My 4 ATI cards have completed around 140 ATI APs TODAY. Each completed task was instantly replaced with a new download.
Obviously, the problem only affects certain Hosts or the board would be flooded, and the 'total channels to do' number would not be rapidly counting down.
I don't know why certain Hosts are effected, but, most Hosts aren't.

My 4 ATI cards have completed around 140 ATI APs TODAY. Each completed task was instantly replaced with a new download.
Obviously, the problem only affects certain Hosts or the board would be flooded, and the 'total channels to do' number would not be rapidly counting down.
I don't know why certain Hosts are effected, but, most Hosts aren't.

I wonder if it could be that your rigs are AP-only, so the scheduler doesn't have a choice, whereas my machine will take anything that comes down the pipe. That's why I hadn't noticed the drought, since it was getting enough MB work to keep it busy. (Which for that machine isn't a whole lot!)

Two of my three hosts had a nVidia card, running MBs, and an ATI card up until a couple hours ago. I just removed the two cards from the XP host and installed a 'new' 6870. One machine still has a nVidia card in it running MBs. So, that's not it.

I just pulled an ATI 4670 and NV GTS 250 out of the XP Host, Heavily edited the State file to run the CUDA MB tasks on the ATI card, installed the 6870....and it is working just fine. It was working just fine a few hours ago with the GTS 250 as well.
Shrugs...

Shrugs...

Agreed! I'm sure some other ideas will surface.

From what I'm seeing, individuals running anonymous platforms are not having difficulties receiving AP opencl_ati_100 WU's whereas those of us who are running stock platforms are not.

Anyone else confirm?

Keifer

---

Since 26 November my GPU has only been getting opencl_ati_cat132 for S@H and opencl_ati5_cat132/AstroPulse v6 v6.06 (cal_ati) for S@H Beta.

A new scheduler was deployed on Seti Beta on the 22nd Nov at Seti Beta to fix the Intel GPU anonymous platform Bug, and a few days later at Seti Main, either we're jumped a few untested changesets,
or we're reverted back a bit (There was a problem with AMD GPUs where the the APR was too high and wasn't getting updated because the Wu was unfeasible, and because the APR wasn't updated the APR remained too high)

Claggy

Thanks for that info. It fits my time frame and thoughts concerning events at SETI Beta. May I assume you are suggesting that in a lame attempt to effect repairs on our end it would be unwise for us normal minions to panic by uninstalling and reinstalling AMD GPU, BOINC and Windows drivers and such?

Keifer

---

You really can't hurt anything by running the Lunatics Installer. It might work. The instructions are already posted, Running BOINC as a service

I found another missing and fixed it. Hopefully you'll get the normal versions now.

---

@SETIEric@qoto.org (Mastodon)

I found another missing <max_cal_target> and fixed it. Hopefully you'll get the normal versions now.

Yes, that works (at least at Seti Beta), snipping Wu's out of my client_state.xml got the cal_ati tasks expired, and more sent, doing it now gets them resent as opencl_ati_100

Thanks

Claggy

oops! My bad. I clicked the wrong button.

Rookie mistake!

---

I found another missing <max_cal_target> and fixed it. Hopefully you'll get the normal versions now.

I've been monitoring the threads by both you and Claggy over at Seti Beta. I'm sure I speak for everyone when I say thanks to the both of you for, addressing the stock AP opencl issues in less than a day, and your update postings here as well!

Keifer

---

Well, I reinstalled Lunatics and I got three ATI Astopulse WU's before the well ran dry.

---

Began receiving WU's from the Beta site on 14 Dec.

Thanks again for addressing that.

Still no AP opencl_ati_100 for stock apps on Seti main. The SSP again shows several thousand tasks ready to send.

Anyone?

---

Well, My ATI 7870 running Lunatics is getting AP, but my machine with ATI 4870 running stock app is getting nothing.

---

Well, My ATI 7870 running Lunatics is getting AP, but my machine with ATI 4870 running stock app is getting nothing.

Try the procedure I mentioned earlier with the 4870 host.
Set the Host preferences to ask for only ATI APs, http://setiathome.berkeley.edu/prefs.php?subset=project
In BOINC Advanced View/Projects, hit the Update button until the Event Log shows it is asking for ATI tasks.
Suspend 2 CPU APs, then quit BOINC.
Open the client_state.xml file with Notepad and Find 'Suspended'.
Delete the Results Entry for the 2 Suspended tasks, everything between and including <result>...</result>.
Save the file, then restart BOINC.
The CPU APs should be resent as Lost tasks assigned to the ATI GPU.
Since you don't currently have any GPU tasks on that Host, there shouldn't be a problem with quotas.
If the tasks are Expired instead of resent, there is a problem with the Server.
Have Fun :-)

If the tasks are Expired instead of resent, there is a problem with the Server.

T-Bar, you're out of touch, we know there is a problem with the Server, it's been fixed at Seti Beta, just waiting for it to fixed here,
I'm sure when enough time has passed since applying the fix to Seti Beta, then it'll be applied here.

Claggy

If the tasks are Expired instead of resent, there is a problem with the Server.

T-Bar, you're out of touch, we know there is a problem with the Server, it's been fixed at Seti Beta, just waiting for it to fixed here,
I'm sure when enough time has passed since applying the fix to Seti Beta, then it'll be applied here.

Claggy

Thanks for the Update. Obviously the people here thought the problem had already been corrected. Too bad I had to waste my time when someone else already had the answer.

If the tasks are Expired instead of resent, there is a problem with the Server.

T-Bar, you're out of touch, we know there is a problem with the Server, it's been fixed at Seti Beta, just waiting for it to fixed here,
I'm sure when enough time has passed since applying the fix to Seti Beta, then it'll be applied here.

Claggy

Thanks for the Update. Obviously the people here thought the problem had already been corrected. Too bad I had to waste my time when someone else already had the answer.

The issue was addressed and corrected at Seti Beti on 14 Dec., however there are still ongoing issues that remain here at Seti main. My understanding is that once those in the know can pull themselves away from their current focus regarding the coding issues with the Android apps they will do what they can to address the remaining issue here.

There are only so many grey cells to go around. In respect to all those who do the heavy lifting we all need to have a bit more patience.

My self included!


Keifer

---

If the tasks are Expired instead of resent, there is a problem with the Server.

T-Bar, you're out of touch, we know there is a problem with the Server, it's been fixed at Seti Beta, just waiting for it to fixed here,
I'm sure when enough time has passed since applying the fix to Seti Beta, then it'll be applied here.

Claggy

Thanks for the Update. Obviously the people here thought the problem had already been corrected. Too bad I had to waste my time when someone else already had the answer.

The issue was addressed and corrected at Seti Beti on 14 Dec., however there are still ongoing issues that remain here at Seti main. My understanding is that once those in the know can pull themselves away from their current focus regarding the coding issues with the Android apps they will do what they can to address the remaining issue here.

There are only so many grey cells to go around. In respect to all those who do the heavy lifting we all need to have a bit more patience.

My self included!


Keifer

You tell people they need a little patience around here
and they get their nose outta joint.

Day 24.

I see a new batch of AP's are being split.

Any chance the corrections have been made?


Keifer

---