Adaptec 2940UW/OF support dropped !!!!
Eric W Braeden
Adaptec tech support. Adaptec: "The 2940UW/OF
is supported under Solaris 2.51 and 2.6 only."
As a number of you know by now, the 2940 works
great under Solaris 7 as well, but only with the
32-bit kernel. Adaptec again: "At this time, there
are no plans for a 64-bit Open Firmware driver
for the 2940 product line."
The source code for the 32-bit drivers is not available
to other corporations so those of use who could
build the 64-bit drivers are unable to do so.
I have found the 2940 to be a fast and stable card for
use in the Ultra5s and am rather irritated that Adaptec
has decided not to suport 64-bit Solaris 7.
I suggest that all who care should hit www.adaptec.com
and make them aware that 64-bit Solaris 7 should be
supported. Getting through their on-line tech support is
not fun, but if enough of use hit on them.......
Eric
Sam C. Nicholson !!
>
> I suggest that all who care should hit www.adaptec.com
> and make them aware that 64-bit Solaris 7 should be
> supported. Getting through their on-line tech support is
> not fun, but if enough of use hit on them.......
>
> Eric
How 'bout voting with feet instead. I've had issues like
this with Adaptec for years. New chip/board/whatever from
Adaptec! Sure we'll support your app! (Until we build enough
channel depth in the M$ market, then we'll drop you like a
hot potato...)
Isn't there a dirt cheap SymbiosLogic based PCI card that just
drops in and works on the U5? I never got a chance to eval
them, but I can't believe that any of them could be more than
a point or two faster than the others.
And then there are the folks at Performance Tech. I'm surely
biased in their favor, but they do have good stuff, and their
support is just tops. But it does come at a price.
Begging/Pleading/Hammering in a competitive marketplace just
seems like a waste of precious time.
John Sullivan
Indeed, the Symbios cards are much less expensive, and we've
found them to be faster and more stable as well. In addition,
we can use the same cards with both SPARC (32- and 64-bit
versions) and x86 (Solaris 7 only). Most of the cards we use
are genuine Symbios cards (SYM8751SPE for single-channel and
SYM22801 for dual), but we have successfully used cards by
ASUS, Supermicro, DTC (Data Technology Corp), and Promise
as well; these seem to all be close to Symbios' reference
design. The Diamond Fireport 40 Dual also works, but the single
does not. I've also heard a lot of talk about Tekram's cards
not working well with Solaris 7, so be careful there too...
Eric W Braeden
Sam C. Nicholson !! wrote in message <36CE37...@webrelay.net>...
>Eric W Braeden wrote:
>>
>> I suggest that all who care should hit www.adaptec.com
>> and make them aware that 64-bit Solaris 7 should be
>> supported. Getting through their on-line tech support is
>> not fun, but if enough of use hit on them.......
>>
>> Eric
>
>
>How 'bout voting with feet instead. I've had issues like
>this with Adaptec for years. New chip/board/whatever from
>Adaptec! Sure we'll support your app! (Until we build enough
>channel depth in the M$ market, then we'll drop you like a
>hot potato...)
>
>Isn't there a dirt cheap SymbiosLogic based PCI card that just
>drops in and works on the U5? I never got a chance to eval
>them, but I can't believe that any of them could be more than
>a point or two faster than the others.
>
>And then there are the folks at Performance Tech. I'm surely
>biased in their favor, but they do have good stuff, and their
>support is just tops. But it does come at a price.
>
>Begging/Pleading/Hammering in a competitive marketplace just
>seems like a waste of precious time.
Sam, you are probably right on all points. Yes, I have heard
good things about the Symbios PCI card, but never tried one.
It is supposed to have native support. Know what model it is?
I just hate the way Adaptec seems to be shafting the people
that have already deployed the 2940.
Eric
Eric W Braeden
Toshio Kumagai
John Sullivan wrote:
>
<snip!>
>
> Indeed, the Symbios cards are much less expensive, and we've
> found them to be faster and more stable as well. In addition,
> we can use the same cards with both SPARC (32- and 64-bit
> versions) and x86 (Solaris 7 only). Most of the cards we use
> are genuine Symbios cards (SYM8751SPE for single-channel and
> SYM22801 for dual), but we have successfully used cards by
> ASUS, Supermicro, DTC (Data Technology Corp), and Promise
> as well; these seem to all be close to Symbios' reference
> design. The Diamond Fireport 40 Dual also works, but the single
> does not. I've also heard a lot of talk about Tekram's cards
> not working well with Solaris 7, so be careful there too...
Tekram's 390F (SYM53c875 chip rev. 3) is working well
under Solaris2.6/x86 and 7/x86 with Tekram's driver.
If you need driver for Tekram 390U/W/F, please send
e-mail to me with your OS version (2.6 or 7, FOR X86 ONLY).
390U2B (SYM53c895) also runs well with Sun's ncrs
driver on Sol7/x86.
On Sol2.6, installing 106699-03 patch is required to
run 390U2B/53c895 U2W HBA.
But note that latest BIOS (3.03) on those cards identify
every 53c810 as Tekram's 310.
So, bare 53c810 cards w/o BIOS (such like CI-2520) will not
work with Tekram's 390 series HBA :-(
###
PCI-device: pci1de1,3904@a, dc390uwf #1
Disk53: <Vendor 'IBM ' Product 'DDRS-34560W '>
cmdk53 at pci1de1,39041: target 0 lun 0
cmdk53 is /pci@0,0/pci1de1,3904@a/cmdk@0,0
Disk53: <Vendor 'IBM ' Product 'DDRS-34560W '>
root on /pci@0,0/pci1de1,3904@a/cmdk@0,0:a fstype ufs
Regards,
--
Toshio Kumagai (Toshio_...@Kumasan.ORG), Japan
Achim Gratz
> As a number of you know by now, the 2940 works great under Solaris 7
> as well, but only with the 32-bit kernel. Adaptec again: "At this
> time, there are no plans for a 64-bit Open Firmware driver for the
> 2940 product line."
