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: Patrick Grinsell who wrote (13542)	6/29/1999 7:16:00 AM
From: SFM1	Read Replies (1) \| Respond to of 16960

Patrick, This response is being posted to provide additional info concerning Rambus. In your link to the "GUM" paper the following is an excerpt: "Because memory bandwidth so quickly becomes the limiting factor in these architectures, the most common differentiator (and also predictor of performance) of consumer graphics cards is the bit-width of the shared memory bank. If the rendering controller is well-designed (with today's competitive market, most successful ones are), it is extremely likely that the memory system, not the controller, will limit the performance of typical professional workloads. Micro-architectural features such as multiple pixel pipelines, dual texture caches, and superscalar processing, while perhaps academically interesting, only help to ensure that the memory remains the real bottleneck." Following is a link to a post from the Rambus thread with an in depth article about the Pixel Fusion 150: techstocks.com But the segement concerning "Bandwidth": "PixelFusion's FUZION 150 chip demonstrates the benefits of Rambus technology for graphics applications requiring increased memory bandwidth and performance," said Dave Mooring, vice president and general manager, Personal Computing Division at Rambus Inc. "PixelFusion's FUZION takes full advantage of the performance scaling capability of Rambus technology to deliver powerful graphics computing solution for Windows NT systems." I am by any means no expert but IMHO this is the direction most memory intensive computers/components and other devices will take. SFM1

To: Patrick Grinsell who wrote (13542)	6/29/1999 10:03:00 AM
From: Jeffrey P	Read Replies (2) \| Respond to of 16960

Okay Pat.... I hate to sound impatient... (tapping toe) .... ..... .... But it's tuesday .... ... ... <grin> Jeff "what article?" P.

To: Patrick Grinsell who wrote (13542)	6/29/1999 11:21:00 AM
From: John Farrell	Respond to of 16960

Patrick, Cirrus Logic's success/failure with their Laguna family had more to do with driver performance and 3D technology rather than it's use of Rambus memory, although that was a concern as well. If you think Rambus can't be VERY successful in a graphics application, just look at the success of the Nintendo 64 which has used Rambus memory since it first shipped. One of the large reasons that the PC graphics vendors didn't jump on the Rambus was that a mixed memory type design was not very feasible. Most of the competing integrated graphics parts (S3's ViRGE family, ATI's RAGE family, atc.) at the time had bond-out options that a board vendor could tie one pin of the ASIC high or low and have either an EDO-DRAM or a SD/SGRAM memory interface. Rambus required a completely different ASIC. Cirrus Logic's problems had to do with price/performance which have always been a key player as to who is successful and who gets culled out. It's quite possible that if 3dfx's Banshee hadn't shown high enough 2D performance that they could have died on the vine before getting Voodoo 3 out and without Voodoo 3, I doubt the STB merger would have happened. Everything is a matter of timing and execution and those that forget this (Video 7, Western Digital, IIT, Ark Logic, Sierra Semiconductor, Weitek, Cirrus Logic, Rendition, etc.) are just memories in the PC graphics business. -John