My nvidia run my monitor...

---

to shutoff the nvidia card?

I've got both of my nVidia 550 TIs disabled. I used Windows Device Manager and disabled both cards. No Crunchee for them.

I'll fire them up in The Winter when I want a Little More Heat.

LikeTheHeatOfDecember21st2012FourthAngel

---

Something does not know that you have over 5 days of cache. I don't see it, either. You've current got 138 Valid results, 365 Pending results, and 324 work units in progress. Of your in-progress tasks, you have 10 ATI tasks either running or waiting to run. You have 46 Pending ATI tasks waiting for validation.

I don't know where you found this data. It doesn't match what I see.
http://setiathome.berkeley.edu/host_app_versions.php?hostid=6738595
None the less 300 wu is 5 days worth of work for an 8800gts.
The 10 ati unit have been trying to download all night 2 made it and have been crunched and are already off my task list(uploaded).



Oh ok....on the hypothetical.

Still no one sees an issue with 11:1 nvidia/ati results. When the ati card is literally 5 times faster. Everything is fine? That isn't a red flag? Again 2 cards turned on,on the same day in the same system? 15 days ago long before this current outage.

Thats probably a point I should have made earlier...this system ran for 12 days before I posted. The ati card and Boinc and seti should be able to figure this faster than 2 weeks. Or atleast I should have seen some change over that time. Right?

I am beating the stuffing out of this horse, it was dead on Saturday, this just isn't making any sense to me.

---

[quote]
"Take the case of your asking for a 10 day cache of NVIDIA & CPU work, you have 8 days..... "
My cache was set at 2/2 days....Then I changed it as I was instructed too 3/.01days and now finally to 5/.01 days... and as I calculated and posted that que had over 5 days work in it yesterday. So I don't really know what your talking about.


I was being hypothetical and trying to use easy numbers without researching what your personal cache was set-at.

Something does not know that you have over 5 days of cache. I don't see it, either. You've current got 138 Valid results, 365 Pending results, and 324 work units in progress. Of your in-progress tasks, you have 10 ATI tasks either running or waiting to run. You have 46 Pending ATI tasks waiting for validation.

I'm not sure what problem you think you are having.

Nobody's caches are topped-up right now. We've just been through a multi-day outtage. It takes a while for the caches to fill.

"Okay... at least I'm trying to be honest. I gave you advice that I cannot follow, myself.

You might want to occasionally look in your "Transfers" tab and see if you have downloads that are stuck in queue; "Download retry in 05:42:03."

If so, select one and hit "Retry Now." .............."
I often find 10-20 downloads that I can prod into downloading...and the recent change from 2 at a time to 4 has really helped. This hasen't been my issue. As my system has downloaded hundreds in just the last few days.

"You don't understand, By deselecting 'Use Nvidia GPU' in your Setiathome Preferences doesn't stop Nvidia GPU computing, What it does is stop Boinc asking for Work for the Nvidia GPU, all existing Nvidia GPU work will carry on crunching as normal,

Once Boinc stops aking for ATI GPU work, re-enable Nvidia GPU work fetch,
(I assume you ment "once Boinc starts asking for ati...")
I understood the motivation behind this advise. I just don't think it will work. Sorry for my obstinence, I just(as previously stated) haven't seen anything to make me believe that time is gonna heal this problem. The whole 0 units completed is what grabs my attention. But hey, I'm here asking for help so if the concensous is shutdown the nvidia I gotta atleast try it.

"Take the case of your asking for a 10 day cache of NVIDIA & CPU work, you have 8 days..... "
My cache was set at 2/2 days....Then I changed it as I was instructed too 3/.01days and now finally to 5/.01 days... and as I calculated and posted that que had over 5 days work in it yesterday. So I don't really know what your talking about.

"I would simply disable nvidia GPU in your preferences and only fetch ATI for a while."

How?
There is no labeled controls in Boinc to control spacific gpus its all or nothing.
Online preferences? Didn't we disable those with the config file? What do I need to add to the file to shutoff the nvidia card?

---

Thank you all in advance.

Okay... at least I'm trying to be honest. I gave you advice that I cannot follow, myself.

You might want to occasionally look in your "Transfers" tab and see if you have downloads that are stuck in queue; "Download retry in 05:42:03."

If so, select one and hit "Retry Now." The danger is that it might lengthen the wait. If so, select one and hit "Retry Now" and repeat that process until something downloads or until the retry times cycle somewhat lower.

There's not much else you can do.

We regularly joke about wearing-out mice and seeking medical attention because we've hit retry to much and so often.

Seriously?(chuckle)

The solution to 1 video card not working is to disable the functioning 1? Yeah I don't think so.

Re-think it.

Apparently there is a hierarchy to work fetch requests.