That just shows how full of shit they are. A "64-bit Open Firmware
driver" is complete nonsense, what they were probably trying to say is
that they will not provide a kernel driver for a Solaris kernel
running in 64bit mode. Get a Symbios adapter that is compatible with
the Solaris glm driver and forget that crap.
Achim Gratz.
--+<[ It's the small pleasures that make life so miserable. ]>+--
WWW: http://www.inf.tu-dresden.de/~ag7/{english/}
E-Mail: gr...@ite.inf.tu-dresden.de
Phone: +49 351 463 - 8325
Chris
>
>
> That just shows how full of shit they are. A "64-bit Open Firmware
> driver" is complete nonsense, what they were probably trying to say is
> that they will not provide a kernel driver for a Solaris kernel
> running in 64bit mode. Get a Symbios adapter that is compatible with
> the Solaris glm driver and forget that crap.
>
One could argue that it's the os vendors job to provide the driver. If I
go out and buy Digital Unix, OS/2 Warp 4, Linux, even win 95, the
Adaptec driver comes with the Os.
Why don't you hassle Sun instead of Adaptec ?...
Rgrds, Chris
Joe Picone - The Terminal Man
> One could argue that it's the os vendors job to provide the driver. If I
> go out and buy Digital Unix, OS/2 Warp 4, Linux, even win 95, the
> Adaptec driver comes with the Os.
>
> Why don't you hassle Sun instead of Adaptec ?...
>
I am tired of hearing this argument that it is the vendor's responsibility.
We loved Solaris x86 when Sun had a decent selection of drivers - I
think it was a great product. Then I bought a laptop... then we tried
to get 24-bit video...
Why should hardware vendors support x86 when Sun's own third-party
software developers, who cater specifically to Unix, won't do it?
As far as I am concerned, this strategy of relying on the vendors is
essentially an admission at this point that Linux has won the war of
the Unix operating systems.
Sun's recent offering of "free" Solaris is a last gasp effort to
stay alive - but it is too little too late. I think Sun is backpedaling
on Solaris x86, and we must begin to look seriously at running
Linux on Suns (as well as PCs)...
-Joe
Mathew Kirsch
> > driver" is complete nonsense, what they were probably trying to say
is
> > that they will not provide a kernel driver for a Solaris kernel
> > running in 64bit mode. Get a Symbios adapter that is compatible
with
> > the Solaris glm driver and forget that crap.
If I
> go out and buy Digital Unix, OS/2 Warp 4, Linux, even win 95, the
> Adaptec driver comes with the Os.
> Why don't you hassle Sun instead of Adaptec ?...
Ah yes, the wonderful world of Adaptec. Seems that they're no longer
targeting Linux exclusively with their 2x4-up-the-key attitude.
There's one major flaw with hassling Sun for the driver. They have to
have programming information for the card in order to write the driver.
Adaptec probably provided the 2940 driver for 2.5.1 and 2.6, and hasn't
provided Sun with any driver programming data.
Back in the days when I used Linux regularly, Adaptec pulled this crap.
As the story I know goes, Adaptec originally let Linux programmers have
access to the information necessary to write drivers for the 2940 card.
Then, they suddenly changed their tune, which was okay because they
already had a 2940 driver, right? Wrong. Adaptec kept making changes to
their base 2940 card that were subtle, yet enough to break the Linux
drivers. Every time the Linux driver team would hack the 2940 driver to
get it working with the latest batch of cards, Adaptec would release
another revision, and loads of new users would be broken down again.
--
+-------------------------------+------------------------------------+
|Matt Kirsch, Programmer/analyst| THIS |
|Academic Computing Services | SPACE UNINTENTIONALLY LEFT BLANK |
|SUNY Brockport | |
|ma...@acs.brockport.edu | |
+-------------------------------+------------------------------------+
Sam C. Nicholson !!
>
> One could argue that it's the os vendors job to provide the driver. If I
> go out and buy Digital Unix, OS/2 Warp 4, Linux, even win 95, the
> Adaptec driver comes with the Os.
One could argue anything, I suppose. Problem is, that the adapter
vendor
is the best source of knowledge about their product. If they are just
cloning a reference design, then the OS vendor might be more in the
know.
The OS vendor in the cases that you mention above* is merely the
distributor/integrator. The adapter vendor supplies the OS vendor
with reference docs, sample code, and even fully builtup drivers
if they want to be associated with the OS.
* Well, except for the freeOS folks. They actually do write
drivers for their own OS. And that does prove you point that
Sun *could* do it.
To be fair, one must admit that there is a broad spectrum of
motivators. To make Linux popular, it is important for them
to support all adapters. Adaptec, otoh, is already popular.
They only have to help M$.
And I suppose, therein is the root of the problem. One might
be able to conceed your point in general terms, but too many
folk have specific experiences with Adaptec to cut them much
slack.
> Why don't you hassle Sun instead of Adaptec ?...
Sun should be hassled for leaving the SCSI ctlr off the U5/10s.
And in this NG, at least, they are being hassled. It was a
terrible omission in an otherwise very nice machine. Especially
when the Symbios PCI board costs less than $200.
Along with your point, Sun has a history of ego. Once they make a
decision, they defend it. Even if it's wrong. Here, they made a
mistake in configuration of a machine. They could have easily
addressed this by (for example) bundling in a SCSI controller in
the server edition. (I suppose they would rather you buy an E250.)
Their only answer is a grossly overpriced (for the PCI market)
SunSwift card.
-sam
Eric W Braeden
anywhere but on Adaptec cards. Yes, they are
stable under 2.5.1 and 2.6 and 7 (32-bit only)
but Adaptec seems intent on destroying their own
market for everthing but M$ crap.
They wont support 64-bit Solaris 7 and they wont
supply corporations the source code to build
their own. Damn...that pisses me off.
