Re: tlug: Bus master IDE performance under Linux

["Stephen J. Turnbull" <turnbull@example.com>]

--------------------------------------------------------
tlug note from "Stephen J. Turnbull" <turnbull@example.com>
--------------------------------------------------------
>>>>> "Francis" == Francis Brian O'Carroll <ocarroll@example.com> writes:

    Francis> How well do current versions of linux hande "bus master
    Francis> ide" disks?  Is it a cheaper alternative to scsi with the
    Francis> same performance, or not?

linux-2.0.29/Documentation/ide.txt says:
<quote>
NEW!    - Bus-Master DMA support for Intel PCI Triton chipset IDE interfaces
                - for details, see comments at top of triton.c
[ snip ]
For work in progress, see the comments in ide.c, ide-cd.c, and triton.c.
</quote>

JWT will probably weigh in with a more informed opinion.

    Francis> One of the researchers here did some tests with netbsd
    Francis> (he probably had to do some kernel hacking since he
    Francis> mentioned having to consult the disk controller chip's
    Francis> spec sheet).

Or he may just have needed to specify parameters to the driver (you
can do this with Linux, I assume *BSD permits too).

    Francis> And speaking of benchmarking, are there any system
    Francis> benchmarks for Linux machines apart from BogoMips?

Have seen references, so it's worth doing a net search, I think.  But
I'm going to lunch :-) I know that there were some Linux vs. NT
benchmarks done at DEC for Alpha architecture; some effort went into
it, because DEC figured they could sell a lot of Multias to people who
didn't need bullet-proof servers or at least plausible deniability (to
the boss when the machine crashes) but did want cheap performance.
This was available from DEC's home page, but I don't have the URL
handy.  I don't know if the benchmarks are publically available but I
think they are.

I'm not sure what you're asking for; of course you can do Dhrystones
and Xstones and stuff like that; and there must be a
comp.*.benchmarks.  When you say system benchmarks for Linux machines,
do you want to compare one Linux machine against another, or do you
want to compare Linux against other systems, or what?  And why do you
care?  If it's just bragging rights, BogoMIPs should be fine.  If it's 
work-specific performance, then you should be looking for something
work-specific rather than Linux-specific.  No?

-- 
                            Stephen J. Turnbull
Institute of Policy and Planning Sciences                    Yaseppochi-Gumi
University of Tsukuba                      http://turnbull.sk.tsukuba.ac.jp/
Tel: +81 (298) 53-5091;  Fax: 55-3849              turnbull@example.com

-----------------------------------------------------------------
a word from the sponsor will appear below
-----------------------------------------------------------------
The TLUG mailing list is proudly sponsored by TWICS - Japan's First
Public-Access Internet System.  Now offering 20,000 yen/year flat
rate Internet access with no time charges.  Full line of corporate
Internet and intranet products are available.   info@example.com
Tel: 03-3351-5977   Fax: 03-3353-6096