We spent some time testing ATI and NVIDIA's latest low end parts recently. Today, we are revisiting budget performance. After all, what's more "budget" than dropping the graphics chip on the motherboard?

The real reason for the existence of the 6200 TurboCache and the X300 HyperMemory is to provide an affordable setup up from integrated graphics. Graphics companies have to be quite careful when developing integrated and budget cards to not cannibalize their business for either product. There must be sufficient performance on integrated graphics to run current software so that end users are satisfied, but cheap budget cards also need to provide a sufficient step up from integrated solutions.

It is difficult to get the level of performance that we want to see in integrated graphics, but it is important to consider the fact that users of integrated graphics may not need any 3D support and would certainly not want to pay for it. Cost is a first concern, and compatibility reigns over performance.

So, how does this generation's integrated graphics perform compared to the recent budget cards on the market? Well, that's what we are here to find out.

From ATI, we have the Radeon Xpress 200 for Intel platforms. We will be pitting this solution against the Intel GMA 950 on our 945G board. These integrated graphics cards will be compared to our recent discrete budget graphics tests. Even though the platforms won't be as similar to what we usually want, these tests are bound by the graphics capabilities of the cards almost completely. We should get a good idea of how these parts stack up against each other.

The Test

Our test will involve the Intel D945GTP and the ASUS P5RD1-V Deluxe board. There are a few caveats in looking at these solutions compared to each other and our previous budget card tests. First of all, though the Intel platforms can use the same processor, we are stuck with DDR 400 for our ASUS board. The 945G board uses DDR2 533. As we will see, this won’t change the outcome of our tests. There will be ATI boards featuring DDR2 as well, but the advantage (if there is one) is minimal.

Our setups were as follows:

Intel D945GTP
3.6GHz Pentium 4
1GB 533MHz DDR2 3:3: 3:12
Seagate 7200.7 120GB HD

ASUS P5RD1-V Deluxe
3.6GHz Pentium 4
1GB 400MHz DDR 2:2:2:8
Seagate 7200.7 120GB HD

We will run our tests with high quality settings in order to see if integrated graphics are as capable as the budget add-in cards that we tested recently. This will also demonstrate minimum performance in the games that we test, as end users will likely want to sacrifice some of the eye candy for playability.

Doom 3 Performance
Comments Locked


View All Comments

  • fishman - Monday, August 1, 2005 - link

    You can get these with either shared memory or dedicated memory - which configuration was used in the tests?
  • IntelUser2000 - Thursday, June 2, 2005 - link

    Sis and Via's integrated graphics solution is even more below Intel. They are more than one generation behind Intel's graphics. Even though Sis 660 or something had 2 pixel pipelines and hardware T&L, Intel's Extreme Graphics 2 beat it hands down, and IEG2 only has single pixel pipeline.
  • IntelUser2000 - Thursday, June 2, 2005 - link

    Apparently dual core doesn't help much if you look at other Intel GMA950 benchmarks. The only one helped is HL2 score, which went up by 30%, and 3dmark2005, which is multi-threaded so it doesn't count.
  • ET - Tuesday, May 31, 2005 - link

    I'd find it interesting to see results with Intel's low end dual core CPU. Since the CPU is used for vertex processing, dual core might have a good effect on this.
  • crucibelle - Saturday, May 28, 2005 - link


    I agree with you completely!

    I particularly wish that the reviewer would have ran a benchmark for Sims 2. Perhaps they can do this in the near future? I certainly hope so.

  • tbrooks40 - Friday, May 27, 2005 - link

    nice write up...

    i don't necessarily agree with akozak - memory is cheap enough now-a-days that seeing a basic system with 1gb of memory won't be all that surprising. i doubt that it's the norm but it's conceivable with memory prices continuing to drop.

    i have a question, one i'm sure akozak would roll his eyes to, would a dual core chip increase the performance of integrated graphics?

    i know an entry level system wouldn't likely have a dual core chip - i'm more curious than anything after reading the dual core performance article.
  • ET - Thursday, May 26, 2005 - link

    #19, I don't think it's useless. They're benchmarking the graphics, so seeing how fast it can perform given optimal conditions is helpful. With a slower CPU and less RAM you get other bottlenecks, so the graphics scores will have less of a difference, but this won't really give you a better feeling for the integrated graphics.

    BTW, I'd consider 1GB pretty much necessary for a system running integrated graphics, if they take RAM away from the system. 512MB may be too little in such a case.

    That said, an article about game performance on entry level machines (Semprons with integrated graphics or whatever) might be of some interest.
  • flloyd - Thursday, May 26, 2005 - link

    Thanks #18 but I know about refresh rates. Unfortunately the 915G that I have is much more "stable" and clear at 75Hz than 85Hz so I have to deal with that for now.
  • abakshi - Thursday, May 26, 2005 - link

    Well yes, comparisons to older standalone cards like an R8500/GF4 would interesting, but you'd have to test these integrated chipsets on the same platform to get any meaningful results, which isn't possible here (at least yet) -- PCIe instead of AGP...
  • iwodo - Thursday, May 26, 2005 - link

    Can anybody check if they have updated the benchmark? I am sure it wasn't this bad last time i check. As my news reader inform me something in this article has changed.

Log in

Don't have an account? Sign up now