hello,

i have a question about the following cpu:

SM40 SuperSPARC Module
SS10
501-2358
40MHz

http://au.sunsolve.sun.com/handbook_...PARC_SM40.html

states that, in contrast to the earlier versions of the sm40,

"the SS10 requires Boot PROM 2.12 if two SM40s are installed"

but also that

"The 501-2358 is not recommended for use in MP mode."

erm, this is puzzling...

does anyone have experience with that old module and can tell me what is true?

claus kick

On Friday 19 September 2003 10:52 am in comp.sys.sun.hardware Claus Kick
wrote:

> hello,
>
> i have a question about the following cpu:
>
> SM40 SuperSPARC Module
> SS10
> 501-2358
> 40MHz
>
> http://au.sunsolve.sun.com/handbook_...PARC_SM40.html
>
> states that, in contrast to the earlier versions of the sm40,
>
> "the SS10 requires Boot PROM 2.12 if two SM40s are installed"
>
> but also that
>
> "The 501-2358 is not recommended for use in MP mode."

Goto http://www.sunhelp.org/ and select the Hardware FAQ.
The SM40 came in several incarnations, only some of which
will ork in SMP properly.
This is one of those messy situations but the hardware FAQ has
most of the answers although it admits to being vague about
some cases.

Claus Kick wrote:
>
> hello,
>
> i have a question about the following cpu:
>
> SM40 SuperSPARC Module
> SS10
> 501-2358
> 40MHz
>
> http://au.sunsolve.sun.com/handbook_...PARC_SM40.html
>
> states that, in contrast to the earlier versions of the sm40,
>
> "the SS10 requires Boot PROM 2.12 if two SM40s are installed"
>
> but also that
>
> "The 501-2358 is not recommended for use in MP mode."
>
> erm, this is puzzling...
>
> does anyone have experience with that old module and can tell me what is true?
>
> claus kick

Perhaps (I'm guessing) it's a case like the Intel Celeron, where they
will work in SMP, but perform badly doing so, because they lack some of
the hardware that enables SMP to work well.

Several people have run some multi-threaded code of mine

http://atlc.sourceforge.net/

which runs a small floating point intensive benchmark at the end - the
program is not built a benchmark, but it runs one to determine effects
of compiler flags and SMP performance.

On a dual processor Celeron, it actually ran slower than the single
threaded version which would have used only one CPU. The Celeron is the
only processor where I have experienced this to date. If you look around
the web, you will find several things written about why the Celeron is
so poor in SMP mode. It may, just may be the same with the SM40's.

I don't know whether the SM40 has a cache, but it if does, that means
the caches need to be consistent. If effective hardware to do that is
not present, it could be the reason SMP performance works poorly, but
will work.

Only a guess, I don't know.

--
"The day Microsoft makes something that doesn't suck is probably
the day they start making vacuum cleaners." -Ernst Jan Plugge.

Dr. David Kirkby,
Senior Research Fellow,
Department of Medical Physics,
University College London,
11-20 Capper St, London, WC1E 6JA.
Website: http://www.medphys.ucl.ac.uk/~davek
Author of 'atlc' http://atlc.sourceforge.net/