Zeev,
"Finally got it"???? You mean finally said it! :8)
I doubt that you all will agree with this; but it seems to me that Bilow, you and I are all correct to one degree or another.
I say that Carl's view is correct. (i.e., that the "Big I" figured back in the early 90's that bandwidth the was going to find future pin/costs limitations and bought a ticket on the RAMBus as the Party Line.)
Your theory, that INTC wanted to speed up the klunky DRAM data rates because they were about to fry our eye-balls with blazing CPU speeds to handle video on demand cable systems through humongous (my editorial license:8) packetized klumps of digitized PPV movies and WWF Monday Nite Nitro events, is also true.
But its clear to me that the both of you NEED my Dumbing Down Theory (henceforth known as "The DDT") to make sense of this mess. Here's what really happened:
--------------------------------------------
Back in the early 90's the paranoid within the EE community look at all the nasty, expensive and difficult Pins popping out all over the place, and low and behold they find some calculations that show the ever increasing need for Pin capacity will soon result in a devastating Tornado directly in the path of Moores' Milk Wagon Law.
...After all, there are nervous nellies in every profession and who can blame them. Nobody had ever announced a Moores Law of "Pin Counts."
So they look around (quickly mind, because the Cash Cow was producing so much milk money that they had to find Emergency Expenditures to prevent some of it from bursting out of the Fab and into the parking lot:8).
Amoung the many wild ass investments they find to insure against this impending disaster, is that multi-million dollar equity deposit in the RAMBus Line. Not only does it promise a simple solution for the threatened Milk Wagon run, but it also has the added bonus of providing another "sponge" for the immediate flood of milk emanating from the Cash Cow.
So things are going along pretty well until, by 1996, the EE's tending this dairy herd get wind of some disturbing developments in weather control. The previously forecasted Pin Count storms aren't developing like they should. The packaging guys have begun to figure out how to seed the Pin Count clouds to alleviate the buildup of bandwidth pressure.
But at the same time the entire known world is watching the values of Turner BS, Time, Warner, MCA, Viacom, GE, AT&T, et al., get "Blowed Up" by riding this bandwidth busting "Content" binge. Even MSFT gets involved snarfing up every piece of artistic image IP it can lay its hands on in a move to corner the high brow end of the Content arena to keep its encroaching Browser busy and lookin good for the Cable users of the future.
So there's Andy watching the sunset on his PC centric world at the same time that the post '87 equity market is really beginning to party like its 1999. In fact it has gone positively certifiable at the mere whiff of the bandwidth winds forecasted for the next Sun spot cycle.
Just then the Vice-President of Cash Cow Management and the new CEO of the RAMBus Line burst in with the alarming reports. The Pin Count Clouds have dissipated. The RAMBus Line is Dead, Dead, Dead!!
Well, being the crafty old CEO he is Andy declares "There are always alternatives!" And rushes up to Redmond for a meet with MSFT. There he learns the true secret of top management:
"It makes no difference whether or not your product serves a real consumer need. The only thing that matters is your ability to make them THINK they need your product above all others."
So Andy rushes back down to the San Andreas where he calls in all the Division poobahs to brain storm the situation. "All right guys. How do we maintain our control of the Cash Cow herd, and at the same time make this Bandwidth Market Party before it really IS 1999? Any suggestions? You Claude, how can we work this Bandwidth angle into our Milk products?
Claude: Now look Boss, we can redesign these utters & sputters to give you GHz of milk. That's no problem. We just packetize larger Klumps into every carton. You'll have Milk out the Whazoo as far as Cash Processing Units are concerned. But I can't help you on the data rates from the CPU to the Main Memory. I've already got on die cache and these bridge muxing things with UMA subsystems sticking out all over the place. The most I'm going to be deliver on the FSB by 1999 is a paltry little 400Mb/s. If you want more than that you're talking about putting the Main Memory thing in the carton too, and there's no way that gets done by 1999.
Andy: 400Mb/s!!! Is that all?! Jeeze! I can hear'em now: "INTC crashes Bandwidth party with ancient spin the bottleneck CPUs!" We have got to do better than this.
Damien(from Marketing): Well we still have that RAMBus investment sitting out in the shed without a problem to solve. It looks to me like we just found one for it. If we anoint it as the "solution" to the emerging "PC memory bandwidth bottleneck" before anybody even notices the bottleneck, we'll be the life of the big Equity Party from now 'til 1999 at least.
Andy: By Jiminny! You've got somethin' there Damien! Even the dumbest Analyst can see that "66MHz" is a lot smaller number than 400Mb/s! Call Legal! I want a license deal on that RAMBus and I don't care what it costs. We are the leaders and we are going to lead on this Bandwidth solution thing NO MATTER WHAT!
Claude: Uh,... about that RAMBus sir. I really don't think -
Andy: Shut up Claude. If you want to be Big in this business you've got to take BIG risks. Damien where's that RAMBus guy?!
------------------------------------------
And the rest as they say is woeful history. :8)
0|0 |
