Ok, I have two machines running BOINC Seti@Home, with the following results:

System 1:
Dell Lattitude D600 Laptop
1.4ghz Pentium M
512mb of Ram
WinXP Pro Service Pack 2

System 2:
Apple iMac G5, 1.6ghz
768mb of Ram
Mac OS X 10.3.8

Ok, my question is, is this normal? The Pentium M has a higher "measured floating point speed" while the G5 has a higher "measured integer speed" and which is more important to SETI@Home?

Pentium M:
Measured floating point speed 1203.09 million ops/sec
Measured integer speed 2501.92 million ops/sec

G5:
Measured floating point speed 830.11 million ops/sec
Measured integer speed 2777.51 million ops/sec

Secondly, anyone else with a 1.6ghz G5? What results do you see?

---

Blackhole forums - technology, politics, computers

[...]
> Ok, my question is, is this normal? The Pentium M has a higher "measured
> floating point speed" while the G5 has a higher "measured integer speed" and
> which is more important to SETI@Home?
[...]

It may have to do with the optimizations of each client's respective compiler. This will affect the credit you get for completing WUs. You'll get different results for the *same* processor (not just the same model, the same actual chip) on Windoze and Linux Boinc clients because of the differences between the compilers used on each OS.

---

There's lies...damned lies...and benchmarks.

Basically the compiler that was used for the G5 (The free GNU compiler for unix like systems) made different code choices than the Microsoft MSVC++ compiler for Windows, when compiling the benchmark part of BOINC.

How can you overcome this...look in other similar posts for the ".perry." optimized linux BOINC platform pages...They explain how to download and get a better benchmark score.

---