It appears to be broken in the current builds as well. Any quick and easy fixes you can offer would be appreciated. I think Petri's busy with the AutoCorr problem right now, but, it would be nice to have a working checkpoint. The Older x41p_zi App has basically been surpassed by the newer zi3 versions. The Older version is the best version of the Mac Special App available right now though, and it is listed as 'available for testing'. It also says, "See the Notes in the docs folder...".

Note 5) Restarted tasks could produce Incorrect Results.
Try not to resume suspended/paused tasks, if possible stop BOINC and reset active tasks in client_state.xml.

That's about as much as I can do.

That is, they run for a couple seconds, their Estimated Times may still have say 10 Min of crunching time left; BUT the Units "Finish" at the point of Resuming. In viewing Tasks on the Web, these Units immediately show up in Inconclusives.

TL

This means that checkpointing mechanism is broken in that build.

If app targets for high-end GPUs maybe it's OK to go w/o checkpoint at all. But in this case better to state this directly by state.sah write/read omission and clear exit with error state (computational error) in case app detects resume attempt.
Reporting invalid result is worst way.

---

SETI apps news
We're not gonna fight them. We're gonna transcend them.

Running the Baseline App 2 at a time... Now that's what I call S-L-O-W.
I just clear the Active GPU tasks from client_state.xml every time I restart, it takes about 15 seconds. Another method would be to just Suspend all the non-running GPU tasks when you want to Stop, once the GPU tasks finish, quit. Or best yet, just Don't stop crunching.

If you want to run the Much slower App, fine. If you want to remind yourself you made the right decision, just check these numbers every once in a while. It's a similar machine running similar cards, http://setiathome.berkeley.edu/results.php?hostid=7942417&offset=340

9-11-2016 at 12:51 PM - PDT, regular CUDA75, (same App as what is being tested at Beta), now restored to Andromeda. appconfig.xml restored to crunch 2 Units at a time per GPU.

5 New Tasks in immediate queue; more should DL in the next hour or so.


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

OK - Another Update:

Down to the last six Units from the boatload that got Downloaded. I've been noticing issues with Suspending for Time of Day and Resuming; that when SETI Resumes under the Special CUDA75 App that the Units that are to pick up crunching from where they left off are getting killed. That is, they run for a couple seconds, their Estimated Times may still have say 10 Min of crunching time left; BUT the Units "Finish" at the point of Resuming. In viewing Tasks on the Web, these Units immediately show up in Inconclusives.

I didn't seem to have this issue with the regular CUDA75 App, (the NON-Special App), so; when these last six Units complete today, I will be reinstalling the regular CUDA75 App. At which point; I will also restore my appconfig.xml file for Main to run 2 Units at a time per GPU.


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

Petri just had me complete a run of 19 Units at Beta on CUDA75 running just 1 Unit per GPU. To me, this seemed to run MUCH SLOOOOWWWWER than when running 2 per card. So, I just restored my original appconfig.xml for Beta, and next time will be running 2 Units per card again.

I hope this all helps out...


TL

[EDIT:]

Just resumed my final 6 SETI Main CUDA75 Units on the Special App.


TL

[EDIT 2:]

OOOPS... :-O

Thought I had No New Tasks set... Just Downloaded a boatload of new work.

On the Plus Side, it looks like the Special CUDA75 App is now completing these Units in less than 6 Min per Unit crunching 1 Unit per card!!! That's FRAKKING INCREDIBLE!!!!! :-)

Might just get my RAC boosted if I go back to crunching 15 Hours a day here. We'll see what happens with this new run of work.


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

https://setiweb.ssl.berkeley.edu/beta//forum_thread.php?id=2334#59501

Thanks for that, just posted over there.

btw: Crunching CUDA75 at Beta, I crunched 2 Units at a time per GPU. I crunched 40 CUDA75 Units.


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

Should be able to add my Mac Pro to the list over there in the next few days (probably Tbars build first, then my own after Windows is nailed down)

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

Since 5 September this host did not receive NV work:
http://setiathome.berkeley.edu/results.php?hostid=8018045&offset=80&show_names=0&state=0&appid=
seems plan class correction in place and working right.
Please confirm that.