Take the case of your asking for a 10 day cache of NVIDIA & CPU work, you have 8 days of NVIDIA work onboard, and you are out of CPU work.

BOINC keeps asking for NVIDIA work trying to fill the last two days. Often, and right now specifically, the servers can't provide enough work to fill your cache with 10 days of NVIDIA work.

The solution is to reduce the cache you are asking-for to, say, arbitrarily, 2 days. Now, your cache onboard is full x 4 for NVIDIA work and BOINC will stop asking for NVIDIA work and will only ask for CPU work since you have none.

Your BOINC stops sending "...fetch work for NVIDIA and CPU..." and starts asking only for CPU work.

So, instead of filling your download backlog with NVIDIA tasks, you'll ask for CPU tasks which improves the odds that any downloads that actually happen without stalling are CPU-tasks.

Because of fact that BOINC sort-of gets paralyzed by stalled downloading tasks, and there are so many of them, currently you aren't asking for CPU tasks as often as you could be if *all* you were asking-for were CPU tasks.

It's going to be the same with the ATI tasks. If you can stop BOINC from asking for CPU or NVIDIA tasks by telling it "I'm full," then it will only ask for ATI tasks and THAT will improve your odds of not-having a CPU or NVIDIA task stuck in the download queue preventing BOINC from asking for ATI tasks.

So, yes, telling BOINC you don't *want* anymore NVIDIA or CPU tasks really does improve your odds of getting ATI tasks in an almost accidental way.

BUT, BUT, BUT... ***right now*** I'm not sure that anything is going to do you much good for the very simple reason that there don't seem to be very good odds of getting *any* tasks to successfully download on the first attempt and not get stuck in queue, or be given absurd retry delay times or for the entire project to ask for a "back-off."

I am running two "mixed" machines. There may be an exception to this rule that I haven't documented, but my experience *seems* to be that if I run out of tasks I run out of ATI tasks first, then CPU tasks, then NVIDIA tasks **unless** it is a new installation, then the order changes to running out of ATI, then NVIDIA (because of their speed), then CPU. (I chalk this up to fewer WUs being scheduled for ATI crunching than CPU or NVIDIA, but I don't know that as a fact.)

The advice you've been given about "stopping the work fetch for the NVIDIA card" is true, correct, and valid **when things are working smoothly** which they obviously aren't right now.

I understand your desire to control this situation. I share it. You and I put-together can't control it. The best advice you've been given seems anti-intuitive, but is nevertheless true: Leave it alone. Walk away. Quit looking. EVENTUALLY the servers will "figure it out" and you will start getting lots of ATI tasks once your NVIDIA and CPU task cache is full.

You may get a "stray ATI task" here and there, now. But, as long as your download queue is full of stalled CPU/NVIDIA tasks it can't catch its breath. That's the idea behind asking it to stop fetching CPU/NVIDIA work. There is no assurance that it will get any ATI work even if you do that because there is no guarantee that there is any ATI work to give you.

I'm as obsessive as they come. My personal, non-technical, advice to you is to walk away and look at it again in a week or so.

We've had an outage. Obviously, we didn't recover smoothly. That could take a week to work itself out. Tomorrow we face a "regularly scheduled outage," and THAT usually takes a day or so to sort itself out once the project comes back up. Now, combine those things and you've got one mell of a hess. As frustrating as it is, I'm tellin' ya... set a reasonable cache for the project and "walk away."

You can expect a similar experience every time there is an outage or a database glitch, or any other excuse for an interruption (due to a limited ability to recover due to bandwidth and other issues) and this project experiences more than its share of power problems, barfing servers, bad ac, and inadequate staffing.

With things as they are right now, that is going to take days, not hours, for things to work-out to your and my satisfaction.

I've got two computers begging for ATI tasks and one begging for NVIDIA tasks. I've got idle cores on one because it can't get CPU tasks as fast as it crunches them. I've got downloads and uploads stuck in queue and machines that have "backed-off" retry times for hours and hours before they'll even try to send again so they can get delayed for even more hours at that point.

There is really nothing to be done but wait. You can "improve" the situation marginally and temporarily, but you and I cannot "fix it" from our end.

I'll repeat my blood-pressure advice: "Walk away. You can change it a little, but you can't fix it. Accept that and be patient."

Your hardware isn´t to shabby to be honest.
I would simply disable nvidia GPU in your preferences and only fetch ATI for a while.

Seriously?(chuckle)

The solution to 1 video card not working is to disable the functioning 1? Yeah I don't think so.

You don't understand, By deselecting 'Use Nvidia GPU' in your Setiathome Preferences doesn't stop Nvidia GPU computing, What it does is stop Boinc asking for Work for the Nvidia GPU, all existing Nvidia GPU work will carry on crunching as normal,

Once Boinc stops aking for ATI GPU work, re-enable Nvidia GPU work fetch,

Claggy

Your hardware isn´t to shabby to be honest.
I would simply disable nvidia GPU in your preferences and only fetch ATI for a while.

