Hi all; Some more of the usual EE-Times articles on computers...
System on a chip makes a giant step forward:
Process said to extract no hit in logic speed -- IBM reports leap in embedded DRAM
Customers in the networking, disk-drive and digital consumer fields already have expressed an interest in taking advantage of the on-chip bandwidth possible with embedded DRAM. Adding DRAM to a logic process usually compromises logic performance, but IBM's plan extracts virtually no speed penalty for the logic.
...
However, embedded DRAM brings advantages for image processing in a digital camera, graphics processing in a notebook computer or data processing in a networking system.
techweb.com
It is this sort of thing that will put companies with real technology, like IBM, in control of the PC industry over the next five years. I am (insert disgusting physical metaphor here) to get into a project with embedded DRAM. My current project is compatible with HDTV. I bet I can get somebody to pay me to have a good time using that sweet IBM technology next year...
Embedded DRAM is of interest to RMBS investors, due to the fact that it eliminates the necessity to use the rambus interface. That is, the rambus interface is most likely to be used by precisely those customers that would also be attracted to embedded DRAM.
In addition to rambus, there are several other techniques for increasing the bandwidth out of DRAM. I had to specify a DRAM for a project a few weeks ago, and reviewed what was available, etc. Sync-link and rambus (SLDRAM and RDRAM, respectively) have interfaces that are too complicated from a signal level point of view (for the small production project I'm on.) But double-data rate (DDRAM) was awfully attractive, I will either use that or standard SDRAM. The cool thing is that I don't have to decide until pretty late in the project which I will use. (I.e. in about 3 weeks.) Here is an interesting article on DDRAM:
IBM samples 256-Mbit parts, joins 10 firms in endorsing format -- DDR picks up steam as next-gen DRAM choice
IBM was among 11 DRAM vendors and large computer manufacturers to voice their collective support for the DDR specification last week, putting the DDR SDRAM back in the race for memory slots alongside the Direct Rambus DRAM (D-RDRAM).
Though microprocessor vendors Intel, Advanced Micro Devices and Cyrix have endorsed the Rambus architecture for PCs, as has Compaq Computer's server division, Rambus has yet to convince most server manufacturers that the narrow, 16-bit Rambus channel can effectively scale to the multi-gigabyte main memories used in larger servers, analysts said.
techweb.com
Incidentally, I feel the same way about the scalability of RDRAM as the paragraph quoted above. The basic idea with RMBS is to get more bandwidth out of fewer pins. But ball grid arrays have revolutionized the packaging problem. The FPGA I am targetting to has nearly 600 pins. With that many available, I would have to be 50% more crazy than I am to bust my ass to reduce pin count.
The more I think of it, the more I like the idea of using DDRAM. I'm going to bring it up tomorrow with the project lead (my favorite drinking buddy), and see if he agrees. My guess is that since we just decided to retarget the FPGA from an ALTR part to a XLNX, which is about 50% faster, he will agree that it is possible, so it is just a matter of getting chips from the vendors...
-- Carl |
