Please report problems in this thread.

I'm assuming the previous settings for hardware and memory limits still apply.

---

Please report problems in this thread.

I'm assuming the previous settings for hardware and memory limits still apply.

Yes. Support for BOINCs suspending features is now implemented and can be tested.

---

_\|/_
U r s

I'm in: http://setiweb.ssl.berkeley.edu/beta/show_host_detail.php?hostid=63614

60 wus done for AMD 6.07 AP app and counting. 12 valid 48 pending 0 inconclusive.
http://setiweb.ssl.berkeley.edu/beta/results.php?hostid=51991

---

_\|/_
U r s

I'm in: http://setiweb.ssl.berkeley.edu/beta/show_host_detail.php?hostid=63614

Betriebssystem Linux 2.6.32-358.6.1.el6.x86_64
BOINC Version 6.10.45

Don't know why you didn't get any work for NV GPU so far, probably the BOINC version is too old.

---

_\|/_
U r s

That's the reason. OpenCL isn't properly detected by BOINC 6. In some cases it's possible to use driver revisions on NVIDIA to tell whether OpenCL is supported, and I may add a change to allow that this week. Unfortunately, BOINC 6's info reported for ATI/AMD cards is too limited to allow something similar without causing too many problems (hangs, driver resets, bluescreens) for people who have no OpenCL support or borken OpenCL support.

---

I'm not sure that BOINC 7 and CentOS is feasible combo.
Had enough troubles with BOINC istallation under this OS already.
So I would prefer to stay with already running one and will try to update only if it's needed.

That's the reason. OpenCL isn't properly detected by BOINC 6. In some cases it's possible to use driver revisions on NVIDIA to tell whether OpenCL is supported, and I may add a change to allow that this week. Unfortunately, BOINC 6's info reported for ATI/AMD cards is too limited to allow something similar without causing too many problems (hangs, driver resets, bluescreens) for people who have no OpenCL support or borken OpenCL support.

So? Will you add BOINC v6 for distribution of OpenCL NV AP or not? All CUDA are OpenCL too. Only driver version should be not too old (263.06 and up). And driver version can be aquired via CUDA APU, OpenCL not required for this, right? So I think NV AP on BOINC 6 hosts could be safe enough. And very needed cause we still have no MB NV for Linux...

Yes, I'll do that.

---

Found at least one NV host running Linux (Ubuntu) that is successful using the new AP6.07 OpenCL app. He's got a 450GTS and uses a driver 313 and BOINC 7.0.27

My hopes to see MORE TESTERS for this NV linux apps has been not yet fulfilled.

---

_\|/_
U r s

yeah...
http://setiweb.ssl.berkeley.edu/beta/gpu_list.php "no GPU reported" for Linux...

I may have to look at the source code for that page. I think it's probably grepping through the log to find those, and hasn't be updated in a while.

---

Better to just add BONC 6 as distribution target and look into Intel GPU issue instead.
This statistics page of no real importance though could be useful indeed.

Still no work for NV AP on Linux under BOINC 6

Received 30 tasks on this host and 28 finished after just a few seconds with the message "Blanking too much RFI? percent blanked: 100.00"

---

Received 30 tasks on this host and 28 finished after just a few seconds with the message "Blanking too much RFI? percent blanked: 100.00"

7 of those already validated ok so it's probably just what the message says

Received 30 tasks on this host and 28 finished after just a few seconds with the message "Blanking too much RFI? percent blanked: 100.00"

7 of those already validated ok so it's probably just what the message says

It's the B3_P1 channel as usual. If we can ever figure out how to get enough participants to contribute cash, one of the staff may be able to go to Arecibo and troubleshoot that intermittent hardware problem. Then at least data recorded after the fix would not be corrupted.

                                                                   Joe

Linux NV AP:

on this host http://setiweb.ssl.berkeley.edu/beta/results.php?hostid=63614&offset=0&show_names=0&state=0&appid=14 11 eligible valids already done, no invalids so far. Currently even no inconclusives.

Looks like app can be released.

Linux NV AP:

on this host http://setiweb.ssl.berkeley.edu/beta/results.php?hostid=63614&offset=0&show_names=0&state=0&appid=14 11 eligible valids already done, no invalids so far. Currently even no inconclusives.

Looks like app can be released.

Just your runtimes are somehow slow. Maybe it is because of your setup/driver install or BOINC usage as root.

When looking through the Linux hosts result list here at Beta, inconclusives seem to be extremely seldom with the current app-versions.

So, releasing the app has my vote.

---

_\|/_
U r s

I have three inconclusives:
5416321
5311674
5416162