I installed the client version 5.2.13 on two Xeon based machines on SuSE 9.3 and SuSE 10. On both machines, the benchmark result are way lower than expected, and of course, the computation takes a lot longer... :((
Aside from that, on suse 9.3 the processors are not corectly recognized as Xeons but as P4... Is there any workaround for this problems ? Thanks.

---

AFAIK it's a known "issue" with BOINC under Linux, but it only affects the speed BOINC *thinks* your computer has. Consequently it'll give you less credits/points than you'd get with the same PC under e.g. WinXP.

Still, the science applications themselves should run fine regardless at your PC's "true" speed.

You can run an "optimized" BOINC-Linux client to remedy this (see the links in the URL in my sig or go to Paul Buck's Wiki)

PS: Try at
# cat /proc/cpuinfo
and see what your Linux kernel knows about your CPU

---

How-To: Join Distributed Computing projects that benefit humanity

I see this on Solaris/SPARC benchmarks too on E@H. Even comparing just the Solaris/ SPARC machines to one another seems funny. I would expect a 500Mhz Sunblade 100 to be way behind a Sunfire 2500 with a 1.2 Ghz US IIIi in Floating Point. According to the benchmark, it about 1/2 as fast where according to spec.org, it is about 1/5. Either that GCC is really bad compared to the Windows /Intel one.

---

I installed the client version 5.2.13 on two Xeon based machines on SuSE 9.3 and SuSE 10. On both machines, the benchmark result are way lower than expected, and of course, the computation takes a lot longer... :((
Aside from that, on suse 9.3 the processors are not corectly recognized as Xeons but as P4... Is there any workaround for this problems ? Thanks.

You can get an optimized x86_64 client @ http://www.guntec.de/Crunch3r/boincx86.html

It's based on 5.3.6.

before you start using the app make sure you did a "ldd boinc_client" to see if there are any libs missing.


regards
Crunch3r

---