Jeff mentioned in the SETI@home forum, that the SETI@home Beta Test project will test that VLAR (AR < 0.12x) WUs will not send to GPUs.

Maybe a well idea to start a thread about experiences and problems.


With also without app_info.xml file my BOINC don't get work:
Message from server: No work sent
Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")

I have allowed only SETI@home Enhanced in the project preferences. CPU & GPU is allowed, because I can't change it (I wanted only GPU).


BTW. 'unable to change project preferences'

---

After reading an other thread, this message could come if only VLAR WUs are available.
But - I have CPU & GPU enabled (preferences & app_info.xml file).. the CPU could calculate this AR WUs.

---

Current..

If BOINC ask for CPU & GPU WUs, BOINC don't get WUs. (this is not well, he should get CPU WUs)
If BOINC ask only for CPU WUs, BOINC get WUs.
If BOINC ask only for GPU WUs, BOINC don't get WUs.

If no WUs, the upper mentioned message from the server.

---

BTW.

The server flops are enabled at SETI@home Beta Test?

Because I have still the flops in the app_info.xml file.

---

looking good so far
if my project prefs are set to just allow GPU tasks get error "no work for selected app"
Change prefs to allow both CPU and GPU tasks got only CPU tasks, all marked as .VLAR wu's

---

My BOINC ask also for GPU WUs, but get no.

Current there are only VLAR WUs available?

---

My BOINC ask also for GPU WUs, but get no.

Current there are only VLAR WUs available?

Would seem so
going to be one of the downsides for GPU crunchers - if theres a VLAR storm could be a few hungry GPU's out there
Think I saw a post at main by Joe that did say that the tapes being split here contained VLARs at the start then move into mid-range wu's - so should only be a matter of time before there are some more cuda tasks here

---

My BOINC ask also for GPU WUs, but get no.

Current there are only VLAR WUs available?

Would seem so
going to be one of the downsides for GPU crunchers - if theres a VLAR storm could be a few hungry GPU's out there
Think I saw a post at main by Joe that did say that the tapes being split here contained VLARs at the start then move into mid-range wu's - so should only be a matter of time before there are some more cuda tasks here

Also noted in the neighboring thread here that the first 2048 tasks are all VLAR, after that all from the same channel should be midrange.

There seems to be a limit of 1 task per request for CPU on the MB work, that's why those first 2048 have not yet all been sent. Getting close, though, a couple of minutes ago the task at position 2024 was sent. It'll be interesting to find out if GPU requests also only yield 1 task each. They've split Results ready to send 130,717, which is almost all new MB work. With that many, there are probably other patches of ".vlar" tasks.

                                                                 Joe

My BOINC ask also for GPU WUs, but get no.

Current there are only VLAR WUs available?

Would seem so
going to be one of the downsides for GPU crunchers - if theres a VLAR storm could be a few hungry GPU's out there
Think I saw a post at main by Joe that did say that the tapes being split here contained VLARs at the start then move into mid-range wu's - so should only be a matter of time before there are some more cuda tasks here

Also noted in the neighboring thread here that the first 2048 tasks are all VLAR, after that all from the same channel should be midrange.

There seems to be a limit of 1 task per request for CPU on the MB work, that's why those first 2048 have not yet all been sent. Getting close, though, a couple of minutes ago the task at position 2024 was sent. It'll be interesting to find out if GPU requests also only yield 1 task each. They've split Results ready to send 130,717, which is almost all new MB work. With that many, there are probably other patches of ".vlar" tasks.

                                                                 Joe

Both my E8500 and T8100 machines got more than one task per request,

Claggy

The last time (17 Jul 2010 6:45:56 UTC) my BOINC got new work, he got 20 VLAR WUs in a bunch for the CPU. He have still 8 VLAR WUs in the overview.
No GPU WUs.

He ask for GPU WUs, but get..

Something interesting, the messages changed:

17/07/2010 18:50:26 (UTC) SETI@home Beta Test Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")

to:

17/07/2010 19:01:02 (UTC) SETI@home Beta Test Message from server: No work sent
..
...
..
17/07/2010 20:35:27 (UTC) SETI@home Beta Test Message from server: No work sent

again to:

17/07/2010 20:40:16 (UTC) SETI@home Beta Test Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")

---

17/07/2010 23:10:59 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:11:19 SETI@home Beta Test Scheduler request completed: got 2 new tasks

17/07/2010 23:12:06 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:12:26 SETI@home Beta Test Scheduler request completed: got 3 new tasks

17/07/2010 23:12:36 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:13:06 SETI@home Beta Test Scheduler request completed: got 2 new tasks

17/07/2010 23:13:17 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:13:37 SETI@home Beta Test Scheduler request completed: got 2 new tasks

17/07/2010 23:14:19 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:14:39 SETI@home Beta Test Scheduler request completed: got 2 new tasks

17/07/2010 23:15:22 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:15:42 SETI@home Beta Test Scheduler request completed: got 4 new tasks

17/07/2010 23:15:52 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:16:28 SETI@home Beta Test Scheduler request completed: got 1 new tasks

