even without changing my boinc or boinc task setting

I didn't change any Settings in BOINC or BoincTasks to run SETI@home v8
I only changed the SETI@home apps by running Lunatics_Win32_v0.44_setup.exe
(I didn't go to stock first, I was using manually edited app_info.xml and now just installed apps from Lunatics_Win32_v0.44_setup)

I use old BOINC 6.10.58 and BoincTasks 1.61

I seen it mentioned about completing 10 or more tasks first,
But i did not understand why that was needed when updating seti 7 to seti 8.

This happens for any new app (not only at SETI@home) - that is how BOINC server Software is written.
When you go to using app_info.xml ("Anonymous platform", Lunatics) it is also considered "new app" by the server.

"Number of tasks completed 11" is needed only for more correct estimated time.
This new estimated time is calculated by the server and sent "together" with the next/new Downloaded tasks.

---

 


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

Hi BilBg
appreciate the help,

(I didn't go to stock first, I was using manually edited app_info.xml and now just installed apps from Lunatics_Win32_v0.44_setup)

I just left everything as it was after installing the apps.
i did not edit any files, (I wouldnt know what to edit)

---

• win7starter
  
• boinc: 7.14.2
  
• boinc tasks: 1.78
  
• Lunatics Win32 v0.44
  

 
OK, I see you experienced the problem I mentioned earlier (and you decided to Abort the tasks):

"The stock CPU app may hang at start on some CPUs/systems - happens randomly (in "Optimal function choice" - unless they changed that code recently - the issue exists for years)
When this hang occurs - the process uses 100% CPU core but there is no real progress (i.e. if the task is restarted it starts from 0%).
The (Lunatics) SSE3, SSE2,... apps don't have that "Optimal function choice" code.
"


Normal stderr_txt of stock CPU app (from another user):
http://setiathome.berkeley.edu/result.php?resultid=4685982280
... have this info:

WU true angle range is :  0.009687
Optimal function choices:
--------------------------------------------------------
                            name   timing   error
--------------------------------------------------------
                v_BaseLineSmooth (no other)
           v_avxGetPowerSpectrum 0.000074 0.00000 
                 avx_ChirpData_d 0.002942 0.00000 
                  v_pfTranspose4 0.001616 0.00000 
                JS AVX_a folding 0.001403 0.00000 

Flopcounter: 44675207566044.867000

I know Your tasks were hang because this info is partial.
This is exactly the place in code ("Optimal function choices:" at tasks starting) where/when the hang happens.
THIS HANG may be the reason you think SETI@home v8 tasks run too long!
(But this is nothing new - the bug was present in SETI@home v6 and v7 stock CPU applications)

See how "Optimal function choice" was hang in the middle
= no real % Progress despite "new" BOINC versions show increasing Progress even if it not exist - when the app don't report any Progress BOINC lies there is.
I don't know if they removed that misleading behaviour in newest BOINC code.

Your 2 hang tasks:

1) http://setiathome.berkeley.edu/result.php?resultid=4680143301

WU true angle range is :  0.011954
Optimal function choices:
--------------------------------------------------------
                            name   timing   error
--------------------------------------------------------
                v_BaseLineSmooth (no other)
     v_vGetPowerSpectrumUnrolled 0.000614 0.00000 
              sse3_ChirpData_ak8 0.013192 0.00000 

2) http://setiathome.berkeley.edu/result.php?resultid=4680143553
... have just this line (hang at very beginning of "Optimal function choice"):

WU true angle range is :  0.011954

This hang will Not happen with Lunatics CPU apps
 

---

 


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

hi BilBg,
So are you saying it was a bug in seti 8 as well as seti 6 & 7.?
or a bug in the stock boinc ?

As far as my tasks go, i only have 8 current tasks im working with.
(i know the website shows more)
I kind of messed up on the last lunatics app install (wrong one)
as a result it downloaded new tasks but then i changed the lunatics app again
after you suggested it and didnt pay attention to those tasks.
after i got the CUDA 3.2 app installed the tasks i had vanished from boinc.
and they were not aborted.
so like 22 tasks out of 30 that show in progress i dont have access to.
(not with current settings) so I dont know what to do about those.