And while I'm steamed....damn Sun for not putting
a SCSI chip on the motherboard of the Ultra5/10.
Eric
wie...@nospam.purdue.edu
>Isn't there a dirt cheap SymbiosLogic based PCI card that just
>drops in and works on the U5? I never got a chance to eval
>them, but I can't believe that any of them could be more than
>a point or two faster than the others.
We've got a Symbios 8760 card in an Ultra 10. Works like a champ -- I
was even able to boot from a Plextor 32X CDROM drive attached to it.
Cost all of $115 from Insight Electronics (www.insight-electronics.com).
This is a narrow Ultra SCSI card, but we just use it with a SCSI scanner.
And SPARC Solaris ships with the required driver...
I don't know that there is a way to get Ultra-2 SCSI on an Ultra 5/10
without the Adaptec card, though. I have no idea if the Symbios card
that supports Ultra-2 SCSI would work (and work in Ultra-2 mode).
I agree that Sun should have included a SCSI controller on board, even
if it was just narrow SCSI.
--
Jeff
Jouko Holopainen
>
> Chris wrote:
> > Why don't you hassle Sun instead of Adaptec ?...
>
> Sun should be hassled for leaving the SCSI ctlr off the U5/10s.
> And in this NG, at least, they are being hassled. It was a
> terrible omission in an otherwise very nice machine. Especially
> when the Symbios PCI board costs less than $200.
Sun should be hassled to hire more device driver programmers.
As I see Solaris does not support USB, Firewire, DVD, CD-RW's,
audio cards, ... Sun do not support CD-ROM's with Joliet
(M$ CDROM file system. OTOH M$ does not support Rock Ridge).
Sun do not support (w.o. Ghostscript) non-PS printers.
Most of those should be about one man year job - depending on
existing "similar" drivers, of course (0.5x - 3x).
> Their only answer is a grossly overpriced (for the PCI market)
> SunSwift card.
They are learning... their memory (for U5/U10) is cheap.
Maybe they got that almost every U5/U10 owner wants SCSI ->
less price -> more sales...
--
Jouko Holopainen X-Net Oy http://www.xnet.fi/
PL 100 FIN-90501 Oulu FINLAND
Tel: +358 8 551 3578 Fax: +358 8 551 3565
jouko.ho...@xnet.fi
Achim Gratz
> Sun should be hassled for leaving the SCSI ctlr off the U5/10s.
> And in this NG, at least, they are being hassled. It was a
> terrible omission in an otherwise very nice machine. Especially
> when the Symbios PCI board costs less than $200.
>
> Along with your point, Sun has a history of ego. Once they make a
> decision, they defend it. Even if it's wrong. Here, they made a
> mistake in configuration of a machine. They could have easily
> addressed this by (for example) bundling in a SCSI controller in
> the server edition. (I suppose they would rather you buy an E250.)
> Their only answer is a grossly overpriced (for the PCI market)
> SunSwift card.
Huh? The SunSWIFT is UltraSCSI plus FastEthernet. There are other
UltraSCSI cards from Sun (single and dual channel controllers)
although I concede that they aren't cheap by any means.
As for the EIDE in the Ultra5_10, get over it. For years people have
bugged Sun to build a cheap machine. When Sun does it, they want one
that is more powerful (never mind that these options exist - they're
called UltraAXi and Ultra60). Compare to the discussion of the iMac
and you see a pattern emerge.
Achim Gratz
> One could argue that it's the os vendors job to provide the
> driver. If I go out and buy Digital Unix, OS/2 Warp 4, Linux, even
> win 95, the Adaptec driver comes with the Os.
The latest Adaptec drivers I've installed all came with the card
because they were newer than the OS.
> Why don't you hassle Sun instead of Adaptec ?...
Because I know that Adaptec has refused to provide Sun with the
necessary information to support their controllers properly. The
Linux folks haven't had any more luck from what I've heard, except
that they don't mind the occasional four weeks of non-support until
someone has hacked up a patch for whatever change Adaptec did to their
designs.
Phillip Fayers
>Chris <i...@dial.pipex.com> writes:
>> One could argue that it's the os vendors job to provide the driver. If I
>> go out and buy Digital Unix, OS/2 Warp 4, Linux, even win 95, the
>> Adaptec driver comes with the Os.
>> Why don't you hassle Sun instead of Adaptec ?...
>I am tired of hearing this argument that it is the vendor's responsibility.
>We loved Solaris x86 when Sun had a decent selection of drivers - I
>think it was a great product. Then I bought a laptop... then we tried
>to get 24-bit video...
I had a look on ATIs site a little while ago and followed a link to
UNIX drivers to find a company statement to the effect that they
do not make detailed information available to anyone to write
drivers. What are these PC companies turning into? Maybe its
too many years dealing with Microsoft.
>As far as I am concerned, this strategy of relying on the vendors is
>essentially an admission at this point that Linux has won the war of
>the Unix operating systems.
I hope not.
>Sun's recent offering of "free" Solaris is a last gasp effort to
>stay alive - but it is too little too late. I think Sun is backpedaling
>on Solaris x86, and we must begin to look seriously at running
>Linux on Suns (as well as PCs)...
I don't think they are back pedaling. Just not pedaling fast enough,
or encouraging enough people to help out with the pedaling. To compete
with Linux we only need a couple of things from Solaris.
i) A `community source' license to open it up a bit more.
ii) A Linux compatilbility system. First for linux binaries and
maybe linux specific source (but I wouldn't encourage that
especially if it is not in one of the UNIX standards). Then
a driver compatibility mode. People who wanted `esoteric'
drivers could run the Linux ones.
--
Phillip Fayers, SunAdmin/Support/Programming/Postmaster/Webmaster(TM)
Dept of Physics & Astronomy, University of Wales, College of Cardiff.
P.Fa...@astro.cf.ac.uk Attribute these comments to me, not UWCC.
