SEO

<URL:http://www.theregister.co.uk/2004/07/26/sun_opteron_finally/>
| At long last, Sun's four processor Sun Fire V40z server and W1100z and
| W2100z workstations have gone on sale. All of the systems run on AMD's
| latest Opteron chip and complement Sun's existing two processor Sun
| Fire V20z server. With four Opteron systems now in its product line,
| Sun has officially become the biggest Tier I backer of AMD's x86-64bit
| product.

John
groenveld@acm.org

> <URL:http://www.theregister.co.uk/2004/07/26/sun_opteron_finally/>
> | At long last, Sun's four processor Sun Fire V40z server and W1100z and
> | W2100z workstations have gone on sale. All of the systems run on AMD's
> | latest Opteron chip and complement Sun's existing two processor Sun
> | Fire V20z server. With four Opteron systems now in its product line,
> | Sun has officially become the biggest Tier I backer of AMD's x86-64bit
> | product.

What does `isainfo -vk` return on those systems?
And how does the naming scheme ("V1100z" etc.) work?

groenvel@cse.psu.edu (John D Groenveld) writes:

> <URL:http://www.theregister.co.uk/2004/07/26/sun_opteron_finally/>
> | At long last, Sun's four processor Sun Fire V40z server and W1100z and
> | W2100z workstations have gone on sale. All of the systems run on AMD's
> | latest Opteron chip and complement Sun's existing two processor Sun
> | Fire V20z server. With four Opteron systems now in its product line,
> | Sun has officially become the biggest Tier I backer of AMD's x86-64bit
> | product.

64-bit Solaris x86 is due "soon". I find that hard to believe, but I
guess it depends what the meaning of "soon" is in this context.

I heard they only just got 64-bit Solaris booting on the platform. I'm
sure it's a major achievement, but I can't imagine a fully tested,
reliable, boxed, guaranteed 64-bit solaris for these machine will be
available "soon" according to my definition!

Chris
--
Chris Morgan
"Post posting of policy changes by the boss will result in
real rule revisions that are irreversible"

- anonymous correspondent

On Mon, 26 Jul 2004, Chris Morgan wrote:

> 64-bit Solaris x86 is due "soon". I find that hard to believe, but I
> guess it depends what the meaning of "soon" is in this context.

I guess it does! But don't forget that most of the hard "moving to
64-bit" problems have already been solved by Sun's Solaris kernel
engineers. With the exception of a few drivers, everything above the
"talks directly to the hardware" level is already 64-bit ready, because
of the single SPARC/x86 source tree.

> I heard they only just got 64-bit Solaris booting on the platform. I'm
> sure it's a major achievement, but I can't imagine a fully tested,
> reliable, boxed, guaranteed 64-bit solaris for these machine will be
> available "soon" according to my definition!

I imagine "soon" in this context is defined as "S10 FCS".

--
Rich Teer, SCNA, SCSA

President,
Rite Online Inc.

Voice: +1 (250) 979-1638
URL: http://www.rite-online.net

Rich Teer <rich.teer@rite-group.com> writes:

> On Mon, 26 Jul 2004, Chris Morgan wrote:
>
> > 64-bit Solaris x86 is due "soon". I find that hard to believe, but I
> > guess it depends what the meaning of "soon" is in this context.
>
> I guess it does! But don't forget that most of the hard "moving to
> 64-bit" problems have already been solved by Sun's Solaris kernel
> engineers. With the exception of a few drivers, everything above the
> "talks directly to the hardware" level is already 64-bit ready, because
> of the single SPARC/x86 source tree.

I don't think it's that simple. The programming model used for Solaris
x86 has always been 32-bit and now they are just adjusting it for
64-bit. I don't know how Sun handles the conditional compilations, but
I can well imagine lots of ifdefs that check to see if it's sparc or
intel and whether it's 32- or 64-bit. If anyone has made mistakes in
making their code "cross-platform" in this sense, those mistakes will
only come out now - for example if someone used the fact of it simply
being intel to choose 32-bit code. Also "Solaris" is much more than
the kernel, for example the 64-bit mode of Sun's compiler for x86-64
is by definition newer and less proven (just like the hardware,
although I'm sure Opteron really is a great product).

Chris

--
Chris Morgan
"Post posting of policy changes by the boss will result in
real rule revisions that are irreversible"

- anonymous correspondent

On Mon, 26 Jul 2004, Chris Morgan wrote:

> I don't think it's that simple. The programming model used for Solaris
> x86 has always been 32-bit and now they are just adjusting it for
> 64-bit. I don't know how Sun handles the conditional compilations, but
> I can well imagine lots of ifdefs that check to see if it's sparc or
> intel and whether it's 32- or 64-bit. If anyone has made mistakes in
> making their code "cross-platform" in this sense, those mistakes will
> only come out now - for example if someone used the fact of it simply
> being intel to choose 32-bit code. Also "Solaris" is much more than

Yes, that IS a good point. I guess time will tell... :-)

> although I'm sure Opteron really is a great product).

Yeah; I'd like a chance to play with^W^Wtest one of those V2100zs!

