OK, I have some weirdness going on here. On identical i7's, except RAM which is 8GB for one, and 10GB for the other.

Both machines run twin 5970's.
Both machines run BOINC 6.10.34.

On the 10GB machine RAM is about 73% or 74xx MB.
On the 8GB machine RAN is about 60% or 49xx MB.

The 10GB machine runs normally, except longer finishing times, and staying at 0% then bouncing to 143% and working backwards.

The 8GB machine runs for hours at 0% before finishing. The GPU usage meter also shows 0%.

Looking more carefully I note on the slow machine the WU's are in the free_5_32xxxx range, on the OK machine the WU's are in the 64 - 65xxxx range.

On my third i7 the usage meter shows averages of 29% over four GPU's, compared to 91-94% on the OK machine.
The third i7 also has 10GB RAM, shows two tasks counting backwards after 2:00 hours run time. and two at 0% after 1:59:49 hours run time.

Holy Crap, I just got a ! Do you want to change the color scheme to improve performance msg. Sheesh, and I thought Cosmo was memory intensive!!!!

EDIT: I brought the resource meter up to see what is going on, and lo and behold Wu's show a normal 100Mb each. Just as I was thinking, "That's weird" they finished, new ones started, available RAM jumped from 77MB to well over a GB and tasks running as normal at 96 - 98% usage. They are taking under 2:30 each. This machine is water cooled and O/C'ed to 800MZ on the cards, So I now got two OK machines and one dud!

---

I expect a 0.22 release tomorrow if I find the time to compile all versions.

Done.

There is a new set of ATI apps fixing the small bug with the progress bar. It was only cosmetical and nothing else material changed. Read the included readme.txt!

I would guess the project uses this fixed version as stock app in a few days. If the broken progress bar doesn't disturb you too much, there is no need to update.

What's left are the CPU apps.

Works a treat..but then, that's no surprise when it come to CP and his programming capabilities!

I am seeing much the same thing on a new 5870 with the DE_13 series WU's. Crunches for 2:11 showing 0% progress, jumps to 140%, then counts backward until 2:52 when it finishes.

I am seeing much the same thing on a new 5870 with the DE_13 series WU's. Crunches for 2:11 showing 0% progress, jumps to 140%, then counts backward until 2:52 when it finishes.

See Cluster Physik post about a new app that overcomes the progress bar issue.

Or my site where they are split out into individual apps.
http://www.arkayn.us/milkyway/index.html

---

in my case, the ones that grant 213.76 are in minority. Half percent take majorly 145 seconds up until 190 seconds and the other half take majorly 770 seconds down until 445 seconds. Still the granted credit is the same. I know this is mentioned before. The increase in crediting isn't approximate with the increase in crediting in case of big wus.

My 5970's read 0.000% then jump to 143% and count back to 100% but what I also noticed was it says .05 cpu required yet the cpu time is only a few seconds less than the total run time. My 295's take a long time but percentages progress as normal and the cpu time is a fraction of the total job time as usual

---

The CPU time displayed for ATI tasks on your results page is actually the GPU time. ATI tasks are shown this way for BOINC scheduling reasons. If you click on the Task ID for an individual task you will see near the end of the stderr out section that the actual CPU time is very low.

ok, so are these new tasks good, bad, or hit or miss?

Also, this inquiry is in regards to the *CPU* app only. I do not have a GPU-capable system. I came looking at this thread because of, well, the subject title... :-)

Brian

As I understand it the WUs referred only have a problem showing progress, or no movement, of the "progress bar percentage". However, when completed they are fully valid when they are crunched and report.

The behaviour I mention is for GPUs, and may not apply to CPU crunched WUs. Either way the work is OK.

---

Go away, I was asleep

The behaviour is also for CPU-apps. I started the thread for 0.20 CPU-app. But I agree with you that the results of the WU are valid and the granted credits are fine.

Hi,

banditwolf send me to this topic with my problem. Here the last reply.

Ok i have also a strange problem with the WU´s.



Should we stop crunching for the moment? Or is it just a problem with the % information?

It seems most wu's are fine but a few have errored. It is being discussed here: http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=1600. If you are using the optimised app a new one was put out that fixes the % problem. The link is in the same topic.

For information, i don´t use any optimiced apps. Always just the offizial apps! Also i have not the problem with all WU`s. Just with a few. I hope this will help you to find the problem.

The old searches did run like 250s here and gave 213.76 credits

I see the new searches run from 461.05s to 663.27s for 417.48 credits

While the 461s for 417 credits is in range, 663s for the same credits is far off and should give around 565.

The progress bar first stayed at 0.0% and was above 130% when finished.

They are ending correct even with the progress bar not reading correct. Just give them some time.

Have you checked your system? They are only going about 50% longer on mine, and with the higher credits the credits per hour is much higher. Stop the computer and get out the canned air (even if it is not time to clean it). While it is open check for anything that may be louse, Etc.

---

Why is anybody even Posting about this when there was a Fix Posted in this Thread on the 14'th for the Progression Bar, install the App 0.22 Fix > http://www.arkayn.us/milkyway/index.html and be amazed at how good the Progression Bar works or don't install it and keep whining about it, it's your choice ... :P

---

STE\/E

Why is anybody even Posting about this when there was a Fix Posted in this Thread on the 14'th for the Progression Bar, install the App 0.22 Fix > http://www.arkayn.us/milkyway/index.html and be amazed at how good the Progression Bar works or don't install it and keep whining about it, it's your choice ... :P

Could it be that some are having that some problem with the CPU units? The fis there for the AMD CPU is for the GUP, not CPU so it is not fixed.

---

Could be, but some are definitely GPU related so the fix should work for them ...

---

STE\/E

Why is anybody even Posting about this when there was a Fix Posted in this Thread on the 14'th for the Progression Bar, install the App 0.22 Fix > http://www.arkayn.us/milkyway/index.html and be amazed at how good the Progression Bar works or don't install it and keep whining about it, it's your choice ... :P

Unless I am mistaken the updated versions fix the progress bar problem in the ATI based apps (eg GPU). However, units being crunched on CPU's are also suffering from this problem.

To the best of my knowledge the 0.19 SSSE3 opt app has not been updated (in fact there does not even appear to be a version 0.20 for SSSE3). Plus the site you are linking to does not have a SSSE3 Win 32bit opt app for CPU's. Furthermore if you look at the following post you will see that the update there is specifically for ATI GPU's.

That's why there is still 'whining' here. Not everyone has an update available to them.

---

Never surrender and never give up. In the darkest hour there is always hope.

I was wrong, but the fix is there for some that Posted here about their ATI GPU's so it should work for them. If the Fix is not Applicable then feel free to continue to Whine ... ;)

---

STE\/E

I have the problem on xp cpu, Boinc 5.10.45.

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.