Running BOINC 5.10.28, various projects. Recently, a CPU benchmark was run giving me 45229 integer operations (C2D 4300). I reran the bencmark, it gave me 22102 integer operations. Another rerun resulted in 1 integer operation, which remains after several reruns of the benchmark, including stop/start service and reboot. Floating point ops appear to be correct.

Looking in client_state.xml, I noticed the following line for p_iops :

<p_iops>1.#INF00</p_iops>

Which explains the value of 1 interger op. Seems the xml gets corrupted somewhere.

BOINC 5.10.28 on VISTA home premium, service-install. Everything seems to work fine except the benchmark scores.

Anyone else seen this? Any suggestions? Never seen this on earlier versions.

... <p_iops>1.#INF00</p_iops> ...

Usually this indicated a division by zero, the result it tried to print to the XML was infinite. That means that the XML file is OK, just the benchmark result was messy.

Running BOINC 5.10.28, various projects. Recently, a CPU benchmark was run giving me 45229 integer operations (C2D 4300). I reran the bencmark, it gave me 22102 integer operations. Another rerun resulted in 1 integer operation, which remains after several reruns of the benchmark, including stop/start service and reboot. Floating point ops appear to be correct.

45229 and 22102 both seem quite high. I have seen few computers with more than 10000. And now it says it has infinite computation power, I want your computer! :)

Is your system clock correct? If the clock auto-adjusts itself while the benchmark is running, it could cause problems (on older versions of BOINC there was a bug that could cause the benchmark to never finish!).

---

... I want your computer! :) ...

Wait until he has checked his electricity bill ;-)


(@Sloompie ... your nick sounds familiar. Could it be from S@NL SETI Classic Gauntlets?)

System clock is correct, synced automatically. No syncing took place during the benchmarks.

@ananas : never been involved in SETI gauntlets.