ugh, Project 1487 again, 7.5 hours per 1%, 31 days

[tmoble]

I thought this was fixed somehow? P1487, Run 0, Clone 317, Gen 19. Looks like 31 days. This has the large clone number deal.

this is an AMD 2GHz machine with 1GB RAM. No other apps running.

First time I got this 1487 millstone it was 30 days or so, they generously awarded 600 points on the deal. Others that started later got like 1400, but they refused to give me that, something about having started too early.

I did get one of these 1487 after that, it was about 2.5 hours per, turned out OK.

Anybody got any ideas how to avoid getting stuck with this deal? This is a Linux box, could I just blow the software off and re-install? I recall there's a question about allowing large WUs, or is that only on the Win version?

[sneakers55]

I thought this was fixed somehow? P1487, Run 0, Clone 317, Gen 19. Looks like 31 days.

this is an AMD 2GHz machine with 1GB RAM. No other apps running.

First time I got this 1487 millstone it was 30 days or so, they generously awarded 600 points on the deal. Others that started later got like 1400, but they refused to give me that, something about having started too early.

Anybody got any ideas how to avoid getting stuck with this deal? This is a Linux box, could I just blow the software off and re-install?

Please don't dump WUs. F@H will be sitting there until the preferred deadline waiting for it to come back, and when it doesn't it'll be reissued to somebody else that has to do it from scratch.

I recall there's a question about allowing large WUs, or is that only on the Win version?

I am quite sure that's common to all the versions.

[tmoble]

Well, I wasn't figuring to dump it and leave it hanging. I'll just write a little script that monitors for the PID and shuts the box down when it goes away or something like that. This machine usually doesn't have a monitor and keyboard so I can't see the screen. I hooked it up when I realized my points weren't accumulating like normal.

Does anybody know if excluding the large WUs will work? I actually don't know if it comprises a large WU.

[sortofageek]

Project 1487 work units are worth 1,741 points.

See the project list here ---> http://fah-web.stanford.edu/psummary.html

If your computer completes at the rate of 2.5 hours per frame, the WU should be complete in about 10.4 days.

[bruce]

Project 1487 work units are worth 1,741 points.

See the project list here ---> http://fah-web.stanford.edu/psummary.html

If your computer completes at the rate of 2.5 hours per frame, the WU should be complete in about 10.4 days.

If the same WU is processed with SSE disabled, it will take about three times as long as if it's run with SSE enabled. That comes pretty close to the 30 days that you reported for the other WU. I doubt that you disabled SSE in the BIOS or you reconfigured the client to disable optimizations, but if your machine crashed and FAH was restarted, FAH can disable SSE, itself, to help protect against the system crashe that it might have caused. Do you always start the client with the -forceasm command line flag?

[tmoble]

Bruce has actual hard info.

I recall that after the last fiasco I used -forceasm, but don't recall doing it this time. Looked at the log, no sign of it. Started 2/3, up to 88%. Got a feeling it wouldn;'t be a good idea to to stop and re-start it with -forceasm at this stage of the game.

Or would it? Could I do that without trashing the WU?

sortofageek, I recall that the first I went through this and ran a 1487 again it took about 2.5 hrs and was about 10 days.

So, It's all my fault again. Dang.

what about stopping and starting with -forceasm?

[tmoble]

too late, already did it. It started from checkpoint, 887419 out of 1000000 steps (88%)

I got:

Extra 3DNow boost OK.
Extra SSE boost OK.

and the usual long 1-4 interactions


Maybe it will do better.

[sortofageek]

Glad you went ahead. Restarting with the -forceasm switch wouldn't be a problem, but you already know that. Unless you suspect you are having hardware problems, it's probably a good idea to always start using that switch, as well as the -verbosity 9 switch option.

[tmoble]

Yeah, we're cookin' with gas now. about 2.5 hours per step.

Guess I'll soon find out if it will carry the options over from WU to WU. As long as it doesn't restart the executable it should.

that verbosity flag, does it refer to the log level? log more with that? nearly empty 200GB drive, ought to take awhile to fill that.

Thanks for taking time to help.

[anandhanju]

If your shortcut has the forceasm switch in it, it'll always use that for each WU, every time.

And yes, verbosity does refer to the amount of logs that get printed to the console/file. Even with verbosity set to the max level (9) and the "humongous log file" bug fixed in the recent beta, the log shouldn't be anything more than a couple of MB at the maximum (if the client is never stopped). My WinXP decides to restart itself (*bleep*) after an update at 3AM every so often and log file sizes have never been a problem.

Fold on!

[sortofageek]

Setting verbosity at 9 makes the log a bit more "chatty." We get a little more info in the logs that way, which can be helpful in troubleshooting.

[tmoble]

it's a linux box, I'm just running the executable like ./FAH504-Linux.exe -forceasm. Not really a shortcut. Guess I could make one.