Phillip Fayers
>"Sam C. Nicholson !!" wrote:
>> Chris wrote:
>> And in this NG, at least, they are being hassled. It was a
>> terrible omission in an otherwise very nice machine. Especially
>> when the Symbios PCI board costs less than $200.
>Sun should be hassled to hire more device driver programmers.
>As I see Solaris does not support USB, Firewire, DVD, CD-RW's,
>audio cards, ... Sun do not support CD-ROM's with Joliet
>(M$ CDROM file system. OTOH M$ does not support Rock Ridge).
>Sun do not support (w.o. Ghostscript) non-PS printers.
You can bet that USB, Firewire and DVD are high up on Suns list,
I wouldn't be at all suprised if the next low end Sun workstation
used USB and Firewire as standard components.
>Most of those should be about one man year job - depending on
>existing "similar" drivers, of course (0.5x - 3x).
>> Their only answer is a grossly overpriced (for the PCI market)
>> SunSwift card.
>They are learning... their memory (for U5/U10) is cheap.
>Maybe they got that almost every U5/U10 owner wants SCSI ->
>less price -> more sales...
Almost every U5/U10 owner? How many U5/10s have Sun sold do
you think? How many of those are being used as a desktop
machine by someone who doesn't give a fig about the individual
components in the machine.
I look after the Sun network in our department at the university,
we are one of those areas which is `price sensitive'. The IDE
disk saves something of the order of 400 pounds UK on our price
(my guess based on UK prices), about the same price as the monitor.
Even the `slow' IDE drives are plenty fast enough for storing
the local operating system and providing a bit more temporary
storage for the users.
Joseph P. DeCello III
>And while I'm steamed....damn Sun for not putting
>a SCSI chip on the motherboard of the Ultra5/10.
>
Uh? You should damn Sun for only putting 256k cache and 4500rpm drive
in there first Ultra5's, but why should everyone pay for SCSI when
they don't need it. I've got 50+ Ultra5's deployed (w/ 4500rpm disk)
and they are much faster than the sparc5/170's w/ scsi and cheaper.
Now they did finally get it right by upping the cache to 2mb and the
scsi disk to a decent 9gb (7200rpm). You can add scsi for $330 (3rd
party) and those who don't need more than one disk don't have to pay
for it. I got one of the new Ultra5/333 w/ 2Mb Lcache, 7200rpm ide,
128Mb ram, 100TPE, and I'm quite happy with it for the price. Yes
they are barely keeping up w/ PCs, but I need to run an OS that transarc
will support an AFS client on, and these things sure beat spending $10k
on an IBM rs/6000. BTW, I did get the 24bit working on the onboard
m64 on the new U5/333, took a patch and running m64config.
Now, the Ultra10 is another issue..... that should have SCSI, and they
should mount the hard drive right side up. The Ultra60 is too high
price for their entry level SCSI system, but the Ultra5 is really right
for IDE.
Whoh, this got off topic.... sorry.
/jpd
--
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Joseph P. DeCello III | Opinions expressed here are mine alone and
Systems Programmer | do not reflect those of any individual or
Michigan State University| department at Michigan State University.
Mathew Kirsch
> I am tired of hearing this argument that it is the vendor's
responsibility.
> We loved Solaris x86 when Sun had a decent selection of drivers - I
> think it was a great product. Then I bought a laptop... then we
tried
> to get 24-bit video...
> software developers, who cater specifically to Unix, won't do it?
Hardware vendors happily provide Windows 95 drivers for their Win95
hardware, as in the case of the original Adaptec 2940UW. Why should it
be Sun's responsibility to provide drivers for hardware specifically
designed to work in Sun workstations, as in the case of the Adaptec
2940UW/OF?
You also need the cooperation of the hardware vendor in order to write
support. If the vendor won't provide documentation, you can't legally
write a driver or provide support for a piece of hardware. How much
credibility would Sun have if most of Solaris's hardware drivers were
reverse-enginnered hacks?
Mathew Kirsch
> distributor/integrator. The adapter vendor supplies the OS vendor
> with reference docs, sample code, and even fully builtup drivers
> if they want to be associated with the OS.
> * Well, except for the freeOS folks. They actually do write
> drivers for their own OS. And that does prove you point that
> Sun *could* do it.
Re-read the first paragraph. Free OS folks need the reference docs,
sample code, or (preferably) built-up drivers just as much as the big
corporate OS folks. Linux wouldn't be where it is without that
information. The corporate cement-heads that think some penniless hacker
is going to take that information, steal their design, ramp up a
multi-billion-dollar facility and put them out of business need to make
a reality check.
> To be fair, one must admit that there is a broad spectrum of
> motivators. To make Linux popular, it is important for them
> to support all adapters. Adaptec, otoh, is already popular.
> They only have to help M$.
Adaptec is popular BECAUSE of M$. Because M$ is installed on 99.9999999%
of the PCs going out today, they only need to support it to make huge
profits (ever looked at the price of an Adaptec board?). The marginal
additional profit gains from supporting other OSes are not worth the
time and expense that would be spent on development.
Mathew A. Hennessy
Eric W Braeden <bra...@erinet.com> wrote:
>Sam, you are probably right on all points. Yes, I have heard
>good things about the Symbios PCI card, but never tried one.
>It is supposed to have native support. Know what model it is?
Dunno bout the Symbios-branded model, but the chipset is 53C875
(and IIRC 53C876). They're pretty common cards (cloners, Mylex) and
priced at least 35% less than comparable Adaptecs.. I've got a
non-booting cloner card doing 8mm and CD-R duty in my U10 and it works
swimmingly.. I can encode an MP3 track and burn an hsfs filesystem in CDE
at the same time without even thinking about it..
>I just hate the way Adaptec seems to be shafting the people
>that have already deployed the 2940.
