Linux-Hardware Digest #692, Volume #14 Fri, 27 Apr 01 01:13:09 EDT
Contents:
Re: Ultra160 (Piet)
Re: AMI MegaRAID 1600 + RH 7.1 not working ("Roberto Migliorati")
Webcam ("Peter")
Re: Ultra160 (Piet)
general noise (Piet)
Re: display problem (Dances With Crows)
Re: general noise (Hal Burgiss)
Re: HP OfficeJet G85 is not recognised by RedHat 7 (Dances With Crows)
Re: Switchboxes for keyboard, mice, video? (Jonadab the Unsightly One)
Re: Switchboxes for keyboard, mice, video? (Jonadab the Unsightly One)
Re: Switchboxes for keyboard, mice, video? (Jonadab the Unsightly One)
Re: general noise (Dances With Crows)
phoneline nic ("atlmike")
Stable sound card for SMP system ("Jason Gillis")
Re: Sharing Files. ("Doug")
Re: Stable sound card for SMP system (Rinaldi J. Montessi)
PCMCIA Flash Card Adaptor (Harmon Seaver)
Re: Ultra160 (hac)
Cannot Install RH7.1 - Partition Table Corrupt ! ("J-Pip")
Re: Ultra160 ("D. Stimits")
----------------------------------------------------------------------------
From: Piet <[EMAIL PROTECTED]>
Subject: Re: Ultra160
Reply-To: [EMAIL PROTECTED]
Date: Thu, 26 Apr 2001 23:50:32 GMT
Jerry Broszkowski wrote:
> Piet wrote:
>
>> I keep hearing that hard disks are the slowest components of a computer.
>> About two years back, I read a very small article in a magazine about
>> Ultra160. I believe they called it SCSI2 or something like that, it was
>> supposed to double as fast as a SCSI interface.
>> The article also claimed Ultra160 was compatible with Linux.
>> I'm wanting to hook myself up with a new desktop. Considering I have an
>> IDE interface now, if I can believe this article Ultra160 would be a
>> major performance boost.
>> Can anyone confirm this? Is there anyone here who has Ultra160 running
>> under Linux, and wouldn't mind sharing his experiences with me?
>>
>> Thanks in advance for any feedback
>
> I believe that until 15,000rpm drives become cheap, you're not going
> to get much performance gain from anything above ATA-66. Under SCSI2
> with a Buslogic BT-958 and a Seagate Barracuda (7200 rpm) I was getting
> around 10MB/s (measured with hdparm). With an HPT-366 (ATA-66) and an
> IBM Deskstar, I get close to 30MB/s (again measured with hdparm).
Hmmz. Your IDE is faster than your SCSI ??
> Of course, I could be completely out to lunch. :-(
I'm far from sure myself, but it kinda sounds like you are..
Doesn't matter, thanks for the thought!
Are their any users who have real-time experience with U160?
Thanks!
------------------------------
From: "Roberto Migliorati" <[EMAIL PROTECTED]>
Subject: Re: AMI MegaRAID 1600 + RH 7.1 not working
Date: Fri, 27 Apr 2001 11:12:49 +1000
With RH 7.0, were you using the stock standard kernel?? Some manufacturers
(Dell and others) include customised patches in their kernels to support
their hardware (RAID controllers in particular). If this is your case then
try and visit the server maker we-site and look for patches that will work
with your setup.
Regards,
Roberto.
"Oliver Olsen" <[EMAIL PROTECTED]> wrote in message
news:9c3erj$e1a$[EMAIL PROTECTED]...
> Hi!
> I'm trying to install RH 7.1 on an Intel server (dual P3-1ghz, 512mb RDIMM
> etc) with the AMI MegaRAID 1600-controller (493). RH 7.1 detects the
> controller, I'm able to select the packages wanted as well as specifying
> what partitions/sizes to use. It also says that the sda-partition has to
be
> initialized, and I have to press 5 times in order to continue. However,
when
> I'm done and have pressed the OK button in order to start installing I get
> the message:
> "Unable to format /. This may be a serious error".
