I have had some memory sticks fail when running these bigger work units. I do recommend trying memtest86 to verify that hardware. Then I also like to boot with an Ubuntu livecd and run the Disk Utility to check the SMART status and also check the performance of the hard drive. I had a hard drive fail and cause any computer it was on to come to a crawl. I've also seen a hard drive boot and work until a certain part of the disk was accessed. This one didn't even complete the speed test on the Ubuntu Hard disk utility. I would say it is possible you had some hardware near the end of its life and these larger work units pushed it over the edge. I have sent back several pairs of memory over the years for warranty replacement. The other problem I watch for is bad capacitor(bulged). The area around the CPU is usually the main problem area. I wish you the best of luck on solving your problems.

I have a few boards in the same situation and looking to replace the capacitors from badcaps.net. I still need to upgrade my soldering pencil to a higher wattage. to repair them. no hurry here yet as the electric bill has increased enough to leave most of them off but my fastest crunchers.
Is that how you are going to fix your P4 or retire it for a new AMD 6 core?

those weren't betas, but the problem was just the same: cpu time was reset over and over again:
i had 3 work units, they all set one checkpoint at about 8 cpu minutes and then crunched on to 55-75 % without any further checkpoints.
Selected "Leave application in memory", "Run always", "Run When Computer in Use" and so on... All wu's fell back to the initial checkpoint whenever i opened the notebook lid (computer was running, just locked the screen). even tried running only 1 cep2 wu at a time and reducing cpu usage to 60%, nothing helped. after 25 hours or so, i killed the 3 jobs, always at 8 min cpu time.

And sure enough, with a little work, starting some downloads from MS in the range of 1.3GB (free trial of Office Professional +) and running a X-WLAN backup netting to 350MB... here's the product:

Wed 20 Oct 2010 12:03:31 PM CEST World Community Grid Task E200468_413_A.24.C22H16OSi.1.2.set1d06_1 exited with zero status but no 'finished' file
Wed 20 Oct 2010 12:03:31 PM CEST World Community Grid If this happens repeatedly you may need to reset the project.
Wed 20 Oct 2010 12:03:31 PM CEST World Community Grid Restarting task E200468_413_A.24.C22H16OSi.1.2.set1d06_1 using cep2 version 619
Wed 20 Oct 2010 12:03:32 PM CEST World Community Grid Task E200469_748_A.25.C17H9N5S2Se.25.0.set1d06_0 exited with zero status but no 'finished' file
Wed 20 Oct 2010 12:03:32 PM CEST World Community Grid If this happens repeatedly you may need to reset the project.
Wed 20 Oct 2010 12:03:32 PM CEST World Community Grid Restarting task E200469_748_A.25.C17H9N5S2Se.25.0.set1d06_0 using cep2 version 619
Wed 20 Oct 2010 12:03:33 PM CEST World Community Grid Task CMD2_0875-1HCI_A.clustersOccur-1ONV_B.clustersOccur_10_458795_463007_459950_460386_1 exited with zero status but no 'finished' file
Wed 20 Oct 2010 12:03:33 PM CEST World Community Grid If this happens repeatedly you may need to reset the project.
Wed 20 Oct 2010 12:03:33 PM CEST World Community Grid Restarting task CMD2_0875-1HCI_A.clustersOccur-1ONV_B.clustersOccur_10_458795_463007_459950_460386_1 using hcmd2 version 614
Wed 20 Oct 2010 12:03:34 PM CEST World Community Grid Task CMD2_0875-1HCI_A.clustersOccur-1KU6_A.clustersOccur_127_470491_470850_470568_470603_0 exited with zero status but no 'finished' file
Wed 20 Oct 2010 12:03:34 PM CEST World Community Grid If this happens repeatedly you may need to reset the project.
Wed 20 Oct 2010 12:03:34 PM CEST World Community Grid Restarting task CMD2_0875-1HCI_A.clustersOccur-1KU6_A.clustersOccur_127_470491_470850_470568_470603_0 using hcmd2 version 614

BOINC still suffers from timeouts and it affects all sciences and because CEP2 does wide apart checkpoints it hurts.