Im going to try to complete the 8 that i do have.
although all of them show 19+ hours each.
I just hope its not all bad ones also.

---

• win7starter
  
• boinc: 7.14.2
  
• boinc tasks: 1.78
  
• Lunatics Win32 v0.44
  

So are you saying it was a bug in seti 8 as well as seti 6 & 7.?
or a bug in the stock boinc ?

What is "stock boinc"? Almost all use "stock" BOINC, very few use some custom version of BOINC
The hang bug is in stock CPU SETI@home applications v6, v7, v8

As far as my tasks go, i only have 8 current tasks im working with.
...
after i got the CUDA 3.2 app installed the tasks i had vanished from boinc.
and they were not aborted.

Every time you re-install Lunatics you should select again ALL the apps you want/need, not only "additional"
If you selected only CUDA app - this means no CPU applications in app_info.xml so BOINC will delete all CPU tasks without warning.

Do you have now CPU (SSE3) app running?
If not - re-install Lunatics (over the current) and select both SSE3 and CUDA apps

(not with current settings) so I dont know what to do about those.

No "Settings" can get them back to you, they will time-out and sent to others.

although all of them show 19+ hours each.

Don't pay attention to that, they are maybe CUDA tasks?
You never did CUDA SETI@home v8 by app_info.xml so this is "new app" as seen by server.

I just hope its not all bad ones also.

What are "bad ones"?? (There is no such thing)

The hang was NOT caused by some "bad" WU/task

---

 


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

hi BilBg,

What is "stock boinc"?

my mistake, i misunderstood.

Do you have now CPU (SSE3) app running?

Honestly, i think I do, but not sure, but i will look into that when i finish
the current running tasks. (ive selected not to receive anymore until i do)

The ones i am doing seem to be finishing faster than the displayed time.
so should not be to long to complete them, then i will try what you said.
(they should finish sometime tomorrow)

Was my mistake saying 'bad tasks',
I was making reference to the discussion with Ageless earlier in the post.
where he said

All of these have certain thresholds. Go over and the task drops out early.
They're not really bad tasks, that was a bit of an error from me, but they're tasks which have to be looked at by the project before they're resent to try again.

---

• win7starter
  
• boinc: 7.14.2
  
• boinc tasks: 1.78
  
• Lunatics Win32 v0.44
  

hi BilBg,
Ok, I installed the lunatics application over the old one as suggested.
I selected multibeam v8 (CPU)-includes legacy v7
and set it for SSE3.
(did not select any astropulse or whatever it is, settings)

Then I selected multibeam v8 (nv -cuda) includes legacy v7
and enabled process multibeam tasks on gpu (nv -cuda)
and selected lunatics_x41zi_win32_cuda32.

however this did not appear to make any difference.
everything seemed the same as before i did it.
(i think i already had the same settings and just overwrote it with the same)
i just have 6 tasks showing, all ending with .vlar_0 or .vlar_1
they are averaging 7 hours to complete even though the time shows as 19 hrs.
(i know its not accurate time, i only mentioned what it says)

---

• win7starter
  
• boinc: 7.14.2
  
• boinc tasks: 1.78
  
• Lunatics Win32 v0.44
  

Until you have 11 "normal" tasks completed AND VALIDATED on each of your processors (CPU and GPU) the "estimated" time for that processor will be WILDLY inaccurate, so IGNORE it. Your eyeballed runtimes sound about right, given that my Phenom is about twice as fast as your Athlon X2 and a typical VLAR takes between three oand four hours on that.

VLARS (tasks ending with "vlar_x") do not count to this figure, and frequently take very long run times. Let them run to completion and report.

As Nvidia cards do not like VLARs, and so they are not sent by the server, it is possible that you won't receive any tasks for your GPU until such time as "normal" tasks become available - and that, sadly, is a bit of a lottery.

The availability of a given type of task depends on what the telescope was doing when the data was collected, if it was looking at one point in the sky then we get lots of VLARs.

