This morning I know I had over 200 CUDA tasks queued up to be worked on over the next few days and now, this afternoon, they are all gone from the client! I either have the worlds fastest machine or someone snatched them from me.

What's the story here? I'm using BOINC version 6.6.28 with quad processors and a GTX 260. Now there is no work being passed from the mother ship. Any help is appreciated.

---

I see a ton of errored Results with this error

ERR_UNSTARTED_LATE -233

BOINC will abort tasks that have not started yet and are over the deadline.

I'm not sure why this would occur. are you running a VLAR killer app? That would kill WU's otherwise Lets hope someone has an idea what this means

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

All of the workunits have been "Aborted by user". See you Computer Page for more information.

I'm not sure why this would occur. are you running a VLAR killer app? That would kill WU's otherwise Lets hope someone has an idea what this means

Normally, the VLAR killer mod leaves a note in the Task Result to say it has killed it. I don't see such information in the results of the above computer.

Also looking at your computer, the benchmarks seem incrediably low for a Q6600!!

---

Not a one was aborted by me.

I see stuff coming in to be worked on and then they instantly turn to aborted. What's the story?

---

Not a one was aborted by me.

I see stuff coming in to be worked on and then they instantly turn to aborted. What's the story?

Oh, I think someone looked at the calendar and hit the "OK"-button instead of "Cancel"-button... Check the date and time on your computer...

check you boinc managers messages. perhaps theres a clue there. you might see that there is something missing
/edit

Oh, I think someone looked at the calendar and hit the "OK"-button instead of "Cancel"-button... Check the date and time on your computer...

I was thinking that as well after a while. WHy else would the PC think the work is being done after the return date

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

They are showing that they were all aborted by user as stated, the error code (0xffffffffffffff17) coming up with them is that they are all past deadline or would not report within deadline. Correct me someone if I am wrong in stating that.

There are quite a few vlar's about at the moment, had a few terminated by my optimized app with vlar killer.

---

Good catch. The date was changed to July 22 when I went there earlier today to look at a forward date and probably did exactly that. Not too smart.

---

About the performance of the q6600, I'm not too happy with it either. It is not overclocked. Am wondering what to do about that. Suggestions??

---

get the lunatics unified installer. That will get oyu optimized and running faster in no time

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

I don't think that'll help. Looking at tasks and projects for that computer, it seems SETI@home is only allowed on the GPU and the other projects on the CPU.

The benchmark-scores seems a bit low (about half speed) for this Q6600 when compairing to other Q6600's. Are you sure it's running full speed and not throttling back?

Not a one was aborted by me.

I see stuff coming in to be worked on and then they instantly turn to aborted. What's the story?

I notice you are also getting a few of these:-

<core_client_version>6.6.28</core_client_version>
<![CDATA[
<message>
- exit code -5 (0xfffffffb)
</message>
<stderr_txt>
SETI@home error -5 Can't open file
(work_unit.sah) in read_wu_state() errno=2

File: ..\worker.cpp
Line: 123


</stderr_txt>
]]>

This was due to a bug which is fixed in 6.6.36, where switched CUDA tasks get trashed because it looks for the result file before it is created. If you get enough of these going on you can quickly run down your CUDA cache.

Also your graphics driver is a bit old. 18585 seems to be the most stable/reliable I've come across so far although there is a newer one now available.

---

GPU Users Group

I don't think that'll help. Looking at tasks and projects for that computer, it seems SETI@home is only allowed on the GPU and the other projects on the CPU.

The benchmark-scores seems a bit low (about half speed) for this Q6600 when compairing to other Q6600's. Are you sure it's running full speed and not throttling back?

running the optimized app won't help? how so? True his pc may be getting throttled but that still doesnt address how an optimized app won't help please explain.

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

I don't think that'll help. Looking at tasks and projects for that computer, it seems SETI@home is only allowed on the GPU and the other projects on the CPU.

The benchmark-scores seems a bit low (about half speed) for this Q6600 when compairing to other Q6600's. Are you sure it's running full speed and not throttling back?

running the optimized app won't help? how so? True his pc may be getting throttled but that still doesnt address how an optimized app won't help please explain.

