Date: Fri, 1 Mar 1996 23:55:04 -0800 (PST) From: Jeffrey Hsu <hsu> To: chat Subject: Greenman in the News Message-ID: <199603020755.XAA17467@freefall.freebsd.org>
index | next in thread | raw e-mail
Article: 80808 of comp.sys.intel
Message-ID: <032311Z01031996@anon.penet.fi>
Newsgroups: comp.sys.intel
Date: Fri, 1 Mar 1996 03:22:23 UTC
Subject: Intel fixed an iP6 chipset bug (also known as "Orion")
INTEL QUIETLY FIXES BUG IN MICROPROCESSOR
by Tom Abate
San Francisco Examiner
SAN FRANCISCO - In an incident reminiscent of the Pentium math flaw, Intel
Corp. apparently has fixed a bug in the chips that surround its new Pentium
Pro microprocessor, without explicitly telling customers that its newest
system ever had a problem in the first place.
The flaw apparently involves the Orion chipset that works hand in glove with
the new Pentium Pro or P6 microprocessor. The Orion acts like a pipeline,
carrying computations from the P6 to other parts of a computer system, such
as network devices, hard disk drives and memory chips.
But in complaints that first surfaced on the Internet, the Orion chipset
comes across as a clogged drain, slowing down data transfer to such an extent
that Intel's speedy P6 ends up being all but useless for many tasks that
require high-speed input and output.
Moreover, frustrated users say Intel has been vague about whether there is a
problem or how bad it might be, leaving thousands of early adopters of the
expensive P6 to rely on Internet rumors to guide purchase decisions.
"You cannot get solid information from Intel," said San Francisco State Pro-
fessor Tom Holton. "They have clammed up."
Holton had asked Intel whether the rumored flaw would affect a $10,000 Pen-
tium Pro workstation he wanted to buy for his speech recognition experiments.
Intel spokesman Tom Waldrop said Thursday his company had not hidden the
Pentium Pro's input-output problems. He said Intel posted a note on the
World Wide Web Nov. 6. It said that certain circumstances "can reduce the
throughput of the entire I/O system." However, the statement did not mention
the Orion chipset.
David Greenman, a programmer in Portland, said members of Intel's P6 devel-
opment group had confirmed the Internet rumors after he used an inside
contact in the group to ask about the sluggish transfer rate of a P6 system
he bought in December.
"I found out very quickly there was a known problem with the Orion chipset,"
Greenman said.
He said Intel engineers had been kind enough to loan him a P6 system with a
revised chipset that solved his transfer problem. Greenman is known as a
leading author of a free version of the Unix operating system called FreeBSD
and helps maintain one of the world's largest Internet sites.
Intel's Waldrop said Greenman's problem was that he had been using his Pen-
tium Pro as a server computer to dish out information on the Internet. Early
versions of the chip were not designed for that. But because Greenman was a
noted programmer running an important Internet site, Intel loaned him a proto-
type of a P6 system with chips designed for network use.
"You should not construe this as a bug," Waldrop said.
help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199603020755.XAA17467>