>
> Anyone familiar with this? I've tried RH 7.0 (2.2.16-22enterprise kernel)
> which works great. However, as I'd like the 2.4 kernel in order to run
> Vmware G6-server. I then compiled the 2.4.3-ac9 kernel and tried booting
up.
> The kernel recogizes the controller and disk-array, but says:
>
> SCSI device sda: 286875648 512 byte hdwr sectors (146880 MB)
> sda: unknow partition table
>
> blah-blah-blah......
>
> VFS: Cannot open root device "801" or 08:01
> Please append a correct "root=" boot option
> Kernel panic: VFS: Unable to mount root fs on 08:01
>
> ___________________________-
>
> This looks like a 2.4 kernel problem and my controller. I've tried mailing
> AMI-support, but not much response from them yet.
>
> Anyone??
>
> Best regards,
> Oliver Olsen
>
>
------------------------------
From: "Peter" <[EMAIL PROTECTED]>
Subject: Webcam
Date: Thu, 26 Apr 2001 23:55:17 GMT
How do I use my webcam?
I found a driver for my logitech quickcam express usb, and it seems to
detect the camera fine, I've also compiled my kernel with Video4Linux,
and created the /dev/video[n] devices using MAKEDEV.
But then what?
I downloaded and compiled xawtv, but the "webcam" program therein says:
v4l: oops: got sigalarm
ioctl: VIDIOCSYNC(0): Interrupted system call
can't set capture format
What simple thing/kernel option/program did I miss?
All help appreciated
/Peter
------------------------------
From: Piet <[EMAIL PROTECTED]>
Subject: Re: Ultra160
Reply-To: [EMAIL PROTECTED]
Date: Thu, 26 Apr 2001 23:55:53 GMT
Steve Wolfe wrote:
>> I keep hearing that hard disks are the slowest components of a computer.
>> About two years back, I read a very small article in a magazine about
>> Ultra160. I believe they called it SCSI2 or something like that, it was
>> supposed to double as fast as a SCSI interface.
>> The article also claimed Ultra160 was compatible with Linux.
>> I'm wanting to hook myself up with a new desktop. Considering I have an
>> IDE interface now, if I can believe this article Ultra160 would be a
>> major performance boost.
>> Can anyone confirm this? Is there anyone here who has Ultra160 running
>> under Linux, and wouldn't mind sharing his experiences with me?
>
> Yes, you can find U160 SCSI controllers that work under Linux. However,
> it only provides a theoretical bus throughput of 160 megabytes/second.
> Realisitcally, you'll still be limitted to the speed of the drive in a
> single-drive situation, or the speed of the drives in a RAID array. Even
> using ultra-wide SCSI 2, offering 80 MB/second, you've still got more than
> enough bandwidth for a single drive.
>
> steve
I quite don't understand. I don't know much about hardware, but if the
ultra-wide SCSI 2 you mention provides (more than) enough bandwith, why is
there even a thing like U160 at all?
Surely I don't believe the websites promising me heaven on earth, but U160
can be assumed to provide some advantages over older hardware?
Thanks!
------------------------------
From: Piet <[EMAIL PROTECTED]>
Subject: general noise
Reply-To: [EMAIL PROTECTED]
Date: Fri, 27 Apr 2001 00:00:02 GMT
Hi again
I have a small other question.
I currently have a p-III @ 550Mhz, and this thing is really _incredibly_
noisy :(
I have to shut it down when I go to bed, where I would rather just leave it
on all the time.
Do recent machines still have this major disadvantage?
Thanks once again
------------------------------
From: [EMAIL PROTECTED] (Dances With Crows)
Subject: Re: display problem
Reply-To: [EMAIL PROTECTED]
Date: 27 Apr 2001 00:45:50 GMT
On Thu, 26 Apr 2001 15:30:08 -0000, khalid staggered into the Black Sun
and said:
>I have p111 800MHz with s3 savage4 vga card 32MB and phillips 15" monitor
>(model no 105E) 54KHz horizontal scanning,flicker free displays of
>800x600at up to 87 Hz.well i am manage to install linux 7.0 from cd but
>when it goes to x window(GUI)when i scroll down or up the screen ise that
>pixel starts breaking up so that i can read any thing after some time
>because alphabet overlap each other.
