🚀 go-pugleaf

In article <71lmco$v0q$1@piglet.dstc.edu.au>,
jason andrade <jason@dstc.edu.au> wrote:
>cdewick@lios.apana.org.au (Craig Dewick) writes:
>
>>the SS-20 (and the SS-10 for that matter). I know that it goes like this:
>
>> SM-51 (50 MHz), SM-61 (60 MHz), SM-71 (75 MHz), ???, ???
>
>>What comes after the SM-71 in terms of speed, processing ability, etc.? Are
>>the HS11 (100 Mhz), HS21 (125 MHz) and the SM-151 (150 MHz) next along the
>>increasing speed/performance curve of SS-10 and SS-20 compatibility? Are
>>there any CPU modules which are faster than the 150 MHz SM-151's that will
>>work in SS-20's?

	Depending on what criteria you use, the ordering
	is approximately:

	SM51 SM61 HS11 SM71 HS21 SM-151  Ross180  Ross200  

	Ross had various permutations of HyperSPARC modules 
	that Sun did not resell, for example the 125 and 142MHz 
	HyperSPARC modules with 1MB cache. Depending on
	the application you are using, the ordering here
	would differ.  (for example, the larger caches in
	the SM71 would possibly allow it to be a better
	choice than the HS21 for Oracle or Sybase.) 

	The 200MHz HyperSPARC was not supported in the SS10.

>>Also, what are the differences between the SuperSparc and HyperSparc
>>processors apart from higher raw clock speeds?

	The differences are too numerous to iterate here.  The
	topic has been covered in various publications and
	a search of the 'net is probably your best bet.

>the SM81 will work, but its unsupported due to the heat output i believe.
>2 in one machine would definitely not be good, unless you have some way of
>doing better cooling inside the machine.  i am still debating whether to
>go for an SS20/712 or play it safe and stick with it being an SS20/612.

	I've used SS20/712 machines with Fast Ethernet and Fiber Channel
	cards, heat wasn't a problem.
	
>the hypersparc has less cache than the supersparc cpus (256K ? and then
>upto 512K for the 125/150Mhz hypersparcs?).

	See above, there were various permutations (at least 3 that I
	can remember) of HyperSPARCs with 1MB of cache.

>observations i have seen is that a 150Mhz * 2 Hypersparc is comparable to
>a 170Mhz Ultrasparc, but that at high loads and/or where cache is important
>even the supersparc shows better performance than hypersparc
	
	YMMV. :-)


	-- Larry

-- 
--
Larry Beaulieu			beaulieu@beaulieu.ne.mediaone.net				
	All flames gleefully ignored from here too.