Having just upgraded a couple of dedicated crunchers running Xubuntu from v7.10 to v8.04 (64b) can anyone explain this new "SCHEDULE" parameter to me. For now I'm assuming (and have changed the default from "1" to "0") this should not be used on a dedicated cruncher. I'm fairly new to Linux. Oh, the package now installs v5.10.45 of the core client.

Here's the start-up script config comments/entry

# Set this to 1 to enable advanced scheduling of the BOINC client and all its
# sub-processes (reduces the impact of BOINC on the system's performance).
SCHEDULE="0"

Then, at the end of the start function it does the 'schedule' function if turned on...

if [ "$SCHEDULE" = "1" ]; then
    schedule
  fi

Here is the actual startup script function

schedule()
{
  log_begin_msg "Setting up scheduling for $DESC and children:"
  if ! is_running; then
    log_progress_msg "$NAME not running"
  else
    if [ ! -x "`which ionice 2>/dev/null`" ]; then
      log_progress_msg "ionice not found,"
    else
      if ionice -c 3 -p $pid 2>/dev/null; then
        log_progress_msg "idle,"
      else
        log_progress_msg "ionice failed,"
      fi
    fi

    if [ ! -x "`which schedtool 2>/dev/null`" ]; then
      log_progress_msg "schedtool not found"
    else
      children=`ps --ppid $pid -o pid= | tr 'n' ' '`
      (schedtool -n 19 -D $pid $children >/dev/null 
        && log_progress_msg "idleprio") || 
      (schedtool -n 19 -B $pid $children >/dev/null 
        && log_progress_msg "batch") || 
      (schedtool -n 19 -N $pid $children >/dev/null 
        && log_progress_msg "normal")
    fi
  fi
  log_end_msg 0
}

---

- da shu @ HeliOS,
"A child's exposure to technology should never be predicated on an ability to afford it."

Did you ever get anywhere with this?

I'm in the process of converting from Windows to Ububtu v8.04 and use the BOINC that installed with it.

I find that even with SCHEDULE="0" in the defaults config, my Q6600 cores all run at 1.6GHz unless I manually change the "nice" values from 19 to 0 in "System Monitor" - it then runs at 2.39GHz

From a console, run the following two commands:
cd /sys/devices/system/cpu/cpu0/cpufreq
cat scaling_governor scaling_cur_freq scaling_max_freq scaling_min_freq

Then post its output here.

---

jfwall@SER2:~$ cd sys/devices/system/cpu/cpu0/cpufreq
jfwall@SER2:/sys/devices/system/cpu/cpu0/cpufreq$ cat scaling_governor
scaling_cur_freq scaling_max_freq scaling_min_freq
ondemand
2394000
2394000
1596000
jfwall@SER2:/sys/devices/system/cpu/cpu0/cpufreq$

OK, thanks for pointing me at those commands (I'm very much a Ubuntu NooB)

I see that Ubuntu is using the speed stepping technology to slow down the low priority tasks (anything below nice=0). So, for the time being, I have disabled it in the BIOS and things have speeded up.

I'm now getting 2.027 GFlops per core as opposed to ~1.3 GFlops with speedstep enabled. (Measured by BoincView monitor from a different machine)

I think I need to set "performance" in the Ubuntu scheduler rather than "ondemand" - there is some discussion on how to do that in the Ubuntu communities that I will pursue.

I'll also try to figure out why Ubuntu is delivering 2.027 GFlops per core when Windows is giving ~2.3 GFlops (That's 1.2 Gflops per machine more for Windows - nothing shows in the system monitor using that amount of CPU)

I run the BOINC benchmarks from the other machine (again using BoincView) when there is no user logged on and still get a max of about 2000 Dhrystones (again windows gives 2350 or so) (Incidentally, Whetstones are about 15% higher in Ubuntu than Windows)

Also the BOINC Client releases are slightly different - 5.10.22 on Windows and 5.10.45 on Ubuntu

For the next few days I'm going to let this machine run without speedstep and compare the overall thruput with another identical box running Windows.

One other thing - I think the script in the original posting tests if "schedtool" is running before using it. It is not installed by default in Ubuntu 8.4 so the script has no effect.

Once again, thanks for your help - any other suggestions would be most welcome.

I'll also try to figure out why Ubuntu is delivering 2.027 GFlops per core when Windows is giving ~2.3 GFlops (That's 1.2 Gflops per machine more for Windows - nothing shows in the system monitor using that amount of CPU)

Probably because your benchmarks differ between Windows and Linux. The GFlops number is calculated from benchmarks.

---

Kathryn :o)

OK, thanks.

Is the difference in the benchmark results due to the different versions of BOINC I'm using (5.10.22 on Windows and 5.10.45 on Ubuntu), or because different benchmarks run on different OS's?

Is the difference in the benchmark results due to the different versions of BOINC I'm using (5.10.22 on Windows and 5.10.45 on Ubuntu), or because different benchmarks run on different OS's?

Probably both...

---