You've just found out one of many reasons why it's best to avoid S3
video cards. They Don't Work Well. You can probably fix this by
editing your XF86Config file and inserting the following line
Option "no_accel"
or
Option "sw_cursor"
within the block of text that starts with the line
Section "Device"
. Try the sw_cursor one first. Or, if you're not running XFree86
4.0.3, you can try that out... it probably has better support for that
particular card, and might even implement some of the acceleration
features correctly. HTH,
--
Matt G|There is no Darkness in Eternity/But only Light too dim for us to see
Brainbench MVP for Linux Admin / Workin' in a code mine, hittin' Ctrl-Alt
http://www.brainbench.com / Workin' in a code mine, whoops!
=============================/ I hit a seg fault....
------------------------------
From: [EMAIL PROTECTED] (Hal Burgiss)
Subject: Re: general noise
Reply-To: Hal Burgiss <[EMAIL PROTECTED]>
Date: 26 Apr 2001 21:26:28 -0400
On Fri, 27 Apr 2001 00:00:02 GMT, Piet <[EMAIL PROTECTED]> wrote:
>
>I currently have a p-III @ 550Mhz, and this thing is really
>_incredibly_ noisy :( I have to shut it down when I go to bed, where I
>would rather just leave it on all the time. Do recent machines still
>have this major disadvantage?
90% of noise is fans. You can buy quieter power supplies and other fans.
A little more money ...
--
Hal B
[EMAIL PROTECTED]
[EMAIL PROTECTED]
[EMAIL PROTECTED]
Spamtrap: [EMAIL PROTECTED] and [EMAIL PROTECTED]
--
------------------------------
From: [EMAIL PROTECTED] (Dances With Crows)
Subject: Re: HP OfficeJet G85 is not recognised by RedHat 7
Reply-To: [EMAIL PROTECTED]
Date: 27 Apr 2001 02:04:08 GMT
On Thu, 26 Apr 2001 14:30:22 +0100, Xiaoming Cai staggered into the
Black Sun and said:
>The HP OfficeJet G85 is not recognised by RedHat 7, although both
>parallel & USB connections are set up. Even simple command of
>'cat file > /dev/lp?', where ? was tried as 0, 1, and 2, gives
>the error of 'Device not configured'.
>Could anyone help me out? How to configure a port?
Was the *real* error message "/dev/lp0 : No such device"? If so, then
the problem is that the lp module (and maybe its supporting modules) is
not loaded. Typically, when you try to access a device, a kernel thread
sees this attempt, reads /lib/modules/$KERNELVERSION/modules.dep and
/etc/modules.conf , and figures out if a module needs to be loaded or if
the right one's already loaded or what.
Try doing "modprobe lp" and then cat'ing things to /dev/lp0. If that
works, put the following line in your /etc/modules.conf file and run
"depmod -a":
alias char-major-6 lp
which should get the right thing done. I thought this was on by default
in RedHat's distros.
If modprobing lp gives you funny errors, then do it the old-fashioned
way: insmod parport , insmod parport_pc , insmod lp . (NOTE: This
assumes you have an i386 or Alpha; Suns use something other than
parport_pc.) HTH,
--
Matt G|There is no Darkness in Eternity/But only Light too dim for us to see
Brainbench MVP for Linux Admin / Workin' in a code mine, hittin' Ctrl-Alt
http://www.brainbench.com / Workin' in a code mine, whoops!
=============================/ I hit a seg fault....
------------------------------
From: [EMAIL PROTECTED] (Jonadab the Unsightly One)
Crossposted-To: comp.sys.ibm.pc.hardware.chips
Subject: Re: Switchboxes for keyboard, mice, video?
Date: Fri, 27 Apr 2001 02:16:46 GMT
chrisv <[EMAIL PROTECTED]> wrote:
> Hell 16-bit looks as good as 24-bit. Extremely close, anyway.
