Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : 3DFX -- Ignore unavailable to you. Want to Upgrade?

To: Chip Anderson who wrote (4986)	7/1/1998 10:40:00 PM
From: R Stevens	Read Replies (3) \| Respond to of 16960

First S3 Savage Benchmarks Revealed (Just in case there are any TDFX & S3 shareholders on this thread) next-generation.com First S3 Savage Benchmarks Revealed Next Generation Online was invited yesterday to S3's Santa Clara offices to play with the Savage 3D. Here's what we've found. July 1, 1998 The Savage 3D was one of the biggest surprises at the recent GDC this past May. With a little more time to work on their drivers and a new revision of the chip, S3 has found itself with a very competitive product. All tests were conducted upon Pentium 2 400's with 64MB RAM and vsync on. The Savage card in question was an 8MB unit and the Voodoo 2 control specimen was a 12MB reference board. Silicon tested was not final (a new rev should be arriving this week) and in certain areas, performance increases should range from 5 - 40 percent. The image quality right out of the gate was solid with good color saturation though the DAC seemed slightly darker than the 3Dfx Voodoo 2. In terms of performance the chip was again solid. One of the first tests we benchmarked was a custom made large texture Quake 2 demo running at 640x480. The test is designed to show the benefits of S3's texture compression. The level in question consisted of 20MB worth of textures (more than a 12MB Voodoo 2 card can hold without swapping) and the results were surprising. The Voodoo 2 clocked in at 23.2 frames per second due largely to the constant texture swapping. The Savage came in at 37.1 frames per second. Is this how an average Quake 2 game is going to function? No, as they generally don't take advantage of so many textures, but as games begin to push the boundaries of texture caches (e.g. Unreal) the benefits of texture compression are clear. The next test was a D3D test with Incoming. Initially, the 16-bit rendering of the Voodoo 2 wasn't quite as clean as the 24-bit rendering of the Savage. Lens flares didn't suffer from some of the cheesecloth patterns that Voodoo 2 offered. Both ran extremely fast. When both cards were set to 16-bit (for purposes of being consistent) the Voodoo 2 turned 58.98 frames per second while the Savage posted 65.31 frames per second. The next test was another D3D test using Forsaken. At 800x600, the S3 pulled 73.87 frames per second while the Voodoo 2 hit 72.54. We then ran the demo again with the S3 at 1024x768 where it managed 55.09 frames per second. A single Voodoo 2 doesn't support resolutions that high thus there is obviously no corresponding score. The editors also witnessed some software MPEG2 decoding for DVD which was nice. Motion compensation reduced CPU usage significantly (from 60% to 30%). Naturally, there is also support for hardware MPEG2 decoding should an OEM be so inclined. All in all, the tests were impressive and boards using the chip should cost less than the editors had originally thought, making it an even more viable competitor in the low-end market. According to S3, consumer products should cost between $99 and $150 depending upon the RAM configuration.