At 01:08 PM 2/15/01, Stan Sieler wrote:

>rally the users to complain about the software crippled CPU of the
>A-Class HP e3000.  (The 440 MHz CPU is software reduced to 110 CPU)


HP have for many years produced roadmaps showing that they aim to increase
processor performance by about 30% annually.

The 917 was introduced in 1992, running at 32 Mhz.  Compounding 30% growth
a year since then should now give us an entry level system over 1200% faster,
equivalent to a 917 running around 400 Mhz.  Guess what: the entry-level
HP9000 A400 runs at 440 Mhz.

The new entry-level system, the A400-100-110, actually delivers only 65%
improvement over the 917, about as fast as a 917 running at 50 Mhz.

The 8500 inside the A400 is a third generation PA-RISC 2.0 CPU, vastly more
powerful than the (PA7000 ?) CPU inside the 917.  For example, the 8500
contains 1.5 MB cache, while the 917 came with a 64KB cache.

Given the 8500 CPU, I'd bet that HP could achieve a 65% speedup by running
an 8500 at 40 Mhz.  Of course, they'd have a hard time finding the
parts:  8500
CPUs are manufactured to run between 360 and 440 Mhz, and HP has (to my
knowledge) never put the 8500 inside anything rated at less than 300 Mhz.

I'm curious how they chose 110 Mhz as the 'rated' speed of the A400.   I
wonder
how much work it took to slow down an 8500 to the level of an e3000 A400.

I know I'm being somewhat unfair: getting maximum performance out of an
8500 would require recompiling MPE/iX and applications with PA-RISC 2.0
compiler optimizations.  Maybe HP could have gotten around to this too, by now.

Paul Taffel