Re: Pentium bug and Oracle
Date: Sat, 03 Dec 94 21:17:43
Message-ID: <3br9j9$cbh_at_dallas1.connect.net>
In article <12a.8253.905_at_compudata.com>,
<barry.roomberg_at_compudata.com> writes:
> Path:
news.connect.net!news.sprintlink.net!hookup!swrinde!pipex!uunet!cmp
udata!barry.roomberg
> Distribution: world
> Newsgroups: comp.databases.oracle
> Subject: Pentium bug and Oracle
> From: barry.roomberg_at_compudata.com (Barry Roomberg)
> Message-ID: <12a.8253.905_at_compudata.com>
> References: <3bnuls$olh_at_crcnis3.unl.edu>
> Date: Sat, 03 Dec 1994 12:16:00 -0500
> Organization: Compu-Data BBS -=- Turnersville, NJ -=-
609-232-1245
> Lines: 14
>
> -> For folks running Oracle on a Pentium system (such as the
NCR/AT&T
> -> 3000 series), is there any reason to be concerned about
Intel's
> -> Pentium FPDIV bug?
>
> Don't know, but please post any followups for the rest of us.
> I'm just moving to Pentium after 486, and am seeing a 7 to 1
> speed increase on my new system. I'd hate to find out I
> can't use it!
>
>
>
> Internet: barry.roomberg_at_compudata.com (Barry Roomberg)
> This message was processed by NetXpress from Merlin Systems Inc.
>
I am not concerned with ORACLE itself. I suspect there is minimal floating point instruction use within the database engine itself, refer to your DBA guide, internal representation of numbers and dates and you will see that numeric data manipulation is integer based internally, even for floating point numbers.
What is your application? If it is life support I think I'd have
strong reservations. If you're running accounting for your
company,
I would think a benefit/risk analysis would weigh in favor of
the pentium platform. If this is not a very quirky condition
we would have heard a lot more, much sooner. I understand that
Intel is releasing chips certified to correct the error, give
yourself a chip swap and upgrade that clunker your using for
development to a pentium and put the new pentium in your production
box.
Received on Sat Dec 03 1994 - 21:17:43 CET