--
Rich Teer, SCNA, SCSA

President,
Rite Online Inc.

Voice: +1 (250) 979-1638
URL: http://www.rite-online.net

In article <86r7qyr810.fsf@elrond.bloomberg.com>,
Chris Morgan <cm@mihalis.net> writes:
> Rich Teer <rich.teer@rite-group.com> writes:
>
>> On Mon, 26 Jul 2004, Chris Morgan wrote:
>>
>> > 64-bit Solaris x86 is due "soon". I find that hard to believe, but I
>> > guess it depends what the meaning of "soon" is in this context.
>>
>> I guess it does! But don't forget that most of the hard "moving to
>> 64-bit" problems have already been solved by Sun's Solaris kernel
>> engineers. With the exception of a few drivers, everything above the
>> "talks directly to the hardware" level is already 64-bit ready, because
>> of the single SPARC/x86 source tree.
>
> I don't think it's that simple. The programming model used for Solaris
> x86 has always been 32-bit and now they are just adjusting it for
> 64-bit. I don't know how Sun handles the conditional compilations, but
> I can well imagine lots of ifdefs that check to see if it's sparc or
> intel and whether it's 32- or 64-bit. If anyone has made mistakes in
> making their code "cross-platform" in this sense, those mistakes will
> only come out now - for example if someone used the fact of it simply
> being intel to choose 32-bit code. Also "Solaris" is much more than
> the kernel, for example the 64-bit mode of Sun's compiler for x86-64
> is by definition newer and less proven (just like the hardware,
> although I'm sure Opteron really is a great product).


You may be forgetting the early Merced port they'd already done.

I doubt that 64-bitness of x86 source is going to be the problem as
much as the low-level stuff and driver mods.


--
mailto:rlhamil@smart.net http://www.smart.net/~rlhamil

Chris Morgan <cm@mihalis.net> writes in comp.sys.sun.hardware:
|I don't think it's that simple. The programming model used for Solaris
|x86 has always been 32-bit and now they are just adjusting it for
|64-bit. I don't know how Sun handles the conditional compilations, but
|I can well imagine lots of ifdefs that check to see if it's sparc or
|intel and whether it's 32- or 64-bit. If anyone has made mistakes inD
|making their code "cross-platform" in this sense, those mistakes will
|only come out now - for example if someone used the fact of it simply
|being intel to choose 32-bit code.

Most of Solaris is a common source base, using the platform-independent
_LP64 #ifdef to check for 64-bit mode. Some code admittedly mistakenly
checked for _sparcv9 instead, but we've been working on cleaning that up
in the months we've had to prepare for this in the user-land software
while waiting for the kernel and compiler to be ready for us to build
our bits on top.

--
__________________________________________________ ______________________
Alan Coopersmith * alanc@alum.calberkeley.org * Alan.Coopersmith@Sun.COM
http://www.csua.berkeley.edu/~alanc/ * http://blogs.sun.com/alanc/
Working for, but definitely not speaking for, Sun Microsystems, Inc.

It is nice to see Sun's uniprocessor SPEC results finally looking
competitive again. For example, for the W1100z Sun reports CINT2000
base = 1434, CINT2000 Rate base = 16.6. Both of thes numbers are
quite respectable compared to the competition (though they're not the
absolute top), and they blow away the best uniprocessor SPARC numbers
that Sun has published (642 and 7.44, respectively, for a 1.2 GHz Sun
Blade 2000). Too bad specbench.org isn't publishing the floating
point numbers yet.

I noticed a few other things:

Solaris x86 64-bit support is not available until Solaris 10 ("early
2005" -- could that mean as late as June 2005?).

<http://www.sun.com/desktop/products/graphics/> hasn't been updated to
reflect the new graphics options available from Sun. (It'd be amusing
to see the NVIDIA Quadro FX 3000 in a head-to-head comparison against
the Sun XVR-4000....)

You can shoehorn only 4 GB RAM into a W1110z now, as 2 GB DIMMs
aren't available until next month. The W2100z holds double that.

The W2100z uses Ultra320 SCSI. Is desktop FC-AL dying?

PS. I wish Sun wouldn't keep mentioning its "disruptive pricing
model" in its press releases. I don't wanna be disrupted.

Paul Eggert wrote:
> Solaris x86 64-bit support is not available until Solaris 10 ("early
> 2005" -- could that mean as late as June 2005?).

Gee, I sure hope not -- if the engineering team has got a multi-user
boot happening by now surely there isn't that much work left to do
in order to get the port fully operational?

> <http://www.sun.com/desktop/products/graphics/> hasn't been updated to
> reflect the new graphics options available from Sun. (It'd be amusing
> to see the NVIDIA Quadro FX 3000 in a head-to-head comparison against
> the Sun XVR-4000....)

um, only if the prices weren't mentioned?

> The W2100z uses Ultra320 SCSI. Is desktop FC-AL dying?

not at all, it's just that the interface costs are still pretty
hefty for FC devices, so for price-point competitiveness you
are unlikely to see FC on the desktop.

Not that I wouldn't mind FC on my desktop but we've gotta be
realistic about this.



mcbofh