[SunRescue] SS10 vs SS20

[James Lockwood]

On Mon, 17 Jan 2000, Earl Baugh wrote:

> Interesting.  Even with the heat issues?  We went thru 2 or 3 SS20 motherboards
> at the office (that had dual drives) before we got that little fan kit from Sun.
> Even with the kit, the SS20's seem noticably hotter inside even with a single
> drive.  A faster box sounds pretty good if the prices are close.  I'll have to
> ask Fred when I stop by.  Thanks for the suggestion...

The SS20 has slightly poorer cooling over the board but better cooling
over the drives.  The important thing is keeping the vents free from dust.
If you are running it in a hot area then I recommend removing the plastic
sidepanels completely (they just snap out).  Use external drives if
possible.  If you can keep the inside air temperature under 110F then you
should be ok.

> >Don't knock the SM50.  It's easily faster than a SS5/110 when installed in
> >a single configuration in a SS20.
> 
> Really?  I'd think the cache would offset the MBUS speed.  How would it
> compare to a SS10 with a SM51?  Faster?  Close to the same? (assuming 
> same framebuffer...the SS20 would have a faster drive with the SCA connector,
> correct?)

SCA doesn't make the drive any faster.  The SS10 and the SS20 both have
identical SCSI implementations, fast/narrow (10MB/sec) SCSI-II.

The SM50 in a uniprocessor config gives comparable performance to a
SS5/110 and a bit better than the SS10/51 in int work.  It's better than
the SS5 for fp and very slightly poorer than the SS10 (most fp benchmarks
benefit from the cache). SM50's don't scale nearly as well as modules with
e-cache in multi-CPU configurations as they saturate the mbus easily. 
Still, they provide a high bang for the buck, and I would recommend that
if you put a module with e-cache in a SS20 you go for a SM61 or higher
(the 51 will drop the mbus speed to 40MHz losing you the advantage of the
SS20). 

> As a side note here, I've heard a lot of discussion on which Sun CPU's are
> faster, and I've not found any "good" complete matrix of processors listing 
> performance across the Sparc machines (say from SS1 thru the SS20).  
> It would be a REALLY useful matrix.  I feel I got lucky and saved a lot of $$'s
> by asking the rescue folks before I bought a hypersparc...(the dual SM51 feels 
> snappier than the single Hypersparc 125 at the office...)  Anybody know of
> such a reference? (perhaps we could include it in the sun-ref)   A Ultra matrix
> would also be nice (though, I have a feeling that it's a little easier to compare 
> across that line...or am I mistaken?)

I've posted lists like this to sunhelp before, a dig through the archives
should be helpful.  Check www.spec.org for lots of benchmark tables, they
should give you a good starting point.

Here's a look at some older boxes:

System            CPU        ClkMHz  Cache      SPECint SPECfp  Info
Name              (NUMx)Type ext/in  Ext+I/D      92      92    Date
================= ========== ======= ========== ======= ======= =====
Sun SS/ELC        FJMB86903  33      64           18.2    17.9  Nov92
Sun SS/IPC        FJMB86902  25      64           13.8    11.1  Nov92
Sun SS/IPX        FJMB86903  40      64           21.8    21.5  Nov92
Sun SS2           RT601      40      64           21.8    22.8  Oct92
Sun SS2/PowerUp   WeitekPwUP 40/80   16/8         32.2    31.1  Jun93
Sun SS10/20       SuprSP     33      20/16        39.8    46.6  Nov92
Sun SS10/30       SuprSP     36      20/16        45.2    54.0  Apr93
Sun SS10/40       SuprSP     40      20/16        50.2    60.2  Apr93
Sun SS10/41       SuprSP     40/40.3 1M+20/16     53.2    67.8  Apr93
Sun SS10/51       SuprSP     40/50   1M+20/16     65.2    83.0  Apr93
Sun Classic,LX    MicroSP    50      4/2          26.4    21.0  Nov92
Sun Voyager       MicroSP2   60      16/8         43.2    36.2  Mar94
Sun SS4/70        MicroSP2   70      16/8         59.6    46.8  Jan95
Sun SS4/85        MicroSP2   85      16/8         65.3    53.1  May95
Sun SS5/70        MicroSP2   70      16/8         57.0    47.3  Mar94
Sun SS5/85        MicroSP2   85      16/8         65.3    53.1  May95
Sun SS5/110       MicroSP2   110     16/8         78.6    65.3  May95
Sun SS5/Turbo160  TurboSP    80/160  512+16/16   132     111    Nov96
Sun SS20/50       SuprSP     50      20/16        76.9    80.1  May95
Sun SS20/51       SuprSP     40/50   1M+20/16     81.8    89.0  May95
Sun SS20/61       SuprSP     50/60   1M+20/16     98.2   107.2  May95
Sun SS20/71       SuprSP2    50/75   1M+20/16    125.8   121.2  Jan95
Sun SS20/HS11     HyperSP    50/100  256+8/0     104.5   127.6  Nov94
Sun SS20/HS21     HyperSP    50/125  256+8/0     131.2   153.0  May95
Sun SS20/151      HyperSP    50/150  512+8/0     169.4   208.2  Nov95

-James