"Casper H.S. Dik" wrote:

> Martin Bochnig <mbeinsx@cs.tu-berlin.de> writes:
>
> >But what about MaxCpu == 400MHz (U5) vs. 440MHz (U10).
>
> The internal handbook describes the differences as "tested speeds";
> the 440MHz modules were never tested in the U5.

Funny. (no comment)

>
>
> There are more differences between the U5/U10: different power supplies,
> different fan assemblies.

Ok, point for you.

>
>
> >And what about MaxRAM == 512MB (SunPCiII) vs. 1024MB (SunPCiIIpro)?
>
> They're different boards, there are many cases of PC boards where
> some systems support more RAM in certain slots than others; may
> depend on the number of traces in the motherboard.

I think they aren't different boards (no prove here, however: They do have the
same internal code name called "Chimera").
They seem to have the same PCB and the same chipsets installed.
Just another preinstalled cpu (Celeron2 733MHz vs. 600MHz) and 128MB memory
preinstalled instead of 64MB. Both of them also take any 100MHz FSB Coppermine
Pentium3's up to 1100MHz (with additional cooling required / 1100MHz part may
be unstable).

>
>
> >Why is it required to give a workstation a new name just when they added
> >(or rather removed) some screws?
>
> It's not required; ultimately it's up to our marketing department how
> boxes are labeled; surely that's there prerogative?

Of course, I agree.
But never forget that most of your dissatisfied customers won't call your
hotline in order to complain.
They just won't buy any of your products anymore.
That is what I'm afraid of :-(

>
>
> >Let's take the SB100 vs. SB150: Oh, great: They really managed to
> >"speedbump" the IIe cpu Ecache size from 256KB to 512KB and the internal
> >clock from 500MHz to 550/650MHz.
>
> So that's what they did between the SS10/SS20.
>
> >However - SB1000 vs. SB2000 is 100.000000000% the same (except cosmetics).
> >Furthermore any SB2000 will identify itself as "Sun Blade 1000" ('uname
> >-i').
>
> They shipped with different motherboard revisions, even though there is
> some overlap

"Some overlap", please don't make me angry again:

--->>

Hi,

in other words 501-6230 and 501-6560 did ship both in SB1000 *and* in
"SB2000" at the very same time!!!
Together 501-6230 and 501-6560 were the "current" SB1000/2000 boards from
December 2001 (release of 501-6230) until the appearance of the 501-6768
in August 2004.
The "SB2000" 's lifetime was March 2002 to July 2004. So what do you want?
Is there any difference between the SB1000 and the "SB2000" products other
than their color/logo ?
If so, what? The price

Well, there are also two older SB1000 boards: 501-4143 and 501-5938.
But that doesn't mean anything here (even if they weren't listed as SB2000
FRU, but they are).
The real problem I see is, that SUNW used IDENTICAL boards in SB1000 and so
called "SB2000" from December 2001 till Summer 2004 while marketing them as
two completely different products {to make things clear: I don't claim
501-6230 and 501-6560 are identical, but ([501-6230 in SB1000] ==
[501-6230 in "SB2000"]) and ([501-6560 in SB1000] == [501-6560 in
"SB2000"]) }
They even offered an "Upgrade Path" from SB1000 up to SB2000 (can you
actually upgrade something to the very same thing?).

One is losing customers that way.
Not to mention the brain-dead "pricing model".
Where did SUNW's execs get their MBA, if any?

> and they all mechanically fit.
>
> >Even the default OBP banner-name is 'SUNW,Sun-Blade-1000'.
>
> So? A different banner means that we would have needed to rev the
> OS.
>
> >Do you have a SF280R available? If so, perform the following simple
> >experiment: Just temporarily disconnect the RSC. Then power the box back on
> >and OBP will tell you "Blade 1000" instead of SF280R !
>
> Yes; that's what the OBP notes say; if it can't probe the RSC or the
> Alarm, it will power up as a Blade 1000.
>
> Note also that the 1200MHz CPU requires a later different OBP because
> the system may fail to detect certain overheating conditions.

True, didn't I allude that. Just as Andrew Tyson as well as other posters did !

Well, it's a matter of 5 minutes.
Moreover newer Solaris revs. are required for the faster processors to operate
flawlessly.

>
>
> >By means of RSC available or not the Blade 1000 boards are able to "know"
> >if they sit in a server or in a workstation.
> >However - they don't have any means to determine whether they sit in a
> >SB1000 A28 or in a SB2000 A29 chassis (which for themselvesare almost the
> >same).
>
> Quite.
>
> Ultimately it's the label on the system which determines what it's being
> sold at.
>
> Remember that we used to sell "SPARCserver 20"s and "SPARCstation 20"s
> and "Ultra 2" workstations and servers.
>
> Identical hardware, except that the server had no graphics and was *much*
> more expensive.

Indeed, that seems to be a general rule.
Just compare the former prices of Sun Fire 280R vs. basically identical
SB1000/"2000".
(Nowadays the prices of used/refurbished sf280r vs. sb1k still reflect that.)


Martin