rec.autos.simulators

Delay seen for Intel's Camino chip set

Worker Workin

Delay seen for Intel's Camino chip set

by Worker Workin » Sat, 13 Feb 1999 04:00:00


> Delay seen for Intel's Camino chip set

> SAN FRANCISCO Difficulties in taking the Intel Camino chip set and the Rambus DRAMs it
> supports to the speeds necessary to accommodate a 133-MHz system bus may delay the rollout of
> personal computers based on the full-speed Rambus technology, sources said. Instead, Intel
> Corp. may support a 100-MHz system bus for its initial Pentium III processors, using Direct
> Rambus DRAMs operating at 600 MHz. The Pentium III, or Katmai, will formally launch next
> week. Analyst Mark Edelstone at Morgan Stanley Dean Witter said the 820 Camino chip set
> that's needed to support the 800-MHz Rambus memories will be delayed to the third quarter. PC
> OEMs, he said, will go ahead with June introductions of Rambus-based systems, but with the
> memory speeds slowed to 600 MHz.

Who in the world is being limited by RAM speed?  Are there ANY
applications that end up waiting on 100mHz RAM?  Am I understanding this
right?  Wouldn't it be more worth their while to be focusing on PCI
speeds or even coming out with an 800mHz CPU before they worry about
800mHz RAM?
Randy Cassid

Delay seen for Intel's Camino chip set

by Randy Cassid » Sat, 13 Feb 1999 04:00:00




> > Delay seen for Intel's Camino chip set
...
> Who in the world is being limited by RAM speed?  Are there ANY
> applications that end up waiting on 100mHz RAM?  Am I understanding
this
> right?  Wouldn't it be more worth their while to be focusing on PCI
> speeds or even coming out with an 800mHz CPU before they worry about
> 800mHz RAM?

Uuuhhhhh, who in the world _isn't_?  Profiles of GPL, for one, clearly
show large CPU usage spikes, and virtually all of them are cache misses
(ok, every once in a while there's an arctan or something).  When
running GPL on a 400MHz machine, probably 20%-30% of the CPU time is
spent twiddling its thumbs during cache misses (and, yes, we did
optimize these portions of code (and the data they operate on) to
reduce both the number, and impact, of the misses).

Randy

Cowen Wilso

Delay seen for Intel's Camino chip set

by Cowen Wilso » Sat, 13 Feb 1999 04:00:00

No they are talking about the Bus speed of the processor and memory.  That
is what regulates how fast information is transferred too and from the
processor.

Cowen

Chath

Delay seen for Intel's Camino chip set

by Chath » Sun, 14 Feb 1999 04:00:00

Delay seen for Intel's Camino chip set

SAN FRANCISCO Difficulties in taking the Intel Camino chip set and the Rambus DRAMs it
supports to the speeds necessary to accommodate a 133-MHz system bus may delay the rollout of
personal computers based on the full-speed Rambus technology, sources said. Instead, Intel
Corp. may support a 100-MHz system bus for its initial Pentium III processors, using Direct
Rambus DRAMs operating at 600 MHz. The Pentium III, or Katmai, will formally launch next
week. Analyst Mark Edelstone at Morgan Stanley Dean Witter said the 820 Camino chip set
that's needed to support the 800-MHz Rambus memories will be delayed to the third quarter. PC
OEMs, he said, will go ahead with June introductions of Rambus-based systems, but with the
memory speeds slowed to 600 MHz.

"Although these systems will not offer the full bandwidth benefit inherent in the Rambus
memory architecture, we believe that PCs with the 600-MHz version of Direct Rambus DRAMs will
offer twice as much bandwidth as PC100 versions when they run bandwidth-intensive
applications like video and 3-D graphics," Edelstone wrote in a report. Major PC makers had
expected to introduce initial full-speed Rambus systems in June, moving to full-volume
shipments in the second half. That schedule, while never officially endorsed by Intel, may
now be pushed back, with a 100/600-MHz specification inserted.

"We are on track to introduce the Rambus technology in 1999," said an Intel spokesman. At
Rambus, marketing vice president Subodh Toprani said that the 600-MHz specification referred
to by Intel as "the down bin" has been in the Rambus product plan since 1997. "At this point,
there is not enough data. No one can say there won't be an 800-MHz introduction as planned,"
Toprani said. Edelstone wrote that his firm still believes "Intel is on track to introduce a
version of the chip set that will support 600-MHz versions of Direct Rambus DRAMs toward the
end of the second quarter. However, we believe difficult technical hurdles will likely cause
upwards of a three-month delay in Intel's ability to deliver a version of the 820 chip set to
support the full Direct Rambus DRAM spec of 800 MHz."

Intel has said it will reveal modified plans for its Rambus program at the Intel Developer
Forum set for Feb. 23-25. A likely scenario is support for a two-stage introduction, with the
initial Pentium III processor linked to a 100-MHz system bus by a Camino chip set that
supports RDRAMs operating at 600 MHz. Once Camino silicon is ready that can support the 133-
MHz system bus, and once yields on the full-spec RDRAMs reach sufficient quantities, PC OEMs
could introduce the 133/800 systems. The issue is how to balance the overall performance of
the system, sources said. Without a chip set that can support a 133-MHz system bus, there is
little need to move to the 800-MHz Rambus technology.

The 600-MHz Rambus parts can provide 1.2 Gbytes/second of peak bandwidth, compared with 1.6
Gbytes/s for the 800-MHz RDRAMs and 500 Mbytes/s for the PC100 SDRAMs. The bandwidth is
divided among the system bus, the Accelerated Graphics Port and the PCI bus. PC OEMs would
like to refresh the top ends of their product line in the second half. But one source said
that some IT managers are reluctant to move to the Rambus technology while the year-2000
problem is putting demands on the software used in corporate America. Another source noted
that delays in the delivery of full-spec Rambus parts would make more room for the PC133
SDRAM specification to be adopted in the personal-computer sector.


rec.autos.simulators is a usenet newsgroup formed in December, 1993. As this group was always unmoderated there may be some spam or off topic articles included. Some links do point back to racesimcentral.net as we could not validate the original address. Please report any pages that you believe warrant deletion from this archive (include the link in your email). RaceSimCentral.net is in no way responsible and does not endorse any of the content herein.