X
Business

Intel's 3-D chip shows potential

More than a year ago, Lockheed Martin's 3-D business unit signed a deal with Intel to bring Lockheed's high-end 3-D technology to the consumer market. Since then, very little information about the Intel project -- code-named "Auburn" -- has been available.
Written by Loyd Case, Contributor

More than a year ago, Lockheed Martin's 3-D business unit signed a deal with Intel to bring Lockheed's high-end 3-D technology to the consumer market. Since then, very little information about the Intel project -- code-named "Auburn" -- has been available.

About the only hard information is that the chip will be called the i740, but anything else you may have read about the chip has been strictly rumor or speculation.

We managed to play with an i740 board at Comdex. Bear in mind, this is beta hardware working with beta drivers. Even so, it looks to be very interesting.

We ran 3D Winbench 98, which let us take a close look at the feature set supported by the i740. One immediately noticeable result was image quality, which was quite good - especially in MIP mapping.

The i740 didn't support the higher-end MIP-mapping features (for example, linear/mipmap/linear in 3D Winbench 98 parlance), but the more common MIP mapping modes showed none of the odd streaking or artifacting visible in RIVA 128 cards.

The chip also supports anti-aliasing, though this feature didn't seem to be turned on in a way that 3D Winbench could recognize.

The performance was equally interesting. The card we looked at was an AGP card with 4M byte of video memory, running on a 266MHz Pentium II. The score returned by 3D Winbench was 460. This places it between the 3Dfx-equipped Monster 3D (which gets around 370) and the RIVA 128 cards, which post scores around 490.

Given the overall image quality -- quite good -- this is not a bad score at all for beta hardware and drivers.

The real issue is what impact the i740 will have on the overall graphics chip market. If Intel can deliver a chip at a relatively low cost with reasonable 2-D and 3-D performance, some more marginal players may be muscled out of the way.

Unfortunately, we couldn't check out 2-D performance or real games, but a canned demo looked good -- but then again, canned demos always look good. The bottom line is that this test drive left us cautiously impressed, particularly if the price point is right (pricing isn't yet available).

Editorial standards