hooziewhatsit said:
about the gromacs core...
I've noticed (maybe someone can verify) that if I shutdown windows twice, letting windows close the client, folding will detect 2 improper shutdowns, then delete the entire WU.
therefore, always close the client first using ctrl+c
I think I've lost two wu's like this, one at 90% and another at 35%.
my duron 1000@1168 is doing a de novo at about 10:40 a frame
my tbird 700 is doing a de novo at about 19 min. a frame.
fold on!
Thanks for the heads up hoozie. I have a unique problem with my new 1800+ in that it locks the screen as I go through it with the KVM. Folding keeps working in the background (EMIII still visibly operating) but I lose all keyboard and mouse function. Eventually power management turns the screen off and I lose visual but work continues. An improper shutdown is the only recovery and now with your info I know not to do that more than once during a WU. Good looking out. I might go to USB keyboard/mouse on that box.
Edit: Guess I would be the perfect candidate for a test. Can't do it on this unit tho. It's too far along. I'll try to force it on the next one.
Ok. Test verifies the termination after two improper shutdowns. On second shutdown the work folder the following occurred :
The core examined the work files. When it saw two improper shutdowns it automatically went into delete mode (Stanford terminology) and deleted the work and frame files. It then executed a "Special Exit". At this point the program continues. It finds that the work files are missing, it removes the work from queue, deletes the current work unit and continues. The assignment server gives it work and it proceeds to work on the new work unit.
I thought I had given sufficient time for the WU to complete before trying to reboot my system, but I was in error. Lost 5 points being impatient. Please learn from my mistakes. We can't afford but so many of these. Be very careful if you have moderately stable overclocks. This core will not tolerate an imperfect system. Hope Stanford is not trying to make this an exact science.
BTW: DENOVO PROTEIN in water
Tbird1.4@1580 8.67min/frame / 14.43 hrs/WU (Win98/ working FAH only, DDR PC2100 (MIcron), 256MB, K7S5A Rev. 3.1) (this WU failed at 35% and was shut down by the core as causing instabilities. It states all of this in a narative on the console. The wu was closed and sent back to glean what info they could from the attempt. The core did all of this automatically).
XP1800+@1610 - 6.43min/frame / 10.73 hrs/WU (WinME, working FAH only, DDR PC2100 (crucial) 256MB, KR7A-R-133)
XP1800+@1633 - 6.26 min/frame / 10.43hrs/WU (WinME/ working FAHonly, DDR PC2400 (Corsair), 256MB, KR7A-R)
The assignment server has not issued one of these proteins to the P4 yet. Will post here when it does.