Adaptec is really annoying, and I definitely prefer the savings
and reduced irritation of Symbios-based systems. Now only if they had a
supported Ultra2-LVD chipset! ;)
Good luck,
--
Mathew A. Hennessy (henn...@thoughtcrime.com)
'So the next time someone says, "I have a 50K file for you," your next
exclamation needs to be, "Wow, that's cold!"' - da...@falstofe.net
Achim Gratz
> i) A `community source' license to open it up a bit more.
The edu source licenses are about as close as it gets. I doubt
they'll go further than that, although I'd certainly welcome a less
restrictive licensing scheme.
> ii) A Linux compatilbility system. First for linux binaries and
> maybe linux specific source (but I wouldn't encourage that
> especially if it is not in one of the UNIX standards). Then
> a driver compatibility mode. People who wanted `esoteric'
> drivers could run the Linux ones.
Bleah. Linux is compatible with everything and nothing. A nightmare
to support the system and to port applications that have been
developed using this trait. I don't even think that the writing of
the `esoteric' drivers as such is much of a problem. But supporting
them without having definitive information from the manufacturer of
the device is dubious at best.
Achim Gratz
> Now, the Ultra10 is another issue..... that should have SCSI, and they
> should mount the hard drive right side up. The Ultra60 is too high
> price for their entry level SCSI system, but the Ultra5 is really right
> for IDE.
The Ultra10 has the same motherboard as the Ultra5, so if one had
SCSI, the other had it too. The drive is mounted "right side up",
more precisely it's having much less difficulty getting the heat off
the chips if the PCB is on top. If you look at the datasheet you'll
find that you can mount it just about every way you want, in contrast
to many drives from an earlier era which couldn't even stand being
mounted sideways.
Sam C. Nicholson !!
> > And in this NG, at least, they are being hassled. It was a
> > terrible omission in an otherwise very nice machine. Especially
> > when the Symbios PCI board costs less than $200.
> >
> > decision, they defend it. Even if it's wrong. Here, they made a
> > mistake in configuration of a machine. They could have easily
> > addressed this by (for example) bundling in a SCSI controller in
> > the server edition. (I suppose they would rather you buy an E250.)
> > SunSwift card.
>
> UltraSCSI cards from Sun (single and dual channel controllers)
> although I concede that they aren't cheap by any means.
>
Yep.
> As for the EIDE in the Ultra5_10, get over it. For years people have
> bugged Sun to build a cheap machine. When Sun does it, they want one
> that is more powerful (never mind that these options exist - they're
> called UltraAXi and Ultra60). Compare to the discussion of the iMac
> and you see a pattern emerge.
>
I am over it. I just placed an order for 10 of these critters.
But it does bug me that I can get a Dell P2-400 with USCSI for about
the same price. What's wrong with wanting cheaper, faster, better?
Incompatible desires drive progress. I'd gladly buy the Dell if
Solaris x86 were up to the task. Unfortunately, driver support
for the particular config wasn't there when I placed my order
Say, isn't that what got us here in the first place?
I'll pay more for Sun, as I feel I get more. But not too much more.
But I don't need a 270Mhz USparc CPU. What I *need* is a quality
piece of hardware, at a 3000-ish price that runs a well-supported
UNIX. I'd *like* a more flexible disk/backup subsystem. And I'd
prefer to get it from one source. If I have to integrate it myself,
I might as well multi-source the pieces, and run BSDI.
Hmm, I'm wondering, not as a conspiracy theory, but I'd wager that
the high prices of Sun's adapters inhibits development of drivers
for cheaper alternatives. Unlike M$, Sun has no incentive to push
cheaper hardware; quite the opposite, as they are still positioned
as a HW manufacturer.
-sam
Sam C. Nicholson !!
>
> "Sam C. Nicholson !!" wrote:
> > The OS vendor in the cases that you mention above* is merely the
> > distributor/integrator. The adapter vendor supplies the OS vendor
> > with reference docs, sample code, and even fully builtup drivers
> > if they want to be associated with the OS.
>
> > * Well, except for the freeOS folks. They actually do write
> > drivers for their own OS. And that does prove you point that
> > Sun *could* do it.
>
> Re-read the first paragraph. Free OS folks need the reference docs,
> sample code, or (preferably) built-up drivers just as much as the big
> corporate OS folks. Linux wouldn't be where it is without that
> information. The corporate cement-heads that think some penniless hacker
> is going to take that information, steal their design, ramp up a
> multi-billion-dollar facility and put them out of business need to make
> a reality check.
I agree with what you say, entirely. I said "*could*" in reference to
the
fact that Sun *could* write a driver if a Linux/*BSD/WhateverIX hacker
could.
I still think it very unlikely, as few would want to base a product
launch
on a shifting platform, as one would with an Adaptec ctlr in the box.
>
> > To be fair, one must admit that there is a broad spectrum of
> > motivators. To make Linux popular, it is important for them
> > to support all adapters. Adaptec, otoh, is already popular.
> > They only have to help M$.
>
> Adaptec is popular BECAUSE of M$. Because M$ is installed on 99.9999999%
> of the PCs going out today, they only need to support it to make huge
> profits (ever looked at the price of an Adaptec board?). The marginal
> additional profit gains from supporting other OSes are not worth the
> time and expense that would be spent on development.
Sounds like we're agreeing violently :)
-sam
Jeffrey Keith Boulier
Phillip Fayers <Phillip...@astro.cf.ac.uk> wrote:
[adaptec driver wars turning into Solaris vs Linux struggle (aargh!) ]
>I don't think they are back pedaling.[on Sol x86 -jkb] Just not pedaling
>fast enough,
>or encouraging enough people to help out with the pedaling. To compete
>with Linux we only need a couple of things from Solaris.
>
ZDNET reports in
http://www.zdnet.com/zdnn/stories/news/0,4586,2215357,00.html that
"It's not quite open-source software, but developers are about to get
access to Solaris without paying for it. Indeed, Sun Microsystems Inc.
plans to put all of its platform software-including Solaris-under a
Community Source License, the company said this week. The license is
likely to cover any Sun software that is not specific to end users or
vertical markets."
