[rescue] sparc 2 and 10 things

Romain Dolbeau romain at dolbeau.org
Thu Dec 15 10:28:30 CST 2016


2016-12-15 17:09 GMT+01:00 Sandwich Maker <adh at an.bradford.ma.us>:
> i've read the sm81s do too in ss10/ss20 chassis.

<http://mbus.sunhelp.org/systems/sun/ss10.htm#h>
<http://mbus.sunhelp.org/systems/sun/ss20.htm#h>

2 SM71 in a SS20 are already too hot for me to dare run them in summer
in a system with a 18 GiB 7200 rpm drive.

2016-12-15 16:23 GMT+01:00 Doug McIntyre <merlyn at geeks.org>:
> Clock speed does tend to win all arguments. A HyperSPARC 150 was
> much faster than dual SM51s, especially if you had the dual.

HyperSPARC were clocked faster, but needed it since they were narrower
issue IIRC. They also usually had smaller caches compared to the 1 MiB
in cached SuperSPARC modules. There is some numbers here
<http://mbus.sunhelp.org/modules/index.htm#perf> ; I haven't run
benchmarks recently, but they seem to roughly tally up with my limited
experience.

Also, the SuperSPARC had an awesome double-pumped design in many
places that made it really, really good for its clock speed. But ended
up causing all sort of issues :-( Ultimately, the SuperSPARC II (SM71
and faster) replaced the double-pumped register file by a conventional
double-ported one, among other changes.BiCMOS wasn't Sun's best
decision either...

Wish I could find a Weitek PowerUP for my SS2 (or, rather, Axil HSW210
since it's a clone). It was still booting its SunOS 4.1.4 from the
three (third one taking over the floppy disk spot) internal disks last
summer... nigh indestructible those things are.

Cordially,

-- 
Romain Dolbeau


More information about the rescue mailing list