Seriously?(chuckle)

The solution to 1 video card not working is to disable the functioning 1? Yeah I don't think so.

---

I forgot that 'my ATI' is not shown properly by 'my BOINC' ;)
(AMD ATI unknown (1024MB) driver: 1.4.1646 == Cat 11.12 XP32)

It is AMD Radeon HD 6570

Do you have flops included in your appinfo BilBg ?

No
And the part for r390 is almost standard (the only change (for less lag): -period_iterations_num 80):

    <app>
        <name>setiathome_enhanced</name>
    </app>
    <file_info>
        <name>MB6_win_x86_SSE3_OpenCL_ATi_HD5_r390.exe</name>
        <executable/>
    </file_info>
    <file_info>
        <name>MultiBeam_Kernels_r390.cl</name>
        <executable/>
    </file_info>
    <app_version>
        <app_name>setiathome_enhanced</app_name>
        <version_num>610</version_num>
        <avg_ncpus>0.05</avg_ncpus>
        <max_ncpus>0.05</max_ncpus>
        <plan_class>ati13ati</plan_class>
        <cmdline>-period_iterations_num 80 -instances_per_device 1</cmdline>
        <coproc>
            <type>ATI</type>
            <count>1</count>
        </coproc>
        <file_ref>
            <file_name>MB6_win_x86_SSE3_OpenCL_ATi_HD5_r390.exe</file_name>
            <main_program/>                           
        </file_ref>
        <file_ref>
            <file_name>MultiBeam_Kernels_r390.cl</file_name>
            <copy_file/>
        </file_ref>
    </app_version>

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

Your hardware isn´t to shabby to be honest.
I would simply disable nvidia GPU in your preferences and only fetch ATI for a while.

---

With each crime and every kindness we birth our future.

BilBg....
I just want my video card to crunch data, currently it isn't. At this point the general consensus seems to be do nothing and it will fix itself. I am not seeing any indication on my end that this will work…I.E. a few more WU or some or any other kind of improvement.

I am one handed at the moment so I can’t really work on the hardware side of things. None the less I am researching which video card I will replace my ATI with when my hand is a little more functional. Since I can’t seem to get to the bottom of its lack of performance replacing it seems to be the path of least resistance.

The time it takes me to decide will provide a few more days for the problem to fix itself(I Wish).

GTX 480 or 2 gtx 460s hmmm my budget it really only about 125$, well my budget is really nothing. I am sure my wife would flip if she found out I was buying hardware to donate processing time....

---

Do you have flops included in your appinfo BilBg ?

---

With each crime and every kindness we birth our future.

Any idea how long I should wait?

For correct estimates (time remaining) of ATI tasks:
Until you see on this page:
http://setiathome.berkeley.edu/host_app_versions.php?hostid=6738595

... at least 10 'Number of tasks completed' for these:
SETI@home Enhanced (anonymous platform, ATI GPU)
AstroPulse v6 (anonymous platform, ATI GPU)

Note:
What you think you 'completed' is not the same what server shows as 'Number of tasks completed'
(those 'tasks completed' have to meet certain criteria)

no new ati wu have downloaded or are scheduled to attempt to download.

Does BOINC ask for ATI work? (look in Event Log)

I actually think ATI MB outier detection is Broken with the ATI r390 MB app, same as Fred J. Verster's ATI host in this thread:

Is it hosts or work erring out?

So all the ATI MB work will remain unscaled until it's fixed, Eric's just fixed a similar problem over at Seti Beta, and does know that similar problems are occuring here.

Claggy

For my ATI it works OK:
SETI@home Enhanced (anonymous platform, ATI GPU)
Number of tasks completed 1588
http://setiathome.berkeley.edu/host_app_versions.php?hostid=4832843

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

Average turnaround time 3.77 days (thats not true at all)

Do you know what 'Average turnaround time' is?

Number of tasks completed 0 (?! I watched it finish 30 units or so and uploaded them)
.........
...according to this I haven't completed any, which explains why I am not getting more work.
Why does it show me not completing any work?

Are you really surprised??
This was already explained ("'tasks completed' have to meet certain criteria" & "outlier detection"):
1) Task have to validate (BOINC have to upload and report the task, then your task will/may wait (seconds or months) for another person's same task's result - the 2 results have to be 'similar enough' to validate)
2) Task have to be not 'outlier'
('outlier': >10% blanking; result overflow (>30 'signals', also called 'early exit'))

Note: validated 'outliers' give credit; But they do not count in 'Number of tasks completed'

Max tasks per day 463 (I don't think this computer has even downloaded 100wu since it was built)(Are these stats for my account as apposed to this machine?)

