Blake, re:drip, drip, drip.
First off, I think you're smart enough that you don't really buy into this "Pricewatch" pricing game.
As for RIMM's becoming competitive with DIMM's, during the CC Tate said the 256/288Mb RDRAM was the key to bringing the price of RDRAM to <20% adder to SDRAM. He predicted this would occur in 2H'01.
We don't know if he was referring to the "4i" initiative of Intel, (changing the architecture to 4 banks), or if he meant the current solution. In either case, one issue in the mainstream PC segment is the dual channel i850 for P4. You need two RIMM's and two C-RIMM's in the platform.
And while there will be 16X16Mb RDRAM's, there will also be 16X16Mb SDR/DDRDRAM's. Granularity with 256 Mb is therefore not an issue for SDR/DDR. (Unless you intend to make 64MB DIMM's, which wouldn't make a lot of sense).
For example, you can make 128MB modules using four 16X16Mb chips for any of the solutions. Here's an example of a new 128MB SDRAM module with four 16X16Mb chips. (DDR versions should be in the works).
usa.samsungsemi.com
So in the mainstream segment, I think dual channel is a major issue. I suspect the minimum implementation for RDRAM will be 256MB, (two 128MB RIMM's and two C-RIMM's). This is double the minimum I see for SDR/DDR. (128MB using 4 chip's). I think this needs to be factored into the price comparisons.
I think P4/i850 will be a winner in the Work Station segment, (where more memory is normally specified), but I think the mainstream will eventually use a new chip set for P4. (But even Intel states the PIII will dominate volume through 2002).
BTW, It appears that Intel has already acknowledged an SDRAM solution for P4 using a new chip set.
Will it be SDR or DDR?
eoenabled.com
JMHO's |