For text, it's close. For images, it's orders of magnitude
better than 256, but I can see the difference.
- jonadab
------------------------------
From: [EMAIL PROTECTED] (Jonadab the Unsightly One)
Crossposted-To: comp.sys.ibm.pc.hardware.chips
Subject: Re: Switchboxes for keyboard, mice, video?
Date: Fri, 27 Apr 2001 02:16:47 GMT
Robert Redelmeier <[EMAIL PROTECTED]> wrote:
> I'm highly skeptical of quantifications of any continuous
> scale. I very much doubt the human eye can accurately
> discern "a few million colors."
Put large areas of any two of them next to eachother
and set the border to moving. Even with 24-bit colour,
choosing adjascent colours, you'll probably be able
to see the border.
Granted, this is a pathological case, and for most
applications (if not all) 24-bit color is adequate.
> > Maybe _you_ can't see it, but I sure can.
>
> And so can I! What I see is a contrast of two colors that
> are not-quite-the-same. On 16bit color, a smooth transition
> between two colors can produce a jaggy, digitized pixel line
> (isochrom?) when one color flops over to another. This
> can be highly disconcerting.
Jaggies are only problematic in certain kinds of applications
(mostly having to do with images). For text, blur is actually
more problematic. Which is why monitors are designed to be
sharper than television screens, because you have to read
lots of text on them. For _text_, a CGA monitor is actually
better than a typical TV screen. (For images, of course,
it doesn't come close, although modern monitors do.)
We could now have a debate about antialiasing, but that
would be off-topic.
- jonadab
------------------------------
From: [EMAIL PROTECTED] (Jonadab the Unsightly One)
Crossposted-To: comp.sys.ibm.pc.hardware.chips
Subject: Re: Switchboxes for keyboard, mice, video?
Date: Fri, 27 Apr 2001 02:16:49 GMT
[EMAIL PROTECTED] (Keith R. Williams) wrote:
> > Home, End, et al. are the fault of IBM, not of Microsoft. And to be
> > honest, they're much more intuitive than whatever Unix correspondents
> > there are.
>
> Sure. I'd prefer HOME and END closer to the cursor keys,
> but...
Huh? How could they be closer? They're *adjascent*! Or
by "cursor keys" do you mean something other than "the keys
that move the cursor"?
Unless you mean the useless batch of keys *between* the
main body of the keyboard and the keypad. Those are of
no value. Use the keypad itself. The others are
redundant and pointless. They keypad has everything
you need, cursor-movement-wise, except a ctrl key,
which would be nice to have, preferably where the
thumb can hit it while your other four fingers are
resting on left, center, right, and Enter. (Right
about where the stupid non-keypad right arrow is.)
Come to think of it, if you took out those stupid
redundant cursor keys and moved the keypad back
adjascent to the rest of the keyboard where it
belongs, Ctrl _is_ in that position. The XT
keyboard actually has the 101-key layout beat
in this respect.
- jonadab
------------------------------
From: [EMAIL PROTECTED] (Dances With Crows)
Subject: Re: general noise
Reply-To: [EMAIL PROTECTED]
Date: 27 Apr 2001 02:21:40 GMT
On Fri, 27 Apr 2001 00:00:02 GMT, Piet staggered into the Black Sun and
said:
>I have a small other question. I currently have a p-III @ 550Mhz, and
>this thing is really _incredibly_ noisy :( I have to shut it down when
>I go to bed, where I would rather just leave it on all the time. Do
>recent machines still have this major disadvantage?
The noisiest components in a PC are generally the hard disks and their
fans, the CPU fan, and the power supply fan (roughly in that order, I
think.) It is possible to spin down any IDE hard disks you are not
using with hdparm -Y, but note that you cannot really spin down
the disk that contains your / or /var partition without tweaking the
update daemon so that it doesn't flush buffers every 30 seconds. Modern
i386 CPUs cannot live without fans, but more expensive fan+heatsink
combos are often quieter. If it's the CPU fan that's driving you batty,
well, the PowerPC line don't need fans....
