Windows S5R4 SSE2 power App 6.07 available
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4330
Credit: 251394672
RAC: 35874
29 Nov 2008 13:04:06 UTC
Topic 194063
(moderation:
A SSE2 App for Windows can be found on the Power User's Apps page.
For this App to run your CPU must support SSE2 instructions.
The science code and build process is the same as for the 6.05 and 6.06 Apps, so the speed shoul dbe exactly the same.
This App should fix a problem with checkpointing introduced in the 6.06 Beta App. If it turns out that this problem is fixed, I'll make this 6.07 into a full-featured S5R4 release candidate.
As for now: Only run this if you're sure of what you're doing
BM
BM
Windows S5R4 SSE2 power App 6.07 available
)
Successfully installed and has taken 3 good checkpoints so far on an X2 4600+ system.
Seti Classic Final Total: 11446 WU.
RE: RE: A SSE2 App for
)
Is it any different to the 6.05 power client?
Shih-Tzu are clever, cuddly, playful and rule!! Jack Russell are feisty!
RE: RE: RE: A SSE2 App
)
Not faster, but two bugs should be fixed: the "crash after done" and the "could not rename".
BM
BM
Thanks I have yet to
)
Thanks
I have yet to encounter either of the issues you mentioned.
Shih-Tzu are clever, cuddly, playful and rule!! Jack Russell are feisty!
RE: This App should fix a
)
This is the output for a task (run on a Q6600) that was started with app version 6.05 and finished off with the new 6.07. As you can see, there is no evidence of any problem with checkpointing.
I'm ready to test the SSE version as soon as you're ready to release it.
Cheers,
Gary.
RE: A SSE2 App for Windows
)
I've gone into work and set up a couple of Windows/SSE2 machines (Northwood P4s) with the 6.07 SSE2 app. Both were running 6.05 previously. The upgrading was uneventful and both are writing new checkpoints without issue. Looks to me like the problem is fixed. There are a number of Windows/SSE hosts (Athlon XPs and Tualatin PIIIs) now waiting anxiously in the wings .... :-).
Cheers,
Gary.
RE: The upgrading was
)
Same here, checkpoints seem to be working well. I'm curious how many tests it will take for the "crash after done". I ran quite a few of the 605 before I had the one instance of it. From what I've seen, it doesn't seem to be a constant problem, so I'm assuming it will take more tests to determine if it's still an issue.
Is it me or is 6.07 slow. I
)
Is it me or is 6.07 slow.
I thought the first one partially completed with 6.07 could have been due to lack of checkpoints in 6.06.
But next one is at 18.5% after 2:10 which if linear progression is assumed would be over 11 hr to completion, 6.05 and 6.06 were completed at about 6:15 +/- a bit.
I'm at about bottom of trough.
edit] Just thought I would mention Seti and Seti Beta tasks are progressing as normal.
RE: Is it me or is 6.07
)
Yes, your first one which was started with 6.06 and completed with 6.07 essentially lost all record of the 6.06 crunching when you shut it down to make the switch to 6.07. You could have avoided the loss by deleting the offending checkpoint (.cpt file in the relevant slots directory) and watching for a new one to be written. Doing the switch to 6.07 immediately after this would have avoided the loss. I've posted about what happens if you lose power or shut down your host over in the 6.06 thread.
I don't think there is any real difference between 6.07 and 6.06 as far as crunching speed on Q6600 hosts is concerned. However, I have 6 Q6600s, 5 on Linux and 1 on Windows which do show the odd dramatic slowdown occasionally for just the odd task here and there. I've been recording crunchtimes on all these hosts for a while now so I'll have a look for some examples and post again later when I find some. I suspect you may be seeing the same effect. I posted about this previously so I'll dig out the link to that as well.
Cheers,
Gary.
RE: You could have avoided
)
I read your thread and considered doing it (wish I had now) since I was switching task between a Seti MB and the Einstein 6.06 on my AMD XP. BTW thanks for that tip.
I thought since I'd set Boinc to run with the app in memory while suspended I would be ok, but I guess it decided it didn't like the switching and re-started from 0.00%. Deleting the .cpt file now though, and guess I'll have to for any other test I run with the 6.06 on that pc. I'm starting to question if waiting for a version of 6.07 with SSE might be the one I should wait to test on further.
I don't notice any difference in the one 6.07 I'm doing right now, but I did notice one of the 6.06 I completed did show a longer crunch time than I'd ever seen with the 6.05