The article goes on to elaborate the licensing & utilization
restrictions that will be involved in the "community source" version of
Solaris. It's really very interesting.
Visions of recompiling everything using cc and -xarch=native dance
in my head, displacing the usual sugarplums.
There must have been a pretty impressive fight among
Sun/NCR/Toshiba/Fujitsu/Siemens-Nixdorff about this, weighing access to
source allowing others to mooch off of their intellectual property and
lessening support contract revenue vs. the potential to tap into "free"
labor from outside programmers.
Does anyone know when the official announcement will be? Could
ZDnet have messed this up? I haven't seen information anywhere else about
a Solaris community source license...
Yours Truly,
Jeff Boulier
Steve Kappel
jeff...@gwis2.circ.gwu.edu (Jeffrey Keith Boulier) writes:
> ZDNET reports in
> http://www.zdnet.com/zdnn/stories/news/0,4586,2215357,00.html that
> There must have been a pretty impressive fight among
> Sun/NCR/Toshiba/Fujitsu/Siemens-Nixdorff about this, weighing access to
> source allowing others to mooch off of their intellectual property and
> lessening support contract revenue vs. the potential to tap into "free"
> labor from outside programmers.
I also read somewhere that it may be awhile because there *IS*
intellectual property in Solaris which Sun doesn't own.
--
______________________________________________________________________
Steve Kappel steve....@iname.com http://www.visi.com/~skappel
John Riddoch
> I also read somewhere that it may be awhile because there *IS*
> intellectual property in Solaris which Sun doesn't own.
Apart from anything else, there's sendmail, bind and all the other
open-source stuff which has licencing restrictions which may or may not
be fully compatible with the community licencing. You might be able to
find more info at:
http://slashdot.org/articles/99/02/25/2352205.shtml
--
John Riddoch Email: j...@scms.rgu.ac.uk Telephone: (01224)262730
Room C4, School of Computer and Mathematical Science
Robert Gordon University, Aberdeen, AB25 1HG
I am Homer of Borg. Resistance is Fu... Ooooh! Donuts!
Bill Sanders
mounted in an external enclosure.
I've picked up their 4.5GB and 9GB disks and neither are identified by
Solaris 2.6 or Solaris7. I do not have this problem with Seagate drives.
Any help would be appreciated....
Bill
Bill Sanders
Jeffrey Keith Boulier
Steve Kappel <steve....@iname.com> wrote:
>> There must have been a pretty impressive fight among
>> Sun/NCR/Toshiba/Fujitsu/Siemens-Nixdorff about this, weighing access to
>> source allowing others to mooch off of their intellectual property and
>> lessening support contract revenue vs. the potential to tap into "free"
>> labor from outside programmers.
>
>intellectual property in Solaris which Sun doesn't own.
From where? I know that Solaris contains code from NCR, Siemens-Nixdorff,
Fujitsu, and Adobe. Does it also still have AT&T (now SCO) intellectual
property left over from the shift to SVR4?
Yours Truly,
Jeff Boulier
Alan Coopersmith
There's bits like this still in Solaris:
# Copyright (c) 1984, 1986, 1987, 1988, 1989 AT&T
# All Rights Reserved
# THIS IS UNPUBLISHED PROPRIETARY SOURCE CODE OF AT&T
# The copyright notice above does not evidence any
# actual or intended publication of such source code.
# Copyright (c) 1987, 1988 Microsoft Corporation
# All Rights Reserved
# This Module contains Proprietary Information of Microsoft
# Corporation and should be treated as Confidential.
--
________________________________________________________________________
Alan Coopersmith al...@godzilla.EECS.Berkeley.EDU
Univ. of California at Berkeley http://soar.Berkeley.EDU/~alanc/
aka: alanc@{CSUA,OCF,CS,BMRC,EECS,ucsee.eecs,cory.eecs}.Berkeley.EDU
palo...@my-dejanews.com
al...@godzilla.eecs.berkeley.edu (Alan Coopersmith) wrote:
> Jeffrey Keith Boulier <jeff...@gwis2.circ.gwu.edu> wrote:
> >Steve Kappel <steve....@iname.com> wrote:
> >>> There must have been a pretty impressive fight among
> >>> Sun/NCR/Toshiba/Fujitsu/Siemens-Nixdorff about this, weighing access to
> >>> source allowing others to mooch off of their intellectual property and
> >>> lessening support contract revenue vs. the potential to tap into "free"
> >>> labor from outside programmers.
> >>
> >>I also read somewhere that it may be awhile because there *IS*
> >>intellectual property in Solaris which Sun doesn't own.
> >
> >From where? I know that Solaris contains code from NCR, Siemens-Nixdorff,
> >Fujitsu, and Adobe. Does it also still have AT&T (now SCO) intellectual
> >property left over from the shift to SVR4?
>
> There's bits like this still in Solaris:
>
> # Copyright (c) 1984, 1986, 1987, 1988, 1989 AT&T
> # All Rights Reserved
> # THIS IS UNPUBLISHED PROPRIETARY SOURCE CODE OF AT&T
> # The copyright notice above does not evidence any
> # actual or intended publication of such source code.
>
> # Copyright (c) 1987, 1988 Microsoft Corporation
> # All Rights Reserved
> # This Module contains Proprietary Information of Microsoft
> # Corporation and should be treated as Confidential.
>
Yeah there is about a half a dozen files with MS in it. Never could
figure out the stupidity of proprietary #defines. Someday somebody
will sue somebody for defineing zero as 0 and execpt a lot of
money for it. I suppose you could just remove the MS stuff
and it would change the os. Problem is you can't tell who wrote
what parts in the include files.
---Bob
Bob Palowoda The Solaris x86 Corner http://fishbutt.fiver.net
-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
Raymond N Shwake
>In article <7buiq4$2dr$1...@agate.berkeley.edu>,
> al...@godzilla.eecs.berkeley.edu (Alan Coopersmith) wrote:
>> Jeffrey Keith Boulier <jeff...@gwis2.circ.gwu.edu> wrote:
>> >Steve Kappel <steve....@iname.com> wrote:
>> >>
>> >>I also read somewhere that it may be awhile because there *IS*
>> >>intellectual property in Solaris which Sun doesn't own.
