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.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: Scumbria who wrote (120679)	7/21/2000 11:32:29 AM
From: Ali Chen	Respond to of 1572620

<Rambus' product (DRDRAM) is performing very poorly in the marketplace, so Rambus has instead decided to hold the SDRAM industry hostage with a couple of very broad, obvious patents that probably never should have been granted.> Actually, none of the RAMBUS "inventions" ever worked, in reality. All Rambus claims are around a bus of single-line memory chips spaced at 0.1" within the same time domain, all on the 3-4" long wires running at fairly low clocks of 250MHz (as I recall). They also proposed these short ram-busses to branch like a tree, and this apparently did not work either. The original Rambus memory supposed to share the bus between addresses, commands, and data. This never worked as proposed because of collisions. It seems that when it came to implementations, Intel and RAM makers requested more manufacturable RIMM arrangement, where chips are mounted flat. The flat-like mounting requires much more spacing between chips, and the bus has grown to 20" long, and therfore signals started to cross time domains breaking the theoretically nice picture. With target frequency rising higher and higher to compete with plain SDRAM, they started to run into all ugliness of "signal integrity". As an example, a single chip connection on a RIMM has 3 sources of inhomogeneity along the transmission line. Hence a 16-chip RIMM has 48 sources, plus at least four potential impedance mismatches across the connector. So, a 2-RIMM memory system has 100 inhomogeneities along the signal propagation path, and the clock has 200. Even if the mismatch is 1% in random directions, imagine what kind of garbage you may get in result. Actually, today' the clock has 100 too, since the original invention (to pass a single clock wire down from the bus to controller and forward it back to chips) did not work either - Intel controllers capture the incoming clock and re-issue the forward clock with internal logic. In conclusion, every Rambus "invention" was non-working. No wonder they have to resort to other style of business... - Ali

To: Scumbria who wrote (120679)	7/21/2000 11:54:25 AM
From: semiconeng	Read Replies (3) \| Respond to of 1572620

Scumbria, Re: Dolby has a successful technology That's true....now....., but it wasn't in the beginning of Dolby Labs. When I was working for a Medical Instrument Company in Palo Alto Ca, in the early 90's, I worked with a senior engineer who was on the design team at AMPEX. At that time, they were developing the first Video Tape Recorder. No, not the video cassette recorder, the original video tape recording technology. I found out he was involved, when I went to a party at his house, ummmmm...mansion, in the Palo Alto Hills, near Stanford University, and spotted an Emmy on his mantle. "What's that for?" says I, and the story unfolded. Then he said... "If you're that interested, let me introduce you to a friend of mine who worked with us...... Thomas Dolby. Yup, that Thomas Dolby. And let me tell you, according to him.... NOBODY was beating a path to his door for the original Dolby Noise Reduction technology..... in the beginning..... My how times have changed..... :-) SemiconEng