This is for this particular machine+app.
It means you are allowed to get this many tasks of this type per day,
not that your BOINC will actually ask for so many nor it ensures tasks will be available at the moment it asks.

Number of tasks today 27 (also not true its been 1 set of 10)

Then you maybe have 'ghost' tasks (which will be resent to you later)

"Note: What you think you 'completed' is not the same what server shows as 'Number of tasks completed'
(those 'tasks completed' have to meet certain criteria)"
And whatever that criteria is I apparently am not meeting it.

Not you, that depend entirely by the tasks themselves (e.g. >10% blanking or too many signals).

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

Some interesting and note worthly results from : http://setiathome.berkeley.edu/host_app_versions.php?hostid=6738595

SETI@home Enhanced (anonymous platform, nvidia GPU)
Average turnaround time 3.77 days (thats not true at all)

SETI@home Enhanced (anonymous platform, ATI GPU)

Number of tasks completed 0 (?! I watched it finsih 30 units or so and uploaded them)
Max tasks per day 463 (I don't think this computer has even downloaded 100wu since it was built)(Are these stats for my account as apposed to this machine?)
Number of tasks today 27 (also not true its been 1 set of 10)
Consecutive valid tasks 363 (363 valid 0 completed)
Average turnaround time 0.00 days (this doesn't make anysense either)

"But I hasten to add, you aren't assured of keeping the NVIDIA busy, either.
...or the CPU for that matter."
Cureently my work que has:
40 CPU 6.03 WU
260+ Cuda 6.10 WU
2 AP CPU 6.01 WU

"aren't assured of keeping the NVIDIA busy"
oh yeah....at about 30 minutes each... that 5.4 days of work....

"So more ATI units you finnish so more you will get."
Well this makes a little more sense...according to this I haven't completed any, which explains why I am not getting more work. Why does it show me not completing any work?

"Note: What you think you 'completed' is not the same what server shows as 'Number of tasks completed'
(those 'tasks completed' have to meet certain criteria)"
And whatever that criteria is I apparently am not meeting it.

---

Any idea how long I should wait?

For correct estimates (time remaining) of ATI tasks:
Until you see on this page:
http://setiathome.berkeley.edu/host_app_versions.php?hostid=6738595

... at least 10 'Number of tasks completed' for these:
SETI@home Enhanced (anonymous platform, ATI GPU)
AstroPulse v6 (anonymous platform, ATI GPU)

Note:
What you think you 'completed' is not the same what server shows as 'Number of tasks completed'
(those 'tasks completed' have to meet certain criteria)

no new ati wu have downloaded or are scheduled to attempt to download.

Does BOINC ask for ATI work? (look in Event Log)

I actually think ATI MB outier detection is Broken with the ATI r390 MB app, same as Fred J. Verster's ATI host in this thread:

Is it hosts or work erring out?

So all the ATI MB work will remain unscaled until it's fixed, Eric's just fixed a similar problem over at Seti Beta, and does know that similar problems are occuring here.

Claggy

Any idea how long I should wait?

You have to wait for it to settle-out and even then you are not assured of keeping your ATI card busy.

But I hasten to add, you aren't assured of keeping the NVIDIA busy, either.

...or the CPU for that matter.

I have two computers, like yours, with mixed blessings in the PCIe slots, two sets of video drivers loaded, etc.

One of them ran out of ATI work. There were NVIDIA and CUDA WUs stuck in the download cache. I cleared them, BOINC asked for work, and BINGO!, I got some more ATI work, it downloaded, and it's off and running.

For how long? I don't know.

Another computer that used to be mixed but no longer is, used to get great gobs of ATI work and seemed never to be in danger of running-out.

In addition to BOINC's messing with us, there is, as has already been explained, a feast-or-famine situation with the availability of some types of ATI work.

So you can fiddle with the cache, or disallow certain work, or ask for voodoo priestess help, but sometimes there is no work to be had at any price.

The specific control you and I would like to have, "Hey, server! I need 40 ATI AstroPulse tasks over here!" doesn't exist.

So more ATI units you finnish so more you will get.

Like i said before disable CPU and nvidia in preferences so you should get only ATI work.
Increase work buffer til it settled.

---

With each crime and every kindness we birth our future.

Any idea how long I should wait?

For correct estimates (time remaining) of ATI tasks:
Until you see on this page:
http://setiathome.berkeley.edu/host_app_versions.php?hostid=6738595

... at least 10 'Number of tasks completed' for these:
SETI@home Enhanced (anonymous platform, ATI GPU)
AstroPulse v6 (anonymous platform, ATI GPU)

Note:
What you think you 'completed' is not the same what server shows as 'Number of tasks completed'
(those 'tasks completed' have to meet certain criteria)

no new ati wu have downloaded or are scheduled to attempt to download.

Does BOINC ask for ATI work? (look in Event Log)

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)