The true "optimised" apps - as distinct from minor CUDA tweaks mainly greared towards smoother running - are for CPU tasks only. If the OP doesn't run SETI tasks on his CPU, then the optimised apps are completely irrelevant: hence useless in his working environment, as stated.

Your benchmark @ stock:

Measured floating point speed 1108.92 million ops/sec
Measured integer speed 3263.38 million ops/sec


My benchmark at 2.79GHz:

Measured floating point speed 2746.09 million ops/sec
Measured integer speed 6073.32 million ops/sec


If this is a OEM PC, no overclocking for you. If you built it, check to see if the CPU is being throttled back due to settings in the BIOS. Can't remember what the 'feature' is that you want to disable, but I am sure someone will post it...

---

Clk2HlpSetiCty:::PayIt4ward

I don't think that'll help. Looking at tasks and projects for that computer, it seems SETI@home is only allowed on the GPU and the other projects on the CPU.

The benchmark-scores seems a bit low (about half speed) for this Q6600 when compairing to other Q6600's. Are you sure it's running full speed and not throttling back?

running the optimized app won't help? how so? True his pc may be getting throttled but that still doesnt address how an optimized app won't help please explain.

The true "optimised" apps - as distinct from minor CUDA tweaks mainly greared towards smoother running - are for CPU tasks only. If the OP doesn't run SETI tasks on his CPU, then the optimised apps are completely irrelevant: hence useless in his working environment, as stated.

I was unaware that he was only running CUDA. He made mention of the CPU and GPU. I assumed he was running both.

However using the lunatics unified installer would at least give him the VLAR killer which would save him some time on CUDA unfriendly WU's

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

I don't think that'll help. Looking at tasks and projects for that computer, it seems SETI@home is only allowed on the GPU and the other projects on the CPU.

The benchmark-scores seems a bit low (about half speed) for this Q6600 when compairing to other Q6600's. Are you sure it's running full speed and not throttling back?

running the optimized app won't help? how so? True his pc may be getting throttled but that still doesnt address how an optimized app won't help please explain.

The true "optimised" apps - as distinct from minor CUDA tweaks mainly greared towards smoother running - are for CPU tasks only. If the OP doesn't run SETI tasks on his CPU, then the optimised apps are completely irrelevant: hence useless in his working environment, as stated.

I was unaware that he was only running CUDA....

Look at the first line that has been preserved through all these quotes (emphasis added). -ShEm- did the research for us.

you'll note that was Shems comment not Peters. Peter asked for help in boosting performance. One could assume that hes looking to do more with seti by his comment about boosting performance. OCing will boost overall performance seti optimized apps will boost performance on seti.

simply put it's be easy to OC a q6600 to 3.2 GHZ. there are many sites that will describe how its done.

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

you'll note that was Shems comment not Peters. Peter asked for help in boosting performance. One could assume that hes looking to do more with seti by his comment about boosting performance. OCing will boost overall performance seti optimized apps will boost performance on seti.

simply put it's be easy to OC a q6600 to 3.2 GHZ. there are many sites that will describe how its done.

The larger issue is that Peter's Q6600 is not running as well as it could. Putting an optimized app on there would help his workunit times, but it won't solve the actual performance problem with the CPU.

Overclocking should only be suggested to advanced users, as should optimized apps.

---

About the performance of the q6600, I'm not too happy with it either. It is not overclocked. Am wondering what to do about that. Suggestions??

Have you had any luck with this?

Loking back over the posts, you seem to have had three issues.

1. Accidental time adjust killed all tasks.
2. New CUDA tasks aborted as soon as they arrived
3. Low benchmark results.

1. was caused by the time error.

2. the tasks I looked at at the time were all -5 errors (no result file). These were caused by a bug in the version of BOINC you are running which errors on CUDA tasks which are switched for whatever reason. In this case it was probably new download tasks taking precedence over an existing task. In this mode all new tasks coming in will get errored as soon as they arrive. This can be fixed by upgrading to 6.6.36 of BOINC.

3. Low benchmarks.
What make/model of motherboard are you running. It may be possible for us to look at the BIOS settings to see possible cause for your low values.

Just as a test I downgraded a 6.6.36 machine to 6.6.28 and it gave the same benchmark values so it isn't just a version oddity.

---

GPU Users Group