Scumbria,
It is FAR TOO SOON to conclude that RDRAM is failing in the market place, when processors that will benefit from RDRAM have yet to even be marketed.
Suppose that with the introduction of Pentium 4 it becomes clear that at and above 1.5 GHz, Rambus provides a real 15% performance boost over any other competing technology, and that this performance boost increases with increasing clock speed. It could well be the case that within 2 years, there are NO processors being sold with speeds below 1.5 GHz. Further, Samsung says (and they should know) that by January of 2002, the cost/price premium for RDRAM will be down to 5%.
People routinely pay nearly double for processors with performance premiums in the 20% range (right now, a PIII 866 is about $500 while a 933 is about $750). In the environment described above, which is not at all out of the question, one would assume that virtually everyone would be using RDRAM. Even if the cost differential was a LOT more than 5%.
I have absolutely no argument with those who say that RDRAM, in and of itself, provides no benefit over SDRAM with current (under 1 GHz) processors. However, until very recently (the advent of the 815E chipset), the fastest SYSTEMS used RDRAM, not because it was necessry, but because the fastest systems used Intel's 820 or 840 chipsets, which only use RDRAM. So, if I wanted the fastest system, I used RDRAM. The advent of the 815E makes this less clear. There are no benchmarks that I am aware of comparing the 815E to the 820E in otherwise comparable systems running comparable software benchmarks.
[The benchmarks on the Intel site compared the 815E to the 820 (not the 820E)), and were not controlled to eliminate other variables].
It is my expectation that Willamette with Tehama (and therefore RDRAM) will be the fastest system available by a WIDE margin for a LONG time (six months or more). I expect the P4 RDRAM systems to maintain a significant (10% to 33%) advantage over SDRAM even when SDRAM chipsets become available for the P4. Consequently, I do expect RDRAM to be successful in the marketplace. Time will tell if I am right.
That said, if the patents hold up, it just doesn't matter from an investment perspective. |
