I made the same experience with the 166Mhz, 256kCache box, even the Alpha
I got for a test gets about 120,000 Dhrystones (Version 2.1, Language: C)
But it is too less at all.
One reason could be: L2 cache of 256 KB is too less, should be 1MB !
Running the whole Byte-bench 3.1 on Alpha166 with 32MB and comparing it
with the same on i486,66 Mhz, 16Mb showed:
Dhrystone : 1 : 2 (i486 : Alpha)
All Int Arithm: 1 : 1
Float Arith: 1 : 5
Double Arithm: 1 : 3
Nearly all other Benchmarks between 1:1 and 1:1.2
I then compiled XLispStat under OSF/1 and compared a complex computing
(nonlinear projection of highdimensional data = Sammon Mapping)
i486, 66 Mhz, Linux: evaluation: 1957 sec, garbage collection: 370 sec.
Alpha, 166 Mhz, OSF/1: evaluation: 2077 sec, garbage collection: 454 sec.
Note: i486 has only half the memory (16MB) of the Alpha box (32MB).
After this test I voted __NOT__ to by any Alpha 166Mhz, 256KB !!
BTW, I did the whole Byte-bench 3.1 on the same Alpha box, but under Linux
(was BLADE 0.1) and the results between Linux and OSF/1 are quite comparable.
So I do not expect any better result for my XLispStat test under Linux.
After this, I got an Alpha box (Sirius275), with 2MB L2 Cache, 275 Mhz
and 64 MB memory. And this was, when computing really got fast !
This box was in the XLispStat test factor 3 faster than the Alpha 166 box.
(it is significantly more expensive, too ;-))
So, Alpha processor indeed is fast, but don't expect it of the small boxes.
Before bying an 166 Alpha, I'd buy me a simple Pentium90, and have an
appropriate fast machine for its price !
bye
-- Bernhard
----------------------------------------------------------------------------
Bernhard Walter walter@pollux.edv.agrar.tu-muenchen.de
Techn. Univ. Munich ###### ##### +49-(0)8161-71-5055 (phone)
DVS - Weihenstephan # # # # # +49-(0)8161-71-4409 (fax)
85350 Freising # #### # # (Germany)
----------------------------------------------------------------------------