It might be a good idea to open up the case and make sure that nothing
is interfering with the moving parts (dust, stray cables, etcetera.)
--
Matt G|There is no Darkness in Eternity/But only Light too dim for us to see
Brainbench MVP for Linux Admin / Workin' in a code mine, hittin' Ctrl-Alt
http://www.brainbench.com / Workin' in a code mine, whoops!
=============================/ I hit a seg fault....
------------------------------
From: "atlmike" <[EMAIL PROTECTED]>
Subject: phoneline nic
Date: Thu, 26 Apr 2001 22:57:15 -0400
I am running Suse 7.1. I have an internal phone line network and a cable
modem. I am wondering if anyone knows of drivers for the phoneline nic
cards.
Thanks,
Mike Carlucci
------------------------------
From: "Jason Gillis" <[EMAIL PROTECTED]>
Subject: Stable sound card for SMP system
Date: Fri, 27 Apr 2001 03:48:03 GMT
Hi,
I'm looking for a _stable_ sound card for use in an SMP system (dual
PIIIs). This system will have a heavy playback load, so stability is a
must.
Any suggestions?
Thanks,
Jason
------------------------------
From: "Doug" <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.linux.networking,comp.os.linux.setup,linux.redhat.install,linux.redhat.misc
Subject: Re: Sharing Files.
Date: Fri, 27 Apr 2001 03:53:01 GMT
Or you could try a neat little program called Sharity, it works the opposite
of the way Samba does (access Win shares from Linux). You can download from
http://www.obdev.at/products/sharity/index.html
It's real easy to install and you don't have the config hassles of Samba.
<toor> <[EMAIL PROTECTED]> wrote in message
news:9ane05$8s0$[EMAIL PROTECTED]...
> I would like to share files with my Windows 98 and Redhat Linux 7.0
> computer. How would I do this?
>
> Thanks
>
>
>
------------------------------
From: [EMAIL PROTECTED] (Rinaldi J. Montessi)
Subject: Re: Stable sound card for SMP system
Reply-To: [EMAIL PROTECTED]
Date: Fri, 27 Apr 2001 04:45:06 GMT
Jason Gillis wrote:
>
> Hi,
>
> I'm looking for a _stable_ sound card for use in an SMP system (dual
> PIIIs). This system will have a heavy playback load, so stability is a
> must.
>
> Any suggestions?
>
> Thanks,
> Jason
I haven't found one that wasn't stable. Maybe I missed your point? I'm
running smp dual celerons (533's not overclocked) and have tried several
different cards. Changed for features not stability problems.
Currently using sb 64 awe
Rinaldi
--
We all know Linux is great...it does infinite loops in 5 seconds.
--Linus Torvalds
------------------------------
From: Harmon Seaver <[EMAIL PROTECTED]>
Subject: PCMCIA Flash Card Adaptor
Date: Fri, 27 Apr 2001 04:44:28 GMT
I've got 2.4.3 compiled to do PCMCIA IDE and geric SCSI so I can read my
Compact Flash cards for my camera, but when I put in the flash card in it's
PCMCIA adaptor, I can read it just fine (after mounting /dev/hdc1) but when I
take it out, the machine immediately locks up. I've tried doing a
/etc/pcmcia/ide stop before taking it out, and of course I unmount it too.
I noticed in the messages file that after it starts the ide module for hdc,
it then says:
Apr 25 19:04:19 cybershamanix modprobe: modprobe: Can't locate module nls_cp437
Apr 25 19:04:20 cybershamanix modprobe: modprobe: Can't locate module
nls_iso885 g-1
--
Harmon Seaver, MLIS
CyberShamanix
Work 920-203-9633 [EMAIL PROTECTED]
Home 920-233-5820 [EMAIL PROTECTED]
------------------------------
From: hac <[EMAIL PROTECTED]>
Subject: Re: Ultra160
Date: Fri, 27 Apr 2001 04:46:25 GMT
Piet wrote:
>
>
> Hmmz. Your IDE is faster than your SCSI ??
>
Sure. New IDE is faster than old SCSI.