In the interim could you post the first thirty lines of your BOINC event log, as this will show if your GPU is being detected or not.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Hi Rob,
Yes i know, I was just relaying the results of what i did to BilBg.
i reinstalled the lunatics app as he mentioned since i wasnt sure if i selected
the SSE3 or not the 1st time.

I did not know that about

Nvidia cards do not like VLARs

Here is the 1st 30 lines, from boinc tasks,because of all the spaces etc
i tried to edit it to better fit here, but it is all 30 lines.

cc_config.xml not found - using defaults
Starting BOINC client version 7.6.22 for windows_intelx86
log flags: file_xfer, sched_ops, task
Libraries: libcurl/7.45.0 OpenSSL/1.0.2d zlib/1.2.8
Data directory: F:\Seti@home\data
Running under account Administrator
CUDA: NVIDIA GPU 0: GeForce 9200 (driver version unknown, CUDA version 2.2,
compute capability 1.1, 254MB, 219MB available, 29 GFLOPS peak)

Found app_info.xml; using anonymous platform
Host name: oblivion
Processor: 2 AuthenticAMD AMD Athlon(tm) 64 X2 Dual Core Processor 5000+
[Family 15 Model 107 Stepping 2]

Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush
mmx fxsr sse sse2 htt pni cx16 syscall nx lm svm rdtscp 3dnowext 3dnow

OS: Microsoft Windows XP: Professional x86 Edition, Service Pack 3, (05.01.2600.00)
Memory: 2.25 GB physical, 5.47 GB virtual
Disk: 243.27 GB total, 153.43 GB free
Local time is UTC -5 hours
VirtualBox version: 5.0.10
URL http://setiathome.berkeley.edu/; Computer ID 7609404; resource share 100
General prefs: from SETI@home (last modified 26-Jan-2016 23:24:10)
Computer location: home
General prefs: no separate prefs for home; using your defaults
Reading preferences override file
Preferences:
max memory usage when active: 1841.98MB
max memory usage when idle: 1957.11MB
max disk usage: 15.00GB
suspend work if non-BOINC CPU load exceeds 30%
(to change preferences, visit a project web site or select Preferences in the Manager)
task 04se15ab.4037.13155.6.33.112.vlar_1 resumed by user
task 01no15ab.1386.6207.14.41.177.vlar_1 resumed by user
Message from task: 0

---

• win7starter
  
• boinc: 7.14.2
  
• boinc tasks: 1.78
  
• Lunatics Win32 v0.44
  

Do you have now CPU (SSE3) app running?

Honestly, i think I do, but not sure, ...

It is easy to see, I give you 3 ways:

1) in Windows Task Manager or Process Explorer look which processes run
Process Explorer shows processes as a tree so you'll easy see what was started by BOINC (boinc.exe)
https://technet.microsoft.com/en-us/sysinternals/processexplorer.aspx

2) go to SETI@home directory (<BOINC_Data>\projects\setiathome.berkeley.edu\)
Your is at F:\Seti@home\data\projects\setiathome.berkeley.edu\
Look the names of .exe files - Copy/Paste the names here if you can't understand which is for what

3) in the same directory look for app_info.xml
Copy it somewhere (to not damage the original accidentally)
Open the Copy with Notepad
Ctrl+F for .exe
F3 for next

This way you'll see all the .exe files listed in app_info.xml

however this did not appear to make any difference.
everything seemed the same as before i did it.
(i think i already had the same settings and just overwrote it with the same)

Yes. probably you "already had the same settings"

I did not know that about

Nvidia cards do not like VLARs

The VLARs are currently not sent to any GPU - they make the computer lag too much or even driver restarts
(On "SETI@home Beta" (another project) they are sent to (some?) GPUs for test purposes)

CUDA: NVIDIA GPU 0: GeForce 9200 (driver version unknown, CUDA version 2.2,
compute capability 1.1, 254MB, 219MB available, 29 GFLOPS peak)

(Please don't edit next time, post the log as it is)

Uh Oh - CUDA version 2.2
You have NVIDIA driver less than 190.38 ? Or your GPU really supports only up to CUDA 2.2 ?