---

SETI apps news
We're not gonna fight them. We're gonna transcend them.

https://setiweb.ssl.berkeley.edu/beta//forum_thread.php?id=2334#59501

---

SETI apps news
We're not gonna fight them. We're gonna transcend them.

Update:

Special App is running. "appconfig.xml" modification to v8 work has been successful. ONLY one Unit per card is being run. "appinfo.xml" modification is working, "CUDA75" is now being reflected in the Tasks Tab of BOINC.

Estimated Time to Complete each WU is about 36 Min. We will see what happens.

Thanks, again, TBar and Petri33. :-)


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

...I'd like to try the Special App; but, Petri indicated a couple of Posts ago that I NEED a command line. Now, you're saying I don't. If NO command line is needed and I can just install it straight, (like the CUDA65 and 75 Apps), I will endeavor to make the change to Special App instead of regular CUDA75 App.

Let me know how to proceed.

The Plan Class you see in the Manager is controlled by the entry in the app_info.xml. If you wish, you can change it to your name, it won't matter one bit as long as you don't have any existing tasks. Tasks are assigned to devices depending on your app_info.xml, once you have a task assigned it's best Not to change the app_info.xml version number or plan class.

Since you don't have any tasks, all you would need to do is change the <plan_class>cuda65</plan_class> to <plan_class>cuda75</plan_class>.

   <app_version>
        <app_name>setiathome_v8</app_name>
        <platform>x86_64-apple-darwin</platform>
        <version_num>802</version_num>
        <plan_class>cuda75</plan_class>
        <avg_ncpus>0.1</avg_ncpus>
        <max_ncpus>0.1</max_ncpus>
        <cmdline></cmdline>

That's it. Upon startup BOINC will read the app_info.xml and assign tasks to <version_num>802</version_num> and <plan_class>cuda75</plan_class>.

To run the 'Special' App just remove the files that came in the CUDA65 package and add the files from the 'Special' package, after changing the plan class entry in the app_info.xml. You really don't need to remove anything, the older app_info.xml will be overwritten by the new one and it won't matter one bit if the CUDA65 App stays in the setiathome.berkeley.edu folder. What matters is what your app_info.xml says, that is what BOINC reads to assign tasks and applications.

Just try to avoid any needless restarts, sometimes the Special App will Overflow a restarted task causing an Invalid.

OK - I've got the Special App installed. I've modified the appinfo.xml as you have stated; so, now it will show CUDA75 in Tasks Tab in BOINC. I've modified my appconfig.xml file to ONLY ALLOW 1 v8 Unit per card.

I will try this out tomorrow. Hopefully all goes well. (Presently crunching Einstein for the evening.)

Thank you for the assistance. I will report tomorrow of any issues.


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

...I'd like to try the Special App; but, Petri indicated a couple of Posts ago that I NEED a command line. Now, you're saying I don't. If NO command line is needed and I can just install it straight, (like the CUDA65 and 75 Apps), I will endeavor to make the change to Special App instead of regular CUDA75 App.

Let me know how to proceed.

The Plan Class you see in the Manager is controlled by the entry in the app_info.xml. If you wish, you can change it to your name, it won't matter one bit as long as you don't have any existing tasks. Tasks are assigned to devices depending on your app_info.xml, once you have a task assigned it's best Not to change the app_info.xml version number or plan class.

Since you don't have any tasks, all you would need to do is change the <plan_class>cuda65</plan_class> to <plan_class>cuda75</plan_class>.

   <app_version>
        <app_name>setiathome_v8</app_name>
        <platform>x86_64-apple-darwin</platform>
        <version_num>802</version_num>
        <plan_class>cuda75</plan_class>
        <avg_ncpus>0.1</avg_ncpus>
        <max_ncpus>0.1</max_ncpus>
        <cmdline></cmdline>

That's it. Upon startup BOINC will read the app_info.xml and assign tasks to <version_num>802</version_num> and <plan_class>cuda75</plan_class>.

To run the 'Special' App just remove the files that came in the CUDA65 package and add the files from the 'Special' package, after changing the plan class entry in the app_info.xml. You really don't need to remove anything, the older app_info.xml will be overwritten by the new one and it won't matter one bit if the CUDA65 App stays in the setiathome.berkeley.edu folder. What matters is what your app_info.xml says, that is what BOINC reads to assign tasks and applications.

