 Benchmark version 1.2” Unreal Tournament on the other hand, was benchmarked using the “Primeval” timedemo, downloadable from 3dcenter.org.
In order to ensure consistent results all benchmarks were run through both operating systems several times. After completing the tests with the ATI card, the graphics drivers were completely uninstalled, the ATI card removed, and replaced with the NVIDIA GeForce board. Upon loading each operating system the latest NVIDIA graphics drivers were installed and the machine rebooted. Then benchmarking procedures were run exactly as they had been with the previous board.
In the end both the NVIDIA and ATI graphics boards showed a comparable drop in framerates. However, the drop occurred, not in the older OS (Windows XP), as would have been expected, but in Windows Vista. Though, from Microsoft’s pronouncements, one might expect Windows Vista and Windows XP to offer equal hardware performance, benchmarking results showed quite a different trend.
Windows XP consistently outperformed Vista in all the benchmarking tests. The difference between newer “Games for Windows” certified titles wasn’t as extreme. Older non-certified games however showed wildly disparaging results between the two systems.
For example while the percentage difference between benchmarks for F.E.A.R, a GFW certified game, was under 6% when comparing Vista and XP. Doom 3 came in above 19% on the NVIDIA card and above 34% on the ATI. The other two test games showed similar results. (See Figures 4 and 5) The newer Company of Heroes showed only a small drop in framerate while Unreal Tournament suffered a significantly greater loss.
Several factors might have accounted for the drop in performance witnessed on Vista. First and foremost one must realize that Vista is a far more complex OS than XP. At an idle desktop Vista had over 43 background processes running compared to the 25 of XP. Idle CPU usage was also higher, hovering around 2 to 6% on Vista while barely exceeding 1% on XP. Finally, the idle memory usage on Vista clocked in at around 425MB versus 330MB on XP.
Simply put, the games had to share more resources with the Vista OS, than with its XP counterpart. This was the most likely cause of the slight framerate drops seen in newer games such as F.E.A.R and Company of Heroes.
The major performance drops on older titles such as Unreal Tournament are somewhat more baffling however. It could be that older games are simply not as compatible with DirectX 10 as newer releases.The lack of Vista native Open GL support might also have compromised their performance. After all, a 34% drop in framerate on Doom 3 can't simply be due to increased system overhead.
No matter what the cause, when all was said and done, running older games on Vista synthetically downgraded both test graphics cards by one generation. Such a performance difference should be too great for the hardcore gamer to ignore.
OTHER THINGS TO CONSIDER
It should be noted that the two graphics cards used in these tests aren’t completely on par with one another. While they are of the same generation the ATI has almost double the memory and a higher clockspeed. A more comparable NVIDIA based card would have been the GeForce 7950GT, but due to economic and time constraints the represented hardware was used.
Furthermore, during the testing procedure, Windows Vista presented a few of its own challenges. The Company of Heroes Patch Installer crashed on first use. Uninstalling the ATI graphics drivers halted the system, though seemed to uninstall the drivers successfully. Additionally, some programs did not install correctly. Company of Heroes failed to place an icon in either the start menu or on the desktop, although one did appear in the Games folder. Lastly the user account control window popped up constantly with warnings of such dangerous tasks as copying files, installing patches, inserting new disks, and saving batch files in notepad. This “security feature” soon proved to be the greatest annoyance in the new Vista OS. Although such difficulties may seem minor, and probably did not influence benchmark results, a platform that is being promoted as “The most popular gaming system on the planet” should not have them.
It seems pretty clear that an operating system can adversely affect hardware performance. When faced with the new choices in operating systems today, there are a few things a gamer should consider. Windows XP gives more gaming performance, stability and value now. It also gives great performance with all currently released games. Vista may not work as well with current and older gaming titles but will ensure compatibility with the newest DirectX 10 based games.
It is somewhat disappointing that the only way to get DirectX 10 compatibility is to buy Vista. However, unless there are some future changes, that’s the way it will remain. So, for the time being a gamer’s best option is to have one hard drive with XP and one with Vista. That may not sound like much of a solution. In the end though, when you plug in your shiny new GeForce 8800 GTX, you want all your games, new and old to run at optimal capability.
|9 User Comment(s) • 8 root comment(s)|
| suibhne (65) Mar 16, 2007 - 09:59 am | Edited on Mar 16, 2007 - 11:11 am|
|The benchmarking was fairly simplistic, and this topic has already been done into the ground (including good stuff here at FS). As others pointed out, software support (particularly drivers) in Vista is still immature, so it's currently difficult to compare the two OS's across the board. It's also worth noting that Vista's memory architecture is fundamentally different from XP's. What I missed was a look at not merely why the frame drops occurred, but whether and how they might be mitigated in the future - i.e., which are due to architectural changes in Vista and which are simply due to Vista being brand-new and not yet adequately supported. This is obviously a tall order, and some of these questions can't even be answered yet.|
Your presentation and writing were admirably clear. Sure, there were a few errors, but they were obviously the exception rather than the norm.
» Login to reply to this
| Fedor (19) Mar 15, 2007 - 05:43 am|
|I'm surprised you sort of went out of your way to try to find the reasons for the slower performance in Vista whilst neglecting the widely-accepted view that it is due to the still immature drivers. Heck, nVidia got a real beating from consumers for the fact that they didn't release a driver for the 8800 for Vista for several months after launch, so the one they had to rush out was obviously not going to be perfect, and ATI has been improving theirs but it's also naturally not yet at the same level as XP's since they've been working on that for many years.|
» Login to reply to this
| mrbigulls (1) Mar 14, 2007 - 07:31 pm|
|Well explained and executed, but I wonder if using different benchmarking tools skewed results at all? Would it not have been wiser to use the same third party software on all four games - despite the availability of proprietary tools included with half - at least for consistency's sake? The differences would be insignificant, however, in my opinion. Good job.|
» Login to reply to this
» Note: You need to be logged in to write a comment!Login here, or if you don't have an account with FiringSquad, register here, it's FREE!
My Media-Blog categories
No categories created yet.