Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 8 Apr 1995 17:34:02 +0800 (CST)
From:      Brian Tao <taob@gate.sinica.edu.tw>
To:        FREEBSD-CURRENT-L <freebsd-current@FreeBSD.org>
Subject:   Re: Disk performance
Message-ID:  <Pine.BSI.3.91.950408171607.787F-100000@aries.ibms.sinica.edu.tw>
In-Reply-To: <199504080735.AAA14324@gndrsh.aac.dev.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 8 Apr 1995, Rodney W. Grimes wrote:
> 
> I tend to agree that it does look a little high.  But remeber we still
> have to bcopy the bytes between user land and the kernel, which well
> be somewhat slower on the 486 class of machines due to the data path
> width to memory.  (It does not scale 2X like you think it would, it seems
> the memory desiges in Pentium systems leave a little to be desired :-().

    Then you have some people who think PCI is a load of crock and
think we should have stuck with VME instead.  Note that my lack of
knowledge of bus architectures does not permit me to hold an opinion
either way, so flames will be cheerfully ignored.  ;-)

> Yes, IDE drives load the cpu down far more, as they not only have to
> bcopy the bytes to and from user land, they also have to bcopy them
> to and from the disk.

    I presume that EIDE drives and drivers that support DMA wouldn't
suffer from this particular deficiency then.  Sorry, I don't mean to
pick your brains in public like this.  I hope someone else is learning
the basics of disk transfer too.  :)

> > Does it really make much difference with only one spindle to swap on?
> 
> Probably not much, though you don't want to locate it far away from
> the most active area of the disk, IMHO.

    I suppose not, especially when these systems I have usually don't
need to touch swap.  I remember laughing at someone a few years back
because he wanted to buy a separate 100-meg drive for his NeXT as
additional swap space.  Now it seems like a perfectly reasonable thing
to do.  :-/  :)

> Especially with out detailed information about how the kernel implements
> the buffer copies to and from user land.  I'm pretty sure the SGI boxes
> use page flipping,

    Hmmm, the only page flipping I know of is a double-buffered
animation technique used on the Apple II hi-res graphics screen.  :)

> systems in them from other tests.  AHh.. just ask phk@FreeBSD.org why
> the sun's are so bloody slow.  Most 486 machines can outperform SS1000 :-).

    Indeed:

% sunmodel
Machine gate: 'SPARCsystem 600MP (4 X 390Z55)' '(Model 541)' @ 50.0 MHz, GX
% uname -a
SunOS gate 5.3 Generic_Patch sun4m sparc
% uptime
  5:11pm  up 5 day(s),  7:07,  9 users,  load average: 1.36, 1.29, 1.38
% time dd if=/dev/zero of=/dev/null bs=102400 count=10240
10240+0 records in
10240+0 records out
0.63u 34.44s 0:35.25 99.4%
                                                                                
    About 5MB/sec slower than the 486's.  A damn waste of CPU too... all
I see people do on that machine is read news and mail.  :(

> I expect to be able towards the end of next week show some really
> amazing numbers on a Pentium system with respect to memory speeds.
> Let's all hope that the new EDO and Pipelined Burst SRAM stand up
> to the theories and we start to see 100++MB/sec main memory speeds on
> a Pentium like we should.

    Wow!!!
-- 
Brian ("Though this be madness, yet there is method in't") Tao
taob@gate.sinica.edu.tw <-- work ........ play --> taob@io.org




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSI.3.91.950408171607.787F-100000>