Pages: 1 2
During the month of August, Intel lawyers were reportedly ‘talking to’ motherboard manufacturers who have boards designed with the VIA Apollo Pro133 chipset, implying that they will bring suit if the boards are released for public consumption. In the meantime, they are apparently encouraging these manufacturers to include PC100 SDRAM support on their i820 chipset based boards.
During this same period, a sudden shortage of i440BX chipsets has forced motherboard manufacturers to scramble to meet demands. We have heard reports that a remarking operation in Asia was selling VIA 686A South Bridge chips as BX South Bridge chips (it turns out they are pin-compatible). Many believe that the shortage was intentionally created by Intel to force transition to the i810 and i820 chipsets. Traditionally, Intel has not had such problems getting manufacturers to use their new chipsets, however things are a bit different this time around.
The i810 chipset was supposed to be introduced in the Spring of 1999, however stability problems and other issues have delayed the chipset. In fact, there have been so many delays, that manufacturers are very nervous about the chipset and have not transitioned as quickly as Intel had anticipated. To further complicate Intels plans, the cost of D-RDRAM is almost 10 times that of SDRAM, causing most motherboard manufacturers to resist the Intel force-feeding of the technology to them via the i820 chipset.
As a result of these problems, most manufacturers have been relying primarily upon the i440BX chipset to be the workhorse – which Intel apparently saw as a possible threat to their plans to get everyone on to the new chipsets. At the same time, the VIA Apollo Pro+ and Apollo Pro133 chipsets were made available, giving manufacturers another option, which several took advantage of.
Intel reportedly took a very dim view of these developments and decided to put a little pressure on the motherboard manufacturers to bring them back in-line. According to our sources, Intel managed to irritate these manufacturers even more than usual. In an attempt to appease the manufacturers, Intel has crammed PC100 support into the i820 chipset, but now it seems that the performance will be worse than the BX chipset with this memory.
The next six months will be very telling. The motherboard manufacturers we have spoken to have already put a 2 RIMM / 2 DIMM design onto their i820 based boards, and some have even hinted at a 4 DIMM design in the future if the market demands it. In the meantime, VIA has included support for Virtual Channel Memory into their Slot 1 chipset, which may even result in a better performance boost than D-RDRAM for current high-end systems.
Of couse, the question remains whether D-RDRAM is a viable memory technology for the future. Some industry experts believe that eventually D-RDRAM will be the dominant memory technology for PCs, while others believe that DDR SDRAM or some other competing technology will eventually win out. If history is any indication, Rambus has a very difficult task ahead, as the majority of proprietary technologies ends up losing in the PC marketplace regardless of any technical superiority.
Most manufacturers and many consumers are very reluctant to let one company control the marketplace – and for very good reason. It is very obvious that the advance of processor technology has slowed considerably since Intel gained almost total dominance. The Pentium II core is actually the same as the Pentium Pro, which was introduced in 1995. With the Willamette delayed until 2001, this will be the longest period of time between generations in Intel’s history!
Both AMD and VIA are now poised to bring considerable pressure to bear on Intel’s market share. Though Intel maintains that D-RDRAM will eventually be the dominant memory solution for the PC, VIA and the large memory manufacturers have other plans. DDR SDRAM II designs are already in the works, and many industry experts believe that this, combined with VCM, will provide much superior performance to the average PC.
Where D-RDRAM does seem to make sense is in multi-processor systems running multiple memory intensive applications, such as true multi-media. Unfortunately for Intel, the marketplace does not seem to be willing to pay the price today for technology that is several years away, at best. With the recent IDF announcement of PC133 support, it seems that Intel finally got at least part of the message. It remains to be seen whether they will continue to listen, or will need to feel a little more pain in the form of additional loss of market share.
Be the first to discuss this article!