RE: Does FreeBSD 4.11-STABLE support the 8237R?





--- Mark <admin@xxxxxxxxxxxxxxxx> wrote:


-----Original Message-----
From: owner-freebsd-questions@xxxxxxxxxxx
[mailto:owner-freebsd-questions@xxxxxxxxxxx]
On Behalf Of Danial Thom
Sent: vrijdag 2 juni 2006 18:28
To: Scott Hiemstra; 'FreeBSD-Questions
Questions'
Subject: RE: Does FreeBSD 4.11-STABLE support
the 8237R?

--- Scott Hiemstra <shiemstra@xxxxxx> wrote:

Did you say you are running a server?
That MB is only suitable for
desktop use, as it has the slowest
ethernet controller known to man
on a 32/33Mhz bus. Running this MB as a
server is like putting
cheap, skinny tires on your porsche.

DT

Personaly, I appreciate your dedication to
maximum performance but
please notice this thread is in reference
to swapping a MB for another
MB and coments like yours are not
appreciated.

Would you prefer if I had stated?

"I have the same board in a crappy server
running 4.11 (FreeBSD
4.11-STABLE #0) and no problems to report."

Please notice I never said what the box was
doing nor did I ask for
your opinion of what MB/NIC I use in my
systems. This SERVER is pur-
pose built and runs stable 24/7 as a low
volume outbound mail server
so the performance of the NIC is not my
primary concern. Please keep
your useless comments to yourself as they
do nothing but waste disk
space, CPU time and the valuable time of
people who attempt to help
others on this list.

Scott

So if someone is planning on using a crappy
motherboard as a server its
not appropriate to mention that the
replacement is not suitable for the
task? So since you're replacing the MB, why
not take the opportunity to
use something suitable.

Because it means introducing a whole slew of
new, unknown variables. :)

When I first installed 4.10R, it did not even
support the 8237; and disk
performance on that board was limited to a
terribly slow Multi-World DMA 2
mode (I think it was that; very slow, at
least). So, imagine my delight
when 4.11-STABLE supported the 8237 at last.
Buying a newer type
motherboard for 4.11-STABLE (where would you
find one for socket 754, so
soon replaced by socket 939, anyway?) would
likely mean an unsupported
south-bridge chip, and being back to square
one. Nope. I'm gonna stick
with what works for 4.11-STABLE (as that is
still my preferred FreeBSD
version; and if I cannot find a new motherboard
after the new one dies, I
will just continue to run the whole thing in a
Vmware box).

As for the LAN, since I only have a 100 Mb
network, I see no reason to
assume even a less than ideal performing
gigabit LAN would slow things
down (unless its performance dropped below 10%;
and I'm sure it's not that
bad).

In fact, not to be unnecessarily contrary, but
I would ere say this
motherboard is totally unsuited for desktop use
(I have a shiny P5WD2
Premium for that), and that this board is
rather ideally suited for a
FreeBSD 4.11 system.


Well that's just stupid, but you're entitled to
waste your money in any way you choose. We run
FreeBSD 4.9 and I've never had a problem with
hardware. Of course I know how to choose hardware
and you don't :)

I never said "desktop". The MB isn't really
suitable for anything that uses a LAN
extensively.

Knowing ASUS (whose MBs I'd never use, btw), I'd
guess that the ethernet controller on the P4WD2
is connected to a 1x PCIe which would be a joke.

What you don't "get":

- The slower the bus, the more CPU cycles it
takes to do an I/O. Typically you are doing 1000s
and 1000s of I/Os per second. Thats 100s of 1000s
of cpu cycles wasted per second.
- inefficient controller = more CPU cyles per
access. Maybe MANY more. This translates to
degradation of your CPU. The more traffic, the
more degradation. Whether you're on a gig network
or a 100Mb/s network, the efficiency of the
controller will still eat up your cpu. Of course
if you're just doing IM or email, then you don't
get enough iterations per second to make a
difference. But on a server,or gaming machine or
anything on a broadband connection, you're just
killing your cpu using a crappy controller.

You'd be better off putting up an old 845 chipset
MB with an fxp controller running a 2.6Ghz
celeron than what you're running, for a lot less
money.

DT

__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
_______________________________________________
freebsd-questions@xxxxxxxxxxx mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "freebsd-questions-unsubscribe@xxxxxxxxxxx"



Relevant Pages

  • Re: 2 PCs not starting up properly nor shutting down
    ... Permanent failures are created by manufacturing defects that may ... the motherboard could have two or more ... CPU would not execute BIOS. ... controller issues a command that says, ...
    (microsoft.public.windowsxp.hardware)
  • Re: Installing the Zalman CNPS9700 LED
    ... plastic fan controller to the motherboard is situated so I have to wrap the ... Should I take it off and turn it around so the CPU fan mobo controller is ... Install the CPU chip first, then the backplate, then the cooler? ...
    (microsoft.public.windowsxp.hardware)
  • nForce Pro 2200 losing ticks
    ... 00:04.0 Multimedia audio controller: nVidia Corporation CK804 AC'97 ... Aperture from northbridge cpu 0 too small ... PCI Interrupt Link enabled at IRQ 19 ... # ACPI Support ...
    (Linux-Kernel)
  • 2.6.0-test9-mm1: The hardware reports a non fatal, correctable incident occurred on CPU 0.
    ... 00:1f.5 Multimedia audio controller: Intel Corp. 82801DB AC'97 Audio ... IntelPentium4 CPU 3.06GHz ... mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe cid ... host bus clock speed is 138.0645 MHz. ...
    (Linux-Kernel)
  • Re: [ckrm-tech] Re: [Lse-tech] [PATCH] cpusets - big numa cpu and memory placement
    ... >>cpu bindings, the other uses the temporal dimension of cpu time. ... cpusets have a purpose beyond isolation and that is ... would partition a machine and CKRM would operate within each partition. ... And each resource controller knows best what it ...
    (Linux-Kernel)