Notebooks present some special problems. To achieve the compact size they have to make some design trade offs. Some notebooks can handle the prolonged 100% CPU usage that crunching demands while others cannot unless you tweak a few things.

First, here is some info for you to make some comparisons. I have a 1.2 MHz machine here and it takes 14 to 15 hours to crunch an FAAH work unit from WCG if no other tasks are running. I also have 2 machines with 2.4 MHz CPUs and they take about 6 hours to crunch an FAAH (Fight Aids at Home) WU. Again, that's if no other tasks are running. Your 1.6 MHz computer should be somewhere between, I would estimate 11 to 12 hours.

So if you set up the hibernation to allow BOINC to have the CPU all to itself for 3 hours every night after you go to bed and then go into hibernation you should be able to finish a WU easily within the 7 day deadline and conserve power too.

But there are other hurdles too, for example checkpoints and running hot. We can do something about those too but you'll have to update to the latest version of BOINC, version 5.6.3 available here. That's a test version but I've been running it here for 10 days on 7 different machines and it's stable for me. Lots of other people are running it too and report no problems. It has some features you could use to tweak your setup, features earlier versions of BOINC do not have, for example a throttle so you can do something about the overheating. It also waits for a checkpoint before switching to another project. I think you need those features for your situation. Do you understand the issue with checkpoints?

Try that and let us know how you make out, ok? Get back to us with questions if you have any. I'm sure it will work if we tweak the settings properly.

Below is a link to an article on checkpoints and hibernating your computer with BOINC running. It's kind of a not-2-rough draft that needs a little more polish but it will tell you what you need to know about making the computer aautomatically hibernate at a preset time every day. It's not the Power Scheme method Windows offers you on the Screensaver page, that doesn't work when BOINC is running and I don't know why. Anyway davidhobbs posted this method a while ago, thanks Dave, and I took his method, made sure it works here, changed a few words to make it easier to read, added a suggestion or 2 from a few other posters and put it all togther in this article. so, that should take care of the auto-hibernating but if the instructions are not clear then ask.

The instructions in the file are for Win XP/Pro, it looks like it should work on ME too but I don't have a clue on 98.

Now the throttle to slow the CPU a bit so it runs cooler. First, you need to use it on a flat surface and have the little feet under it so it gets air flow. Got those little feet or did they break off like on my laptop? If you lay on your bed and use it your blankey bunches up around it and cuts off the airflow and the cooling. So f that's all in order and it still runs hot then try the throttle in BOINC 5.6.3. As you may have noticed the throttle is nowhere to be found in any of BOINC's setup screens or option pages. You have to copy 'n paste the following text into a file called global_prefs_override.xml in your BOINC folder. Use notepad to create the file, copy 'n paste, save the file. I'll explain how the file works below.

<global_preferences>
<run_on_batteries>0</run_on_batteries>
<run_if_user_active>0</run_if_user_active>
<start_hour>0</start_hour>
<start_hour>0</start_hour>
<net_start_hour>0</net_start_hour>
<net_start_hour>0</net_start_hour>
<leave_apps_in_memory>0</leave_apps_in_memory>
<confirm_before_connecting>0</confirm_before_connecting>
<hangup_if_dialed>0</hangup_if_dialed>
<work_buf_min_days>0.1</work_buf_min_days>
<max_cpus>2</max_cpus>
<cpu_scheduling_period_minutes>60</cpu_scheduling_period_minutes>
<disk_interval>60</disk_interval>
<disk_max_used_gb>100</disk_max_used_gb>
<disk_max_used_pct>50</disk_max_used_pct>
<disk_min_free_gb>0.1</disk_min_free_gb>
<vm_max_used_pct>75</vm_max_used_pct>
<idle_time_to_run>3</idle_time_to_run>
<max_bytes_sec_down>0</max_bytes_sec_down>
<max_bytes_sec_up>0</max_bytes_sec_up>
<cpu_usage_limit>100</cpu_usage_limit>
</global_preferences>

When BOINC starts it will get your preferences from the website and then it will read the global_prefs_override.xml. Any preferences you have in that file will override the preferences from the website. The trick is, 0 = no (false) and 1 = yes (true). Thus the first line tells BOINC to NOT run when the your laptop is on battery power. If you want BOINC to run on batteries you modify the first line to be

<run_on_batteries>1</run_on_batteries>

The throttle setting is at the bottom line. Other settings you might want to adjust, the disk usage is 100 GB so you may want to bump that down, 2 GB is probably way more than BOINC will ever need, 1 GB would work probably, anyway BOINC tells you how much disk it's using so you can always adjust again later. This one used to be important ...

<cpu_scheduling_period_minutes>60</cpu_scheduling_period_minutes>

.. but not so much anymore because BOINC 5.6.3 waits until it reaches a checkpoint before it switches to a different project WU. I think it should be 120 (2 hours) minimum even with this new BOINC but that's up to you. This next one is the WU cache, lots of people use higher than .1 but for your usage pattern I recommend .1.

<work_buf_min_days>0.1</work_buf_min_days>

On the hibernation... if you set it up like it says on the website then that's all you need to do. You don't have to do anything to BOINC to make the hibernating work. It will just kick in at the hour you specify.

It may be of interest here, that you in reality have 10 days to complete a Work unit in WCG.

I have noticed that when a WU is not returned within the 7 day requirement and a fourth unit is issued for validation, that the fourth unit is given 3 days to complete, but, if the outstanding (overdue) unit is recieved during this 4 day period, it is given credit. This effectively gives a 7+3=10 day deadline.

Cheers ozylynx.

Didactylos wrote:
> To improve efficiency (more science for less crunching) WCG reduced the initial replication to 3. This means that when one
> of the copies isn't returned quickly, it is a long time before an extra copy is sent out to complete the work unit. If the work
> units were allowed to run for a month, then people would have "pending validation" results hanging around forever.
>
> The WCG staff checked the statistics carefully, and determined that nearly everyone could return work units within
> several days. The deadline was set at 7 days as a compromise for those people who aren't connected to the net constantly.

If WCG wants more science for less crunching, it should set the initial replication to 2, and request a third crunch if and only if the first two results aren't in agreement. This would probably boost productivity by 45% or more.

The seven-day deadline is too short. People on dial-up might reasonably want to leave their machines crunching while they leave for a two-week vacation. WCG should consider offering a variety of deadline choices and letting users select choices other than the default choice. For example, the default choice could be seven days. Then you could have a second choice which has the effect of creating a crunching crew with a 14-day deadline. The deadline choice becomes operational for all the users selecting it when enough users from enough different places worldwide have selected that choice to make it into a viable alternative crunching network. All "extra copy" work from all deadline choices should go only into the shortest-deadline crunching crew(s).