The rate that a drive reads bits from the platter depends on the
density and the RPM. Both tend to increase with each generation of
drives. Newer drives are nearly always faster than older drives,
regardless of interface. The key figure of merit is the sustained
transfer rate, which you will only find on the data sheet for the
drive. They never advertise it. The speed of the interface is always
faster, and therefore not usually the bottleneck.
SCSI is better when you have multiple drives. This is always an issue
with servers, and rarely an issue with desktop systems. If your
motherboard comes with two or more ATA66 or ATA100 controllers, they
will work fine with one or two hard drives.
--
Howard Christeller Irvine, CA [EMAIL PROTECTED]
------------------------------
Date: Fri, 27 Apr 2001 12:49:16 +0800
From: "J-Pip" <[EMAIL PROTECTED]>
Subject: Cannot Install RH7.1 - Partition Table Corrupt !
Hello all,
Several attempts at installing RH7.1 failed because my partition
table
is corrupt. I am trying to "repair" it to avoid blowing the whole thing
away.
We've already tried out "dd if=/dev/zero of=/dev/hda count=1 bs=512" on
my friend's computer, which had a similar problem, and saw that it works
fine, but I wanna save it as a very last resort in my case.
Could anybody please help me with this output from "FindPart", there
is a huge "imaginary" partition past the end of the disk that I can't get
rid of
using fdisk from a linux rescue system. I have "EditPart" also. Any links to
sites that can help me better understand the data presented are welcome.
=====FAT CHS =Size Cl ==Root =Good =Rep. Maybe ==Bad YYMMDD DataMB
0 1 33 10000 4 2 10000 0 0 0
010303 1587
457 1 33 Second FAT not found.
638 1 33 6001 4 2001 6001 0 0 0
010302 1063
1465 0 34 Second FAT not found.
1531 1 33 13997 4 2 13997 0 0 0
010302 3151
2424 1 33 9580 16 52400 9580 0 0 0
010302 4226
Partitions according to partition tables on first harddisk:
-PCyl N ID -----Rel -----Num ---MB -Start CHS- --End CHS-- BS CHS
0 1*0B 63 10249407 5004 0 1 1 637 254 63
OK OK
0 2 0F 10249470 67906755 33157 638 0 1 4864*254 63
OK
638 1 0B 63 6152832 3004 638 1 1 1020 254 63 OK
OK
638 2 05 6152895 32130 15 1021 0 1 1022 254 63
OK
1021 1 83 63 32067 15 1021 1 1 1022 254 63
OK OK
1021 2 05 6185025 273105 133 1023 0 1 1039*254 63
OK
1023 1 82 63 273042 133 1023 1 1 1039*254 63
OK
1023 2 05 6458130 7887915 3851 1040* 0 1 1530*254 63
OK
1040 1 83 63 7887852 3851 1040* 1 1 1530*254 63 OK
OK
1040 2 05 14346045 14346045 7004 1531* 0 1 2423*254 63 OK
1531 1 0B 63 14345982 7004 1531* 1 1 2423*254 63 R0
OK
1531 2 05 28692090 39214665 19147 2424* 0 1 4864*254 63 OK
2424 1 0B 63 39214602 19147 2424* 1 1 4864*254 63 OK OK
2424 2 05 67906755 --- --- 4865* 0 1 1023 254 63
NB
Error reading sector 78156225.
Thanks,
J-Philippe.
------------------------------
Date: Thu, 26 Apr 2001 22:56:18 -0600
From: "D. Stimits" <[EMAIL PROTECTED]>
Reply-To: [EMAIL PROTECTED]
Subject: Re: Ultra160
Piet wrote:
>
> Hi
>
> I keep hearing that hard disks are the slowest components of a computer.
> About two years back, I read a very small article in a magazine about
> Ultra160. I believe they called it SCSI2 or something like that, it was
> supposed to double as fast as a SCSI interface.
> The article also claimed Ultra160 was compatible with Linux.
> I'm wanting to hook myself up with a new desktop. Considering I have an
> IDE interface now, if I can believe this article Ultra160 would be a major
> performance boost.
