Hello all Has anyone compared SuSE linux performance on a mac architecture versus a PC(IBM clone)architecture? I'm just curious if there is any difference in speed if comparable machines were used.Assuming the two types can be compared.But we've all seen benchmark tests with each of their "native" OS's . I friend of mine gave me 6.4 for power PC and I've got and old 7600 laying around. So i think I'll give it a try.Any comments would we be extremely appreciated
At 11:12 PM 2/20/2002 -0700, you wrote:
I'm just curious if there is any difference in speed if comparable machines were used.
Now honestly -- think about this -- what is "comparable". Mhz of the processor? Give some thought to it. The speed of a machine is widely dependant on a great number of things.
Assuming the two types can be compared.But we've all seen benchmark tests with each of their "native" OS's .
Using the same OS is a very good idea when testing. However, you have to remember, for instance, that the SuSE-linux-ppc kernel+patch level+libs+mod_utils+other utils might not be the same version, even in the same "version" of SuSE (6.4). If, for example, the x86 version has a slightly newer kernel version, it might has a fix to the VM or scheduler that's not in the PPC kernel, it would give the x86 one an unfair advantage in the test. Also, you must not compare i686 compiled kernel+packages to a generic PPC kernel - i686 include support for SSE (Intel MMX, or AMD's SSE, I forget what it's called -- QuantiSpeed?), which is comparable to the G4's Altivec. Since you will probably never find a good way to compare Altivec to MMX, since there might not be comparable instructions in both. So it's better to compare a generic PPC (non-Altivec-enabled) package to a generic i386 (non-SSE enabled x86) one. OTH, SSEs are where a lot of modern processor power lies. Altivec is part of what makes a G4 "so fast" at a "slower" Mhz. Benchmarks between architectures is just a really hard thing to do, there's no "right" way to do it and very possible, non of the ways to do it are satisfactory. Realizing that even an "untuned" system is tuned to _something_ it might be best to tune each system to it's best performance and compare that (otherwise, by using default installation/settings you might be accidently favoring one or the other without knowing it.), but then, how do you ever know that you've done a "comparable" amount of tuning for each? It's endless....
I friend of mine gave me 6.4 for power PC and I've got and old 7600 laying around.
What is that, a G3?
So i think I'll give it a try.Any comments would we be extremely appreciated
I'd like to hear how it turns out. If you remember you might mail me off list - I don't watch the list too closely anymore. ---------------------------------------------------- Jonathan Wilson System Administrator Cedar Creek Software http://www.cedarcreeksoftware.com
participants (2)
-
JW
-
rev rob