Dhrystones

Mikael Nykvist (viper@ludd.luth.se)
Fri, 18 Aug 1995 18:29:35 +0200 (MET DST)


Howdy ppl!

I got the dhrystones files from Jim Nance..

So here is some data:

Dhry: v2 (With the more used version v1.1 the results should be higher)

Comp: Noname AXPpci33, 21066A @233Mhz,256k 20ns,16MB 70ns
Soft: Linux/Alpha 1.3.18, gcc 2.7.0, Bob's libc-inc 0.35
Pass: 500000 passes
Unloaded system

Results:
-------

My own: 101351 (gcc -O5 -fomit-frame-pointer)
94339 (gcc -O3 -fomit-frame-pointer)

Jim's precompiled binaries: (Compiled under OSF1, but benched under Linux)

dhry.cc.migrate: 192307 (cc -migrate -O5 -ansi_alias -ifo -non_shared)
dhry.gcc.2.6.3 202702 (gcc -O3 -fomit-frame-pointer)

For comparison: (Pentium 90,16MB,256k cache, gcc 2.7.0 -O3 -fomit-fram..)

dhry: 85714.3

(Gateway 2k. Pentium 120Mhz,gcc 2.7.0 -O3 -fomit..)
dhry: 104166.7

So, it looks like gcc2.6.3 produces faster code under OSF1, than gcc2.7.0
produces under Linux/Alpha. And that a AXPboard is about ~2.4 times faster
than a normal P90, and about twice as fast as a P120,when gcc2.6.3 produces
the binaries under OSF1.. Ofcourse dhrystones aint the best bench in the
world...

I think dhrystones depends heavily on cache size/speed, so it would be
interesting to run the same tests on a AXPpci33 @233,64MB,1MB 15ns board
(You read this Bob? ;)

/Mikael

-- 
Mikael Nykvist, viper@ludd.luth.se
Running Linux/Alpha at 233Mhz!!!. http://www.ludd.luth.se/~viper/