No need to change the driver if you like it

If you want to change it - go for 266.58 which is good for old GPUs
Like my: NVIDIA GeForce 8400 GS (511MB) driver: 266.58
http://setiathome.berkeley.edu/show_host_detail.php?hostid=7396088

I didn't check if your GPU is supported by 266.58
Older good versions are 190.38 or 191.07

But you need to change or disable CUDA app with the driver you use now!
The oldest version in Lunatics_Win32_v0.44_setup.exe is CUDA 2.3
It say it needs "min. driver 190.38"

You may:
1) upgrade NVIDIA driver

or
2) deselect "Use NVIDIA GPU" on SETI@home preferences

or
3) re-install Lunatics and select only SSE3 CPU app

or (Advanced)
4) get cuda22 app from SETI@home server and edit app_info.xml manually to add it (not really easy to do if you never edited app_info.xml)

EDIT: cuda22 app exists only for SETI@home v7, not for SETI@home v8:
http://setiathome.berkeley.edu/apps.php

---

 


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

Hi BilBg,
the VLARs might of been left from other tries to get lunatics installed,
But i did have 8 or more of them i did complete a few
but also I aborted a few, as you can probably see by looking at my tasks.

Normally ive only done CPU tasks as mentioned earlier in this post.
ive never knowingly selected to do GPU tasks.
so usually i would not select anything that was obviously for GPU.

Uh Oh - CUDA version 2.2

dont forget as my signature shows, my acer aspire x1200 pc, is kind of custom.
as it does not originally support windows xp, i think thats why the
version is like 2.2 i believe it was a universal driver compatible with xp.
so i think its best to try to make due with what i have until i decide to upgrade my windows.

Also note, i think i made a few posts when i 1st joined seti about issues i had
with tasks related to the same cuda driver thats why i never selected it again,
and only ran CPU tasks.

But you need to change or disable CUDA app with the driver you use now!

The 'Use NVIDIA GPU' is already unselected in SETI@home preferences,
I never selected it before,
but it was selected in the lunatics app when i installed it.
(just never selected on the website) only 'use cpu' is selected there.

Also, I just recieved a bunch of new tasks. 44 in all. (says 9 days worth)
i hope that is just an estimate or something is broken.
in my boinc tasks options i have selected
minimum work buffer = 2.00 days
additional work buffer= 3.00 days
that should be 5 days worth of tasks not 9.

i was just curious about that considering all the other time variations.
its probably nothing but thought id mention it.

---

• win7starter
  
• boinc: 7.14.2
  
• boinc tasks: 1.78
  
• Lunatics Win32 v0.44
  

You have been told on a number of occasions to ignore the estimated times and the allow BOINC Manager do it job.

The estimates will be wildly wrong until you have had 11 normal tasks competed and validated, so just let tasks run to completion, watch the elapsed time if you must watch something.

You will find that v8 tasks take longer to run than v7 tasks did. This is because v8 does more calculations than v7 did.

The way the cache system works is far from obvious - it is not uncommon to get more work than one would expect, because the servers, where the actual work allocation takes place take one "guess" at the processing time, and our PCs are displaying a different "guess". I've found it far better to set a longish minimum work buffer (5 or 6 days) and a very short (0.1 days) additional work buffer - this way I get a constant drip feed of tasks rather than waiting until the cache is nearly empty and getting a large number - which is exactly what you have seen happen.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

hi rob,
1st of all i was not addressing you. if you saw the reply it was to BilBg.
2nd i am not stupid, i have said quite a few times now if you read my replies
that i understood about times not bieng right and so on.
yet you keep hammering me about it even when i only reference it.
i do appreciate the advice and help, but not the hype.

---

• win7starter
  
• boinc: 7.14.2
  
• boinc tasks: 1.78
  
• Lunatics Win32 v0.44
  

Also, I just recieved a bunch of new tasks. 44 in all. (says 9 days worth)
i hope that is just an estimate or something is broken.
in my boinc tasks options i have selected
minimum work buffer = 2.00 days
additional work buffer= 3.00 days
that should be 5 days worth of tasks not 9.

