Hi, Gabriel.

First, losing a small amount when you stop BOINC is normal and inevitable.

Second, you should not be losing many hours of work with a 2.3GHz processor. How long does it take you to complete a task normally?

That's not correct. BOINC doesn't save a checkpoint when you exit.

Reminded me of debsgr8 who wanted a beep for each checkpoint made so she'd know when to gracefully shut down the system right after one. On multi-cores that's a problem as than the question is which job ;?

Generally, e.g. on my laptop I just close the lid and it goes into standby (XP HM, C2D). I don't even bother to pause. It only runs HCC which has an incrementally increasing checkpointing, the last one on slow machines I'm sure exceeding an hour and more towards theend. Anyway, with Leave in Memory [LIM] on and WtD set at 5 minutes (300 seconds), it somehow manages to just resume without loosing a second and always validating. For the Doubting Thomases here's the log from last night and resume of this morning.

11/23/2008 10:55:20 PM|World Community Grid|[checkpoint_debug] result X0000055201265200508240309_0 checkpointed
11/23/2008 11:24:06 PM|World Community Grid|[checkpoint_debug] result X0000055201265200508240309_0 checkpointed
11/23/2008 11:24:34 PM|World Community Grid|[checkpoint_debug] result X0000055241316200508111419_0 checkpointed
11/24/2008 7:43:13 AM||Suspending network activity - time of day
11/24/2008 8:03:47 AM|World Community Grid|[checkpoint_debug] result X0000055201265200508240309_0 checkpointed
11/24/2008 8:17:59 AM|World Community Grid|[checkpoint_debug] result X0000055241316200508111419_0 checkpointed
11/24/2008 8:26:20 AM|World Community Grid|[checkpoint_debug] result X0000055201265200508240309_0 checkpointed
11/24/2008 8:47:33 AM|World Community Grid|[checkpoint_debug] result X0000055201265200508240309_0 checkpointed

Restarts would be logged, the last seen on the 22nd when I booted the system (do it about once per fortnight) Those would show as:

11/22/2008 7:05:39 PM|World Community Grid|Restarting task X0000055190451200509061953_1 using hcc1 version 606
11/22/2008 7:05:39 PM|World Community Grid|Restarting task X0000055151402200509061751_0 using hcc1 version 606

All that I know in setting and handling check-pointing is collected in a Start Here FAQ. The index in in my sig for the shortest route to it.

ciao

Gabriel,

Unfortunately the progress indication is linear whilst the amount of computation effort to do the first 2-3% takes 5 minutes, the last 2-3% can take an hour and longer. That's why looking at the checkpoint before closing is helpful to minimize the time loss. It may be too much but it's technically impossible to save an analysis step in the middle and resuming it at the same point and not running the risk of result corruption. In my and for many using standby, the operating system seems to be able to freeze memory image and continue. In Standby the computer still uses a little power to maintain the memory state.

As for ram availability: HCC uses between 40-57mb per job, it varies but is small compared to other WCG projects, so a 600mb limit should not make a difference. Some jobs are more complex than others. Generally on my laptop they take 6 hours, so if a longer comes along of 7, then for a machine that took 14 hours it becomes 17 etc.

ciao

Open Windows Task Manager by pressing Ctrl+Shift+Esc, select the "Processes" tab, and if you don't see the information you wish about memory, from the Menu bar select View->Select columns to choose the columns you want to see.

Cheers. Jean.