17/07/2010 23:16:38 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:16:58 SETI@home Beta Test Scheduler request completed: got 5 new tasks

17/07/2010 23:17:09 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:17:49 SETI@home Beta Test Scheduler request completed: got 5 new tasks

17/07/2010 23:17:59 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:18:45 SETI@home Beta Test Scheduler request completed: got 5 new tasks

17/07/2010 23:19:47 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:20:07 SETI@home Beta Test Scheduler request completed: got 4 new tasks

17/07/2010 23:20:50 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:21:11 SETI@home Beta Test Scheduler request completed: got 7 new tasks

17/07/2010 23:21:21 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:21:42 SETI@home Beta Test Scheduler request completed: got 4 new tasks

17/07/2010 23:21:53 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:22:20 SETI@home Beta Test Scheduler request completed: got 7 new tasks

17/07/2010 23:23:02 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:23:23 SETI@home Beta Test Scheduler request completed: got 6 new tasks

17/07/2010 23:23:35 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:23:56 SETI@home Beta Test Scheduler request completed: got 0 new tasks
17/07/2010 23:23:56 SETI@home Beta Test Message from server: No work sent
17/07/2010 23:23:56 SETI@home Beta Test Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")

17/07/2010 23:32:53 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:33:13 SETI@home Beta Test Scheduler request completed: got 11 new tasks

17/07/2010 23:33:24 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:33:51 SETI@home Beta Test Scheduler request completed: got 5 new tasks

17/07/2010 23:34:02 SETI@home Beta Test Requesting new tasks for GPU
17/07/2010 23:34:29 SETI@home Beta Test Scheduler request completed: got 3 new tasks

[UTC + 02:00]


All GPU WUs have no .vlar ending.

---

Hi Joe

Reply to your message Message 39675 on next thread ( is this project up and runnig) will post here.

Just downloaded 40 vlar WU's and 8 AP 5.05 (August 11 deadline).

Will concentrate on the vlar WU;s.

Est time on host 23245 at the moment is 7+ hours. Running around 26% after 2 hours 48 miniutes.

Michael.

As I got a machine cranked up with Boinc 6.11.1, I "reset" the project, so it set DCF at 1.0 The VHAR's (AR 2.7) take about 5 min and the DCF is not reducing very quickly (estimated time to complete 1 hr 8 min). Initially it is geting GPU one at a time. As several completed it started requesting a few more at a time. As a I got two midranges (estimated time to complete 3 hr 48 min GPU).

So at this point the machine has no CPU work.

Regards

---

Thanks to Paul and Friends
Please consider a Donation to the Seti Project

...
There seems to be a limit of 1 task per request for CPU on the MB work, that's why those first 2048 have not yet all been sent. Getting close, though, a couple of minutes ago the task at position 2024 was sent. It'll be interesting to find out if GPU requests also only yield 1 task each. They've split Results ready to send 130,717, which is almost all new MB work. With that many, there are probably other patches of ".vlar" tasks.

                                                                 Joe

Both my E8500 and T8100 machines got more than one task per request,

Claggy

Thanks, and with one running stock while the other is anonymous platform that eliminates any possibility the slowdown was intentional. I had of course spot checked a couple of other hosts, just my luck they were only getting 1 at a time too.

Now the midrange work is going out to both CUDA and CPU hosts, so we're past the initial phase of the test and of course working up through the task list much faster.

                                                                 Joe

It's look like (I compared the estimate times and took two WUs) that all DLed GPU WUs are AR 2.77x and AR 0.44x.

---

It's look like (I compared the estimate times and took two WUs) that all DLed GPU WUs are AR 2.77x and AR 0.44x.

Makes sense that the transition from one observing project to the next would require a quick move to where the new observing run wants to look.

WU names starting 18no09aj.xxxxx.72, 18no09aj.xxxxx.481, 18no09aj.xxxxx.890, and 18no09aj.xxxxx.1299 are the VLARs, I guess the 2.77s start 18no09aj.xxxxx.1708 or maybe 18no09aj.xxxxx.2117 also, if the move took longer.

                                                                Joe

My BOINC got:

CPU 0.012x AR WUs: 18no09aj.30466.481.3.13.x.vlar_x

GPU 2.77x AR WUs: 18no09aj.30466.1708.3.13.x_x
GPU 0.44x AR WUs: 18no09aj.30466.2117.3.13.x_x

---

BOINC got inter alia:

CPU & GPU 0.44x AR WUs: 18no09aj.30466.2935.3.13.x_x


On my system the 'VLAR WUs not to GPU' work well.

:-)


It could be enabled at SETI@home.. ;-)

---

How work the new feature?

It's possible to enable this new feature at SETI@home after the current planned server outage?

---

Boinstats, using the data from July 21, indicates there are over 2400 hosts which have credit granted in the last week. Unfortunately that includes all the credit granted as the old work was cancelled, so it doesn't say how many hosts have actually provided data on the ".vlar" test. I expect there were enough to analyze how it worked, assuming they have time to do so.

                                                                Joe