>> >
>> >From where? I know that Solaris contains code from NCR, Siemens-Nixdorff,
>> >Fujitsu, and Adobe. Does it also still have AT&T (now SCO) intellectual
>> >property left over from the shift to SVR4?
Some time after SCO took over Novell's (formerly USL's) UNIX intellectual
property from Novell, Sun paid upfront money to SCO for their SVr4 license.
I think it's safe to assume Sun would be prohibited from publicly giving
that stuff away.
>> There's bits like this still in Solaris:
>>
>> # Copyright (c) 1984, 1986, 1987, 1988, 1989 AT&T
>> # All Rights Reserved
>> # THIS IS UNPUBLISHED PROPRIETARY SOURCE CODE OF AT&T
>> # The copyright notice above does not evidence any
>> # actual or intended publication of such source code.
>>
>> # Copyright (c) 1987, 1988 Microsoft Corporation
>> # All Rights Reserved
>> # This Module contains Proprietary Information of Microsoft
>> # Corporation and should be treated as Confidential.
>>
>Yeah there is about a half a dozen files with MS in it. Never could
>figure out the stupidity of proprietary #defines. Someday somebody
>will sue somebody for defineing zero as 0 and execpt a lot of
>money for it. I suppose you could just remove the MS stuff
>and it would change the os. Problem is you can't tell who wrote
>what parts in the include files.
And last year SCO reached settlement with Microsoft permitting
SCO to dispense with Microsoft code, royalties, et.al. That's why the
latest UnixWare variant has lost Xenix and DOS functionality (like
doscp). Not only can SCO clean out the MS cruft (and charge less, if
they choose) but we're free from looking at Microsoft copyright notices
when we boot up.
Simon Lockhart
>a SCSI chip on the motherboard of the Ultra5/10.
And even worse (or better, depending on your point of view), including
the SCSI on the OEM Ultra 5 boards!
Simon
--
Simon Lockhart | Tel: 01737 836676
Senior R&D Engineer, Online | Fax: 01737 836665
BBC Research & Development | Email: Simon.L...@rd.bbc.co.uk
Kingswood Warren, Tadworth, Surrey. | URL: http://www.bbc.co.uk/rd/
George Dau
]Has anyone found the secret jumper settings for IBM SCSI (68pin) disk drives
]mounted in an external enclosure.
]I've picked up their 4.5GB and 9GB disks and neither are identified by
]Solaris 2.6 or Solaris7. I do not have this problem with Seagate drives.
More info please.
I doubt the presence of secret jumper settings. Special Microcode, yes, but
jumper settings?
In what way are they not identified? What does probe-scsi-all report? What about
format? Model numbers for the IBM drives?
George Herbert
>Eric W Braeden <bra...@erinet.com> wrote:
>>And while I'm steamed....damn Sun for not putting
>>a SCSI chip on the motherboard of the Ultra5/10.
>
>And even worse (or better, depending on your point of view), including
>the SCSI on the OEM Ultra 5 boards!
Inter-Sun politics. Before the Grand Reunification, SMCC started
the Darwin project (Ultra 5/10) but SME, the manufacturing and OEM
company, was interested in slightly upscale systems with the same
CPUs and went out and did the AXi board.
At the origional price, the $300 or so per board it saved SMCC to
do IDE on Darwin didn't make a whole lot of sense. Now, however,
it's nearly 15% of the system pricing, which is a bundle.
-george william herbert
gher...@crl.com
Sam C. Nicholson !!
>
> At the origional price, the $300 or so per board it saved SMCC to
> do IDE on Darwin didn't make a whole lot of sense. Now, however,
> it's nearly 15% of the system pricing, which is a bundle.
Did it really cost Sun $300 bucks to put a scsi controller on the MB?
Or was it a $300 savings to swap out the ctlr *and* the drive?
I remember when Apple went to IDE disks (the 4400, I believe). They
knocked about $500 off of a sub-5000 machine. But they left the SCSI
ctlr on the MB. As I recall, SCSI disks in 95-ish were at a premium,
as they are now. (I have seen them bump to within 2-3 % of IDE at
various
points over the last 10 years, and within a year be 20% more.)
Do you know if anyone at Sun even considered this?
I spent a week discussing this issue on email with a fellow from Sun
( I don't see him posting here, so I'll assume he prefers his privacy),
he made some excellent points about the cost issues. I'd agree that
this
year SCSI on-board *with* SCSI disks would bump the price a fair bit.
But would just having the ctlr mean that much at all? Heck, even Narrow
ctlrs would be nice, and the ESPs are well under 10 bucks in quantity.
BTW, I'm not arguing the transfer rate issue, it's the disconnect-
reconnect and tagged queuing features that I miss. Though the latter
does require newer ctlrs, I'll grant.
Regards,
-sam
Shu-Chen Liu
for software development and running applications Netscape, Star Office etc.
is absolutely great. The 7200 rpm IDE hard drive seems to give wonderful
performance but I am sure it would not be adequate if you were trying to use
the Ultra 5 for a server or something like that.
As for OEM boards, I recently ordered the Cycle upgrade motherboard for
Sparc 5's and Sparc 20's which come with any of the Ultra 5 or Ultra 10
processors but also have SCSI built in! This seemed like a very nice
product. A motherboard with a 270 MHz UltraSparc, 64 MB RAM and SCSI
controller costs $2000. Basically, I agree with an earlier post in that the
expensive part, right now, is the SCSI drive.
Simon Lockhart wrote:
> In article <7asv8e$63o$1...@news.erinet.com>,
> Eric W Braeden <bra...@erinet.com> wrote:
> >And while I'm steamed....damn Sun for not putting
> >a SCSI chip on the motherboard of the Ultra5/10.
