My cache is still full but it takes up to 7 requests here.
Got a batch of 33 GPU tasks one hour ago.

---

With each crime and every kindness we birth our future.

It has been over 4 hours ago when I got two Nvidia GPU wu's... and they were done under 20 minutes... After that, 25 requests and no tasks...

yup no gpu work since hours

---

I came down with a bad case of i don't give a crap

Upped my cache level on my T5500 to pull in some VLARs, i think all but 5 tasks are VLARs, will do the non-VLARs first, then resends,
with 100 tasks on board this is about two weeks work, have set NNT for now.

Edit: Did the same on my T8100, all it's picked up so far are VLARs.

With the MB v7 tasks ready send close to 400k, the splitters are eithier still going or there are a lot of Wu's timing out, or becoming inconclusive.

Claggy

Roll on the release of the OpenCL application for VLARs on GPUs.

Thinking aloud - would it be possible to make that application a special application only for use against VLARs? This might overcome the performance hit that was being suffered in Beta when using the application for normals and shorties.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Upped my cache level on my T5500 to pull in some VLARs...

Did the same on an i5, and topped up to 100 tasks exclusively with first-run VLARs.

We're currently working through the (many) tapes recorded on 16 May: there was some interesting astronomy going on that day, according to the Arecibo Observatory Telescope Schedule.

http://www.naic.edu/vscience/schedule/tpfiles/MinchintagA2048tp.pdf
http://www.naic.edu/vscience/schedule/tpfiles/LorimertagA2854tp.pdf
http://www.naic.edu/vscience/schedule/tpfiles/KaspitagP2030tp.pdf
http://www.naic.edu/vscience/schedule/tpfiles/DenevatagP2859tp.pdf

(though possibly of more interest to Einstein than to SETI)

Roll on the release of the OpenCL application for VLARs on GPUs.

Thinking aloud - would it be possible to make that application a special application only for use against VLARs? This might overcome the performance hit that was being suffered in Beta when using the application for normals and shorties.

I know they asked that question about a week ago. I think there still wasn't enough ATI GPU users testing it to give them a yes or no as to if it would work. Anyone have any further information?

---

It has been over 4 hours ago when I got two Nvidia GPU wu's... and they were done under 20 minutes... After that, 25 requests and no tasks...

Suddenly I did get 50 tasks for Nvidia, after that, zero for several times

Yeah, seems like we caught a break. A bunch just downloaded onto 1 machine. The other got a dozen or some more.

I've just been idly wondering why I have so many validated tasks still showing on my list. And found the cut-off point:

Task 4294795247 --> WU 1858492176 (just a slow wingmate, all good)
Task 4294972138 --> WU 1858575073 (Unable to handle request, can't find workunit)

Anyone recognise the value 4,294,967,296 - between those two task IDs? Yup, 2^32. Oops.

Edit - Task 4294967296 itself has been validated, but is in the "can't find workunit" state.

Task 4294972138 --> WU 1858575073 (Unable to handle request, can't find workunit)

JBird reported 8 of those in Q&A.

---

I've just been idly wondering why I have so many validated tasks still showing on my list. And found the cut-off point:

Task 4294795247 --> WU 1858492176 (just a slow wingmate, all good)
Task 4294972138 --> WU 1858575073 (Unable to handle request, can't find workunit)

Anyone recognise the value 4,294,967,296 - between those two task IDs? Yup, 2^32. Oops.

Yes, the data base purge seems to have stopped right after they changed the size of the WU names. Maybe they forgot to change the purge program to match?

---

Donald
Infernal Optimist / Submariner, retired

I'd reckon the file deletion program is returning an error for result files associated with a ResultID >= 2^32, accounting for the line

Result files waiting for deletion 7,847,326

on the server status page. And if the result file can't be deleted, the result status in the database can't transition to "ready to purge". So both the file storage area on the server disks is filling up (slowly, these are small files), and the table rowcount in the database is growing inexorably. Better put that on their ToDo list for Tuesday.

The dB is so gonna explode if it keeps growing like that

---

I came down with a bad case of i don't give a crap

Yeah, seems like we caught a break. A bunch just downloaded onto 1 machine. The other got a dozen or some more.

That was short one, again only VLAR's. My 560Ti got two tasks, and they are completed in 10 minutes...

I hope that those 21(2/3)my tapes will have more non-VLAR's... if not, then is time to panic :)

EDIT: As I was typing, I got 14 GPU tasks from 16my15am -tapes

The dB is so gonna explode if it keeps growing like that

Both David and Eric have acknowledged my report - they're on the case.

I reckon the dB can hold out until Tuesday - we've been higher than this, with the WU table exploding too, and lived to tell the tale ...

I'd expect they have some script running that warns them when magic number N about the amount of results/tasks waiting for 'X' is passed and that it then plays a trumpet at them. Perhaps that the sound's been turned off.

---

I'd expect they have some script running that warns them when magic number N about the amount of results/tasks waiting for 'X' is passed and that it then plays a trumpet at them. Perhaps that the sound's been turned off.

I think we are the default trumpeters ;)

I just sent in a report that the security of the web sites (BOINC and Seti) is a bit lacking, in that Poodle can attack, the certificate is going to be blocked next year and such fun things. I think that with stuff like that they turn me up. ;-)

---

Roll on the release of the OpenCL application for VLARs on GPUs.

Thinking aloud - would it be possible to make that application a special application only for use against VLARs? This might overcome the performance hit that was being suffered in Beta when using the application for normals and shorties.

There is a CPU App sitting at Beta right now that would basically Double the number of VLARs processed per hour in OSX. Right now the FLOPS reading is;
Mac OS X/Intel 7.00 29 May 2013, 21:14:00 UTC 30,342 GigaFLOPS
That number would Double as soon as the CPU App is released, that's a Lot of VLARs. Yet, the App just sits at Beta even though not a Single Machine has had a problem with it in the Months it's been there.

Strange isn't it...