SSE v2.0 crashes on Windows ME

[filipvanlaenen]

I have been running the client on this machine since 25 August, and installed the SSE v2.0 client as soon as it was announced on this forum (don't remember when, but it was several weeks ago). Since this evening, the client crashes consistently at the following point:

[Wed Nov 24 16:53:24 2004] got k and n from cache
[Wed Nov 24 16:53:25 2004] Intel(R) Pentium(R) 4 CPU 2.00GHz detected. Enabling cpu specific optimizations.
[Wed Nov 24 16:53:28 2004] restarting proth test from cache (k=28433, n=7487977) [0.2%]
[Wed Nov 24 17:12:57 2004] got k and n from cache

(The last line is after the restart).

The error message I get is a generic Windows message saying that SB caused an error, so nothing useful there as far as I can see.

I sent a mail to L. Helm and D. Norris - should I do anything else?

[vjs]

Is your computer coverclocked or overheating, I had the pre sse2 client working on a computer no problem but when the new client was installed it crashed continously.

I think the issue was heat???

It's not a box I normally get a chance to crunch with so I uninstalled the client and went on to something else.

Check for heat, non working fans, bust bunnies, also try to find a program like sissoft sandra that reports your CPU temperature. If it's less than 50C for the CPU heat is probably not a problem. If it's over 55C maybe, but approching 70C find a new heatsink or some extra fans.

All temperatures underload of course.

[Joh14vers6]

What happen if you restart the client after, let's say 5 minutes?

[filipvanlaenen]

I don't think the heat is a problem. The crash is the first thing my computer does, so to say, when I restart it after it has been shut down for more than eight hours.

I have a feeling it must be something in the software, because it always crashes at exactly the same place, namely at the end of the block.

[chvo]

Maybe the cache got corrupted, try deleting the z7487977 (not a big loss since you're only 0.2% far in the test).

EDIT: if the client crashes at the end of a block, try disabling "send intermediate blocks" for a few blocks and then re-enable it. That should solve your problem.

[Joh14vers6]

Originally posted by jjjjL 10-20-2004 12:20 AM
I found the offending code that can sometimes divide by zero.

It has to do with the ETA calculation not any of the important math. I can make a fix but I'm going to investigate a litte first to see how it is this situation can come about in the first place.

This seems to only happen on a restart when zero blocks are done and the save file is in a goofy state.

A workaround for now is to turn off transmission of intermediate blocks transmission for 1 block. You should be able to turn it back on after that.

Cheers,
Louie

It is a known bug. We still wait for the patched client.

[filipvanlaenen]

Yes, thanks, the work-around worked.