309547848 234283116 9 Feb 2011 5:27:07 UTC 9 Feb 2011 6:03:17 UTC Completed and validated 419.19 56.63 0.53 1,164.00 Anonymous platform
309546527 234282837 9 Feb 2011 5:19:06 UTC 9 Feb 2011 5:50:00 UTC Completed and validated 498.20 66.71 0.62 1,385.72 Anonymous platform

309548022 234283290 9 Feb 2011 5:28:14 UTC 9 Feb 2011 6:10:12 UTC Completed and validated 418.20 56.38 0.53 1,164.00 Anonymous platform
309547846 234283114 9 Feb 2011 5:27:07 UTC 9 Feb 2011 6:03:17 UTC Completed and validated 419.20 56.71 0.53 1,164.00 Anonymous platform
309546367 234282688 9 Feb 2011 5:17:55 UTC 9 Feb 2011 5:41:38 UTC Completed and validated 419.17 56.63 0.53 1,164.01 Anonymous platform
309581467 234305042 9 Feb 2011 6:23:50 UTC 9 Feb 2011 6:59:03 UTC Completed and validated 350.56 43.26 0.31 1,164.01 MilkyWay@Home v0.23 (ati13ati)
86_s
79_s
82_s

I wonder if they come with check pointing?

I do not believe that they do. Also none of the ones I've completed have validated yet, hope that changes by morning.

---

I have one validated by now. They running about 750 s or 630 s. So my cache is growing to more than a half hour for two 5850. This is a step in the right direction! Maybe checkpointing for slower GPU would be a good idea!

---

I notice GPU usage has dropped significantly!

I don´t notice that. CCC shows for my cards between 91 and 95 % usage independendly from the WU calculated. I notice CPU-usage is about double more.

---

The new WUs take tens of minutes to complete for me. It would be really nice if they had checkpointing, to put it mildly.

Hello,

the new work units are hugely resourse hungry so much so one of my machines just crashes as soon as you start milkyway
for some reason the wu are doing the same thing as collatz in that box and using all the 4 gig of ram and loads more with everything else switched off including all other projects
thats on a twin 5970 x-fired machine but even on my main machine only using 1 4890 another 1 gig of ram is being used that wasnt yesterday for some reason by MW

as i say collatz does this as well making it unusable on that machine i just hope MW hasnt gone the same way !

best regards
Ian

---

....Please Join team Scotland HERE

Your machine shouldn't crash. Aren't the GPUs overheating (I underclock mine when doing MilkyWay)? Don't you have a faulty memory module in the crashing PC?

http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=248968&offset=0&show_names=0&state=2

Here is my machine with pending long wus (3x6970). I haven't observed a change in GPU load but CPU usage may increase as the timing is higher than previous short wus. Although it might just be due to the increase in size?

Anyways, i do like this idea as it will increase the utilization of GPU and decrease the load on servers.

Thank you for that team:)

Just noticed this a few mins ago and my first thought was that the server was borked! lol

Very cool to have these.

Observations are, slightly higher GPU usage than previously when using the stock app with all CPU cores loaded (I could be wrong here but it's not lower on my 5870). Checking out the completed tasks I'm using overall less CPU time than running a few of the shorter WU's to equal a longer one. And I'm not seeing anything odd with memory usage, it's sitting at about 45mb for the program.

---

Hello,

the new work units are hugely resourse hungry so much so one of my machines just crashes as soon as you start milkyway
for some reason the wu are doing the same thing as collatz in that box and using

as i say collatz does this as well making it unusable on that machine i just hope MW hasnt gone the same way !

Hi Ian, probably not the projects but the card/machine. Have 10 ATI GPUs running MW and no crashes so far. The long MW WUs are using 45k to 55k of main memory each. Collatz runs without errors on all machines. Both use app_info.xml and are at 98-99% GPU. The key to your problem is that both projects are failing and both of these clients are very stable. Run tests on the card, make sure your drivers are installed, try the card in another machine. Best of luck to you!

The long tasks are taking just under 9 minutes on my 5870, compared to 90 seconds for the short units. Nice to see some longer WUs!

I am seeing some of these varying long WUs coming through. The very long ones are taking about 6 x as long, while the shorter long ones are about 4 x as long as the stock WU..

HD3850 temperature a little higher than before.

---

Go away, I was asleep

1st - 4830 from 5~7 Minutes per WU is now taking 30 minutes
2nd - 5830 from 2 1.9 to 2.2 minutes is now taking 10 minutes

Weird, but maybe now the Milkyeay team is increasing the WU information for more processing ( hope the points raise too... ).

Regards!

The thing I'm wondering about is in the news thread about the new NV app Matt says something about WU's being created by accident that were longer...

We may have been too soon in our rejoicing...

---

I did a long one and observed it crunch.
It took 445.60 sec. clocktime and 8.31 sec. CPUtime on my HD6970 (driver: 1.4.900 Phenom II X6 1090T Win 7 64 8 Gig)using anonymous platform
It claimed 0.06 and was granted 1,164.01 credits
The memory used was no more then 47.000kB and GPU-Z 0.5.1 showed between 82% and 86% GPU load.

What is the consensus on these improperly long WU's? With both MW and Seti down for several days, the only thing running in the end is I assume one of these units. It is a 0.50 separation unit running on the CPU. It has wrapped over 100% now several times over. Is it going to finish eventually or should I abandon it?

---

Only a problem with the progress bar (we had this before when the number of streams got increased?) with no influence on crunching the WU. It will finish if you let it run.

Thanks, I meandered over all the message traffic and came to no conclusion other than it might be best to just let it run. I left it alone and just found it out has reported successfully.

http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=309496092

No harm, no foul.

Keith

---