techweb.com
May 17, 1999, Issue: 1160
Section: Semiconductors
Rambus delay stirs quick-fix specter
Mark Ellserry
In early February, Intel Corp. announced it was delaying
the introduction of the Intel RDRAM platform-a technology
that was to feature the 800-MHz Direct Rambus DRAM
interface-until September 1999. That unexpected industry
revelation immediately resulted in a flurry of market
repositioning by all affected parties.
Due to this delay, many PC OEMs are now seriously
considering PC133 as a viable alternative to Rambus for
all system price points. PC133 would certainly fill the gap
in product until the September release of Direct RDRAM.
However, the fundamental question for the industry
remains: Is it worthwhile to look to PC133 SDRAM as a
temporary standard for those high-end systems previously
planned for Rambus, or should the market wait the extra
three months for the Rambus debut?
Looking at this issue in the most positive light, an argument
can be made that PC133 is a viable alternative to Rambus
since the current motherboard architecture remains the
same. Yet for middle and high price-point systems, the
question remains: Why invest resources in a
lower-performance product-possibly not available until
July-when Rambus-based systems will be on the market in
September?
There is no clear indication that there will be enough
manufacturing capability to build 133-MHz chipsets for all
ranges of PCs. The consequence of changing product
direction again is that the PC-buying public-both
corporations and consumers-could become confused.
Further, consider the ramifications of temporarily moving
to PC133. Considerable engineering effort goes into
introducing a new PC platform. In the case of PC133, the
high-end motherboards will not be manufactured by Intel.
Instead, the PC OEM will be required to invest a large
amount of engineering effort to design such a motherboard,
or to purchase it from a PC- board manufacturer.
Once designed, the new product must be qualified for
production and sale. PC OEMs must ask themselves if this
hefty investment is worth a 90-day delay. In these difficult
times, that delay does not justify changing product plans in
midstream.
In the event that the PC industry does move away from
Rambus for high-end systems, we may discover that such
market maneuvering and apparent uncertainty results in a
decrease in third-quarter PC sales.
Market actions of this nature may have a devastating effect
as their consequences cascade down to PC consumers. If
the industry does choose to use PC133 as a near-term
stopgap, PC customers could become confused about which
PC generation to buy, and may even suspend any significant
investment in PC133 systems in anticipation of the release
of the high-performance Rambus version.
When Rambus got the Intel nod over DDR
(double-data-rate) SDRAM in 1997, all Rambus licensees
geared up to produce this high-performance device. We
fully understood that there would be an enormous amount of
technical effort required to make Rambus successful, and
we have worked steadily to reach that goal. Yet it has only
been in the last few months that the hardware has been
available for debugging, system testing, and validation.
The work accomplished through January of this year
produced a stable system with excellent operation at a
600-MHz data-transfer rate. However, PC OEMs decided
they did not want to introduce systems with only 600 MHz.
Intel was therefore forced to delay its chipset until
September, when it would have 700- and 800-MHz
transfer-rate performance to meet PC OEM requirements. This is where we stand today. From my perspective, it is
foolhardy for the industry to try to slip in a temporary fix to
high-end systems offerings when there is so much technical
momentum going for Rambus.
Earlier, the PC market made a resounding commitment to
Rambus, and all the players have been solidly behind this
next-generation DRAM technology. Ramp issues remain,
but there is high confidence within the industry for an
expedient resolution. So let us not lose sight of the progress
we have made and of our Rambus objectives. And above
all, let us not put our well-orchestrated product plans on
the back burner. Rather, let us keep pressing forward with
Rambus.
-Mark Ellsberry is vice president of marketing at the
semiconductor division of Hyundai Electronics America,
San Jose, a division of Hyundai Electronics Industries Co.
Ltd.
Copyright ® 1999 CMP Media Inc.
|