The work request depends on a couple of things:
1. the amount of days in the minimum buffer settings.
2. the amount of days in the additional buffer settings.
3. the percentage for the host details of Fraction of time BOINC is running.
4. the percentage for While BOINC is running, fraction of time computing is allowed
5. the percentage for While is BOINC running, fraction of time GPU computing is allowed (if applicable and a GPU is being used).
6. the estimates that tasks have that they should run for and the correction that BOINC has calculated that should be put on them.

While min 2 + add 3 may be 5 days of work, the combination of all the above may cause BOINC to seemingly over fetch. I would wait it out.
Even when those 11 tasks have run, estimated run times are still being recalculated. You and BOINC have more of a good over view after running several hundreds of tasks.

And then, not all tasks are the same length. Low angle range tasks run longer, high angle range tasks run faster. How much slower or faster depends on your hardware, its capability to set back floating point operations. Any GPU will always be faster than a CPU, because the GPU uses all its internal cores to run the task, while the CPU only runs one task per one core.

---

thanks Ageless,

i just wasnt sure if it might of been an issue or not.
just thought i would mention it.

i was just wondering about install paths.
if you noticed boinc is not installed on my windows drive.
im not sure if thats a good/bad thing.
i dont think it would really matter, or does it?

---

• win7starter
  
• boinc: 7.14.2
  
• boinc tasks: 1.78
  
• Lunatics Win32 v0.44
  

My BOINC programs directory is installed on partition P:, half of my third physical 1TB hard drive. The Data directory is on my old but sturdy 500GB second drive, partition M:

So no, all that doesn't really matter in this.

---

Any GPU will always be faster than a CPU

Well ... not "Any GPU" but most GPUs

NVIDIA GeForce 8400 GS (my) is almost the same speed (in fact slightly slower) as one core of Intel(R) Pentium(R) Dual CPU E2140 @ 1.60GHz (in the same computer) for SETI@home tasks:
http://setiathome.berkeley.edu/show_host_detail.php?hostid=7396088
http://setiathome.berkeley.edu/results.php?hostid=7396088&offset=0&show_names=0&state=4&appid=

I think it is the second version based on G98GS (I may check again but don't want to wake the computer now)
https://en.wikipedia.org/wiki/GeForce_8_series#Technical_summary

I think GeForce 9200 will be about the same speed
https://en.wikipedia.org/wiki/GeForce_9_series#Technical_summary

---

 


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

Normally ive only done CPU tasks as mentioned earlier in this post.
ive never knowingly selected to do GPU tasks.
so usually i would not select anything that was obviously for GPU.
...
The 'Use NVIDIA GPU' is already unselected in SETI@home preferences,
I never selected it before

Then you had me confused by this:

"

Your GPU (NVIDIA GeForce 9200 (253MB)) is probably too slow and low RAM to be of real use (but you can try CUDA 3.2 app if you wish).

well i dont know about my NVIDIA being slow, probably is compared to some of the rigs out there, i just know in seti v7 i had no trouble like this. all the tasks were running smooth, ...
"

(From post: http://setiathome.berkeley.edu/forum_thread.php?id=78853&postid=1759749#1759749)
... which I interpreted as if you "had no trouble" running SETI@home v7 on your NVIDIA


All this means - it was useless (but harmless) to select any CUDA app in Lunatics
Since you don't accept NVIDIA tasks this CUDA app will just stay unused for now (no need to re-run the Lunatics installer)
CPU tasks/app is not influenced in any way - no matter if you have (or have not) installed also CUDA app

---

 


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

Hi BilBg
Im getting a general understanding between GPU & CPU tasks.

When i said

"I had no trouble" running SETI@home v7 on this NVIDIA

Honestly, i recall i did have trouble at first when i 1st installed seti v7.
which i believe you same guys helped me with those ones as well.

But once everything was working ok i had no further problems with it.
(until now when i changed to v8)

But anyway everything seems to getting back to kind of 'normal' for me now.
thanks for the help again.

---

• win7starter
  
• boinc: 7.14.2
  
• boinc tasks: 1.78
  
• Lunatics Win32 v0.44