I just decided to try the 4.40b version and all of a sudden I have a WU (on my AMD x2 3800+) that indicates it is going to take 40 hours to process. Until now the usual process time was ~25 hours. How do I tell if the new software is the cause or if the WU is unusually large? I have read in the forums of people knowing what the size of the WUs are but I have no idea how to find the size of the WUs.
Any help would be appreciated,
Eeyore
Copyright © 2024 Einstein@Home. All rights reserved.
Version4.40 or large WU
)
Hi!
Coinciding with the app version switch, your computer also got a new data pack for a "monster" workunit that lies in the ca. 650 credits range. So it's normal that the expected time to completion went up. The 4.40 version is reported to be as fast as or slighter faster than 4.38.
Happy crunching,
Bikeman
RE: RE: I just decided to
)
But my old WUs were in the 637 credit range and only took about 25 hours to process. I still wonder about what has happened to change the processing speed. With the change in processing time I should be getting credit in the 1000 unit range! This is beginning to sound like government accounting tactics to take my credits and tell me I am making more money than ever (LOL). Actually, I am just curious as to whether the new update has changed things that slowed down the processing. I will continue to crunch since my computers are normally idle anyways.
Thanks,
Eeyore
I thought I read somewhere
)
I thought I read somewhere where Application Changes reset the DCF to 1? If this is so, and your DCF was a lot lower then it'll take time to relearn the speed.
RE: RE: RE: I just
)
Sorry for the additional message but I forgot to mention that the WUs that I am crunching on my Q4 machine haven't changed in their expected finish times and I also changed to the new 4.40b software. Has the software update changed to make it work differently with the Q6600 processor and the x2 3800+ processor?
Thanks again,
Eeyore
RE: Sorry for the
)
Just to clarify a little, the new (and slower) "debugging" code was first installed in v4.38. It was modified slightly in v4.40 (and made ready for S5R3). But, apparently, v4.38 and v4.40 are comparable in speed. However, different CPU's seem to handle the new code differently. For example, the slowdown (from v4.33 to v4.38) on my Pentium M was about 7% but it was about 15% on my P4 Prescott.
RE: RE: Sorry for the
)
RE: But I have a 50%
)
You are claiming over 200 points more for the Wu as well so it might not be that much of a problem.
Hang in there a bit longer to get all the answers. :)
Anders n
Edit
language
RE: RE: But I have a 50%
)
If you divide the credits by the hours of cpu time spend, you'll see that for both results, you get almost exactly the same ratio: 16.4 credits per CPU hour.
So nothing to worry about.
CU
BRM
RE: RE: RE: But I have
)
RE: That is my problem.
)
Have a second look: credits on your AMD x2 have gone up recently from 448 to 653 per workunit:
86790487 34668143 6 Sep 2007 5:27:13 UTC 8 Sep 2007 15:11:11 UTC Over Success Done 142,984.69 653.36 pending
86779363 34664429 5 Sep 2007 11:41:27 UTC 7 Sep 2007 7:42:23 UTC Over Success Done 98,336.16 448.57 448.57
Happy crunching
H-BE