Just try to avoid any needless restarts, sometimes the Special App will Overflow a restarted task causing an Invalid.

Is the Website current??? It's still NOT reflecting the 7 completed tasks I just recently reported; AND, I told SETI to Allow New Tasks again and picked up an 8th Unit that ALSO is NOT being reflected online as In Process...

8th Unit in Process still reflects as CUDA65; NOT CUDA75.

TL

Right now the delay is at Replica seconds behind master: 10,591 So, you will have to wait a while to see the tasks just completed.

Under Anonymous platform the Plan Classes don't matter. What matters is if you change the app_info.xml Plan Class while having Existing tasks you will create Ghosts unless you also change the client_state.xml entries to match the app_info.xml. For that reason the Plan Classes have been left the same. It will Not make any difference if it says 65 or 75. None, nada. So, if you must change it to 75 for some cosmetic reason, you need to wait until you don't have any existing tasks, then change the Plan Class in the app_info.xml to 75.

You really should use the 'Special' App with that machine. The CUDA75 App at C.A. is the Exact same Baseline version that is currently at Beta. It's just a little faster on the VLARs than the CUDA65 version. The 'Special' App at C.A. is the Older version x41p_zi which doesn't use any cmdlines. The 'Special' version at C.A. didn't have some of the previous problems present in x41p_zi3f and earlier, but is much faster than the Baseline version. You still need to run just ONE task at a time with the 'Special' version x41p_zi though.

Right now Petri's latest version x41p_zi3g seems Very Good. So far it appears the previous problems with the earlier version of x41p_zi3 have been solved. We'll have to wait and see how it works out.

!!!!!!!!!!!!!
You really don't need to do a thing, just run the tasks and don't abort anything. It doesn't matter One Hoot if the thing says CUDA65.

Thanks for the info. TBar. No Ghosts created this time. No SETI Tasks are currently in queue.

I just thought that, (with NO current queue), installing the CUDA75, (NOT Special), would state CUDA75 on the Tasks. (That's typically what the average Joe like me expects to happen when changing an App...) Since you say that it doesn't matter; I'll go ahead and reinstall the CUDA75 App.

I'd like to try the Special App; but, Petri indicated a couple of Posts ago that I NEED a command line. Now, you're saying I don't. If NO command line is needed and I can just install it straight, (like the CUDA65 and 75 Apps), I will endeavor to make the change to Special App instead of regular CUDA75 App.

Let me know how to proceed.


Thanks,


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

Is the Website current??? It's still NOT reflecting the 7 completed tasks I just recently reported; AND, I told SETI to Allow New Tasks again and picked up an 8th Unit that ALSO is NOT being reflected online as In Process...

8th Unit in Process still reflects as CUDA65; NOT CUDA75.

TL

Right now the delay is at Replica seconds behind master: 10,591 So, you will have to wait a while to see the tasks just completed.

Under Anonymous platform the Plan Classes don't matter. What matters is if you change the app_info.xml Plan Class while having Existing tasks you will create Ghosts unless you also change the client_state.xml entries to match the app_info.xml. For that reason the Plan Classes have been left the same. It will Not make any difference if it says 65 or 75. None, nada. So, if you must change it to 75 for some cosmetic reason, you need to wait until you don't have any existing tasks, then change the Plan Class in the app_info.xml to 75.

You really should use the 'Special' App with that machine. The CUDA75 App at C.A. is the Exact same Baseline version that is currently at Beta. It's just a little faster on the VLARs than the CUDA65 version. The 'Special' App at C.A. is the Older version x41p_zi which doesn't use any cmdlines. The 'Special' version at C.A. didn't have some of the previous problems present in x41p_zi3f and earlier, but is much faster than the Baseline version. You still need to run just ONE task at a time with the 'Special' version x41p_zi though.

Right now Petri's latest version x41p_zi3g seems Very Good. So far it appears the previous problems with the earlier version of x41p_zi3 have been solved. We'll have to wait and see how it works out.

