cool post of the day:
Open Standards
by: Desi_Russell 8/26/00 11:29 pm
Msg: 151902 of 151971
messages.yahoo.com
If you try to get a committee to come up with an open standard, two things happen:
1. the backbiting/politicking/footdragging means it necessarily takes a lot longer to finish the job than if a single group of motivated designers do it (ever heard of design by committee?) Ever heard of the SDRAM standard as approved by JEDEC?
2. you get a bunch of garbage thrown in to placate the various special interest groups. Ever hear of the SDRAM standard as approved by JEDEC?
Because the committee tried to make everyone happy, the SDRAM standard tried to be everyting to everybody. It was a useless pig when it came out: there were way too many modes to test that drove up testing cost, there were no standard parametric specifications that killed system interoperability and it was badly in need of being put on a diet from an operating mode perspective.
So what happened to it and how did we get to where we are now?
Intel decided after being victimized with poor SDRAM compatibility for a year, that they had to take control of the spec and finished the job that JEDEC spent about five years on but failed to complete. They specified a simplified SDRAM including parametrics that would be suitable for PC use if the specs were followed. But before that, JEDEC refused by a very political vote, to develop SDRAM Lite. There were certain members of the committee that were just too influential for the standard's good. The free market system had to take up the slack and finish the job.
I expect the results of the DDR effort (which DDR? is it DDR, DDR2, AMI, ADT or something else?) to be no better. When you have a group of guys together that are naturally competitors all trying to create a standard, there are many other agendas at play other than the purported agenda of the committee. Design by committee is not very efficient at getting things done. Ultimately the standards get passed, but that doesn't mean they are useful.
Another thing to consider is just because a bunch of memory companies want to make a chip doesn't mean it is going to serve a market need. Remember the first 1Meg VRAMS? Classic case of bumbling about. The JEDEC group that standardized it made it a x4, just as the 256K had been. Really dumb, they just didn't understand what they were doing. But it was a JEDEC standard. How about the RAMLINK standard from the IEEE? Has anyone ever heard of that? Another useless standard from a committee, many members of which helped to make the SDRAM standard at JEDEC and are now trying to make the DDR standard(s). Ever hear of the Synclink, er Synchlink, er SLDRAM standard? Micron even ran ads promoting it. Another useless standard. How about Burst EDO, another failed standard?
I'd be willing to bet that the DRAM companies would be perfectly happy building 16meg EDO devices if the market would buy them. They should be really good at getting yield on them. But no one wants them so they had to make the investment necessary to build SDRAM.
Bottom line is there is the Push dimension from the suppliers and the Pull dimension from the customers. In the final analysis, it is the Pull that is the important angle. Today we have a group of the Push side desperately trying to create the Pull, hence the DDR brouhaha of today.
Intel and Rambus will have to Pull the Memory Makers kicking and screaming into prosperity. DDR just will not make it.
As an interesting footnote, remember that DDR is also the acronym of the Deutschland Democratic Republic, AKA the former East Germany: the failed Communist regime that was responsible for killing many many innocents. I dare say that the DDR boys trying to kill Rambus have lots of similarities: an inferior system, a central controlling committee, their propagandist shills and despite all that, they ultimately were doomed to failure. DDR, yep, that name was chosen well!
-Desi R. |