>
> And even worse (or better, depending on your point of view), including
> the SCSI on the OEM Ultra 5 boards!
>
palo...@my-dejanews.com
rsh...@atww.org (Raymond N Shwake) wrote:
> palo...@fiver.net <palo...@my-dejanews.com> writes:
> Someday somebody
> >will sue somebody for defineing zero as 0 and execpt a lot of
> >money for it. I suppose you could just remove the MS stuff
> >and it would change the os. Problem is you can't tell who wrote
> >what parts in the include files.
>
> And last year SCO reached settlement with Microsoft permitting
> SCO to dispense with Microsoft code, royalties, et.al. That's why the
> latest UnixWare variant has lost Xenix and DOS functionality (like
> doscp). Not only can SCO clean out the MS cruft (and charge less, if
> they choose) but we're free from looking at Microsoft copyright notices
> when we boot up.
>
Well umm ok but when they get into the board room meetings MS might
have something to say considering they have a considerable investment
in SCO. Nothing against SCO in in this case but the money influence
may play a factor.
Achim Gratz
> But would just having the ctlr mean that much at all? Heck, even
> Narrow ctlrs would be nice, and the ESPs are well under 10 bucks in
> quantity.
Aside from the controller you'd need to provide a motherboard
connector (better two if you want internal and external at the same
time) and a way to get out of the box (another connector with a
faceplate and cabling). All this needs to be tested and supported.
If the faceplate is optional, you have to provide an upgrade path (is
it customer installable? FRU? something else?). So it runs you more
than the ten bucks you quote on each box you make and most of the
customers will never use nor pay for it. Those who are willing to pay
have plenty options of getting a Symbios controller (from Sun or third
party).
George Herbert
>George Herbert wrote:
>> At the origional price, the $300 or so per board it saved SMCC to
>> do IDE on Darwin didn't make a whole lot of sense. Now, however,
>> it's nearly 15% of the system pricing, which is a bundle.
>
>Did it really cost Sun $300 bucks to put a scsi controller on the MB?
>Or was it a $300 savings to swap out the ctlr *and* the drive?
Combined (about 75% of that is in the drive, actually...)
>I remember when Apple went to IDE disks (the 4400, I believe). They
>knocked about $500 off of a sub-5000 machine. But they left the SCSI
>ctlr on the MB. As I recall, SCSI disks in 95-ish were at a premium,
>as they are now. (I have seen them bump to within 2-3 % of IDE at
>various
>points over the last 10 years, and within a year be 20% more.)
>
>Do you know if anyone at Sun even considered this?
I don't know. It makes some sense, yes, but I don't know all the
details (the Darwin hardware designs were fixed in stone before
I heard the design details).
>I spent a week discussing this issue on email with a fellow from Sun
>( I don't see him posting here, so I'll assume he prefers his privacy),
>he made some excellent points about the cost issues. I'd agree that
>this year SCSI on-board *with* SCSI disks would bump the price a fair bit.
>But would just having the ctlr mean that much at all? Heck, even Narrow
>ctlrs would be nice, and the ESPs are well under 10 bucks in quantity.
Once you include fab prices and board space, it's a bit more...
but not much. It would probably add under $50 to the system price.
There's no real reason not to use a fastwide or ultra controller...
the silicon they sit on is pretty cheap these days, the risers for
the cables are only a couple of bucks, etc.
-george william herbert
gher...@crl.com
Dirk Hufnagel
>
> I think I found it. First, Adaptec referred me to artecon. After looking for
> some fast Ultra2 PCI SCSI cards, I found ITI-6000 below....80 MB/s per channel
> (up to 160Mb/s)..twice as fast as the adaptec. see below--supports solaris
> 2.5/7
>
> http://www.intraserver.com/
>
> Its new and haven't tried it...it appears to be about 300 bucks....
This card seems to be Symbios based and should be very
similar to other Symbios Ultra2 host adapter.
What surprises me is that there are no Solaris driver for this card
on the Intraserver support website. They only say:
"Support drivers for Solaris are included on the Solaris
2.5.1 (1/97 or later) CD. No additional drivers are needed.
OpenBoot (obp) version 3.1.3 or later is also required."
Does that mean that SUN supports the 895 Symbios Ultra2 SCSI
chipset? I always thought that only the 875/876 chipset
is supported.
If the 895 chipset is really supported, the cheapest UltraSCSI
adapter would be the original Symbios card, either a single
channel SYM8951 for 180$ or a dual-channel SYM22801 for 330$.
dirk
Rene Martinez
SCSI adapters for SPARC systems.
The ITI-6000U2 Series are Ultra2/LVD SCSI adapters for SPARC systems.
These products are shipping now. IntraServer ships new state of the art
drivers with the products.
Single channel and dual channel versions are available.
Regards,
Rene
Reno Marioni <rmar...@earthlink.net> wrote in article
<36EC38F5...@earthlink.net>...
>
> I think I found it. First, Adaptec referred me to artecon. After
looking for
> some fast Ultra2 PCI SCSI cards, I found ITI-6000 below....80 MB/s per
channel
> (up to 160Mb/s)..twice as fast as the adaptec. see below--supports
solaris
> 2.5/7
>
> http://www.intraserver.com/
>
> Its new and haven't tried it...it appears to be about 300 bucks....
>
> --
> Cheers,
> Reno
>
> _______________________________
> Reno Marioni
> CEO, Founder
> Globalistic.com
> The Digital Adventure Media Company
> (415) 665.4220
> cell: (415) 902.RENO
> http://www.globalistic.com
> _______________________________
>
>
Andre Lue-Fook-Sang
equivalent on the IBM drive.
TP power from bus on
terminator enable on
parity off
and then give them unique SCSI id's 0,1,..
(id 7 usually used by controller)
hope this helps
**** Posted from RemarQ - http://www.remarq.com - Discussions Start Here (tm) ****