!!!!!!!!!!!!!
You really don't need to do a thing, just run the tasks and don't abort anything. It doesn't matter One Hoot if the thing says CUDA65.

Due to the CUDA75 NOT reflecting as CUDA75; I stopped all processes and tasks. Aborted Tasks. Exited BOINC, went to Projects' Folder, removed CUDA75 info, reinstalled CUDA65 App. Reinstalled BOINC to reset preferences in MAC OS.

Will try to obtain work later.


TL

P.S. Website doesn't even show the Aborted tasks, yet... Don't know if this is because of Outrage Tuesday and it just isn't up to date yet, or what...

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

Eric implemented plan class restrictions. Please check if affected Macs now don't get OpenCL NV work.

How about the plan class restirictions: Do the affect my cuda? I have set my plan class as opencl_nvidia_sah even though I run cuda?
I love guppi work.

No.
You run under anonymous platform. So can set any plan class name doesn't matter.
This restriction made exclusively for "stock" app distribution.

Thank you!

---

To overcome Heisenbergs:
"You can't always get what you want / but if you try sometimes you just might find / you get what you need." -- Rolling Stones

Eric implemented plan class restrictions. Please check if affected Macs now don't get OpenCL NV work.

How about the plan class restirictions: Do the affect my cuda? I have set my plan class as opencl_nvidia_sah even though I run cuda?
I love guppi work.

No.
You run under anonymous platform. So can set any plan class name doesn't matter.
This restriction made exclusively for "stock" app distribution.

---

SETI apps news
We're not gonna fight them. We're gonna transcend them.

Hi,

if you're running TBar's 7.5 special for MAC ...
1) the tasks do not show yet in the setiathome results page.
2) Please do run one at a time
3) add -bs switch and -unroll 4 or -unroll 5 to your cmdline.
4) Check from the stderr that the version is zi3e or zi3f

Petri

p.s. and yes the do show up as 6.5 until the last downloaded before the change has been processed.

I'm NOT using the Special App, just the original CUDA75 App. My queue from WoW completed yesterday. Today I "Allowed New Tasks" and picked up 7 Tasks that now are all completed and Reported. I don't know why they're not showing up... The 7 tasks just completed ALL showed as CUDA65 in the Tasks Tab of BOINC 7.6.22.

I don't know how, or where, to put the command line you suggest. I don't know, nor understand "-bs" nor where to put "-unroll 4 or 5".

With the Units now Reported and gone, and as you say NOT showing on the Web Page, where do I go for Stderr???


TL

Since you are not running the special app you do not need to take any measures. Forget 1-4. Sorry.

The stderr output is in the setiathome results page under valid, pending and inconclusive labels. Just click the task.

Is the Website current??? It's still NOT reflecting the 7 completed tasks I just recently reported; AND, I told SETI to Allow New Tasks again and picked up an 8th Unit that ALSO is NOT being reflected online as In Process...

8th Unit in Process still reflects as CUDA65; NOT CUDA75.


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

Hi,

if you're running TBar's 7.5 special for MAC ...
1) the tasks do not show yet in the setiathome results page.
2) Please do run one at a time
3) add -bs switch and -unroll 4 or -unroll 5 to your cmdline.
4) Check from the stderr that the version is zi3e or zi3f

Petri

p.s. and yes the do show up as 6.5 until the last downloaded before the change has been processed.

I'm NOT using the Special App, just the original CUDA75 App. My queue from WoW completed yesterday. Today I "Allowed New Tasks" and picked up 7 Tasks that now are all completed and Reported. I don't know why they're not showing up... The 7 tasks just completed ALL showed as CUDA65 in the Tasks Tab of BOINC 7.6.22.

I don't know how, or where, to put the command line you suggest. I don't know, nor understand "-bs" nor where to put "-unroll 4 or 5".

With the Units now Reported and gone, and as you say NOT showing on the Web Page, where do I go for Stderr???


TL

Since you are not running the special app you do not need to take any measures. Forget 1-4. Sorry.

The stderr output is in the setiathome results page under valid, pending and inconclusive labels. Just click the task.

---

To overcome Heisenbergs:
"You can't always get what you want / but if you try sometimes you just might find / you get what you need." -- Rolling Stones