> Can anyone confirm this? Is there anyone here who has Ultra160 running
> under Linux, and wouldn't mind sharing his experiences with me?
>
> Thanks in advance for any feedback
There are a couple of things to consider. FYI, ultra 160 is considered
scsi 3 these days; scsi 2 would be the 80 MB/sec. There are actually a
lot of different terms for various scsi devices though.
One consideration is burst rate, another is continuous transfer. A more
mild consideration might be cpu useage. Easiest point first, cpu useage
doesn't really matter under scsi, the controller uses dma; for IDE, you
have to be sure the dma is enabled to reduce cpu use, or it'll fall
down.
Burst rate is what the 160 Mbyte/sec or other rate usually refers to. It
does matter, despite what people might tell you. Partly because several
disks have onboard ram cache, especially larger or higher rpm scsi
drives. Now imagine that you have more than one disk running at once,
then burst rates matter more, since it means holding the bus for shorter
time for small data. My current ultra 160 has a measured burst rate of
140+ MByte/sec on each disk. Whereas IDE will hang on to a bus the
entire time it is reading or writing, until finished, a scsi device can
let go of the bus and allow other devices to run on it while it is doing
something like fetching. It adds to the responsiveness.
In terms of sustained rates, it depends on the hard drive itself; a
combination of how fast it spins and how high density the storage is.
Larger drives have higher density and tend to have higher sustained
rates regardless of whether it is scsi or IDE. But there are more 10000
rpm scsi drives available than IDE, and only scsi runs at 15k rpm so
far. My 10k rpm ultra 160 drives run at a measured sustained throughput
of just over 34 Mbyte/sec. Because a scsi chain can run up to 160
Mbyte/sec, three at a time have no problem all sustaining that
throughput simultaneously. If you have only a single drive, it won't
matter nearly as much as if you want multiple drives. In the case of
IDE, you'd typically put two drives on two separate cables and enable
udma and be doing just about as well, but it can be somewhat troublesome
to try and get udma set up right...there is no bother on scsi.
A really big caveat here is that most PCI busses are not capable of the
160 Mbyte/sec bus speeds of a full ultra 160 controller. To actually
achieve that without saturating the bus itself you need a 64 bit pci
bus; the standard is 32 bit for most home machines. Within the limited
set of motherboards that have 64 bit pci, many of them use the i840
chipset, which is somewhat defective (IO-APIC problems), though the
serverworks chipsets are supposed to work quite well (i840 chipset is
Intel). Unfortunately, many serverworks chipsets don't have all that
great of a video bus performance, and several boards made with that
chipset don't even support AGP video (they're often assumed to be server
machines when you order 64 bit PCI and ultra 160). If you get the 32 bit
versions of an ultra 160 controller you will be limited in burst speed
to something probably closer to 100 Mbyte/sec, not so different from
udma 100 IDE, but you will still have the advantages of the disk being
able to detach from the bus while doing its own thing.
My overall feeling after running ultra 160 and older scsi, as well as
udma 33 and udma 100, is that it really kicks ass on IDE. The 10000 rpm
units I use seem, in practical day-to-day use, to far outperform udma
100, even though specs might not appear that different (keep in mind
though that the scsi is on a 64 bit bus and the udma 100 machine is only
a 32 bit pci bus). If you run a server, or gigabit ethernet, ultra 160
is the only reasonable way to go (as far as I'm concerned). If you can't
find something with a 64 bit pci bus for ultra 160, you might still
consider it, since the ultra 160 drives themselves tend to appear at a
higher performance in the market, prior to IDE versions coming out. The
price of an ultra 160 drive compared to an ultra 2 drive (80 Mbyte/sec)
isn't that much different, though the price of scsi is definitely higher
than IDE. You can get a 32 bit version of ultra 160 controller, and it
will be limited by that, but if you can afford more money for the
controller, it is probably still worthwhile.
D. Stimits, [EMAIL PROTECTED]
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: [EMAIL PROTECTED]
You can send mail to the entire list by posting to comp.os.linux.hardware.
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Hardware Digest
******************************
