29/12/2008 20:16:20|SETI@home|Finished download of setiathome_6.06_windows_intelx86__cuda.exe

;)

Thanks to developers and testers!

S@H->NNT, new WUs were manually suspended, set to crunching old ones...

We'll see...

Windows/x86 - V6.06 - since 29 Dec 2008 17:05:48 UTC

It's now bug-free?

---

from what ive read at beta, no.

---

from what ive read at beta, no.

Maybe less buggy than V6.05?

---

same as on setibeta: the files are binary the same

---

Matthias

Somebody run an MD5 or SHA1 on 6.05 and 6.06 and see if they're the same.. :p

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

29/12/2008 20:16:20|SETI@home|Finished download of setiathome_6.06_windows_intelx86__cuda.exe

Can anybody explain why some people download CUDA-capable applications automatically, while others do not no matter how they try?

I think it has something to do with the other fundamental question: why some who DID dowload the CUDA application are no longer capable of crunching WUs with it, while others are.

I am proud to belong to both categories: I used to have a working CUDA installation that crunched flawlessly for a couple of days and then stopped. In the effort of having it working again I uninstalled BOINC altogether; when I reinstalled it, my machine dowloaded the normal Multibeam CPU application and now it does not even try to get the CUDA-enabled version.

Is there any sensible explaination for that?

---

You could always try a manual download

BOINC

Bernie

---

Can anybody explain why some people download CUDA-capable applications automatically, while others do not no matter how they try?

Explained here

when I reinstalled it, my machine dowloaded the normal Multibeam CPU application and now it does not even try to get the CUDA-enabled version.

Is there any sensible explaination for that?

Depends what version you re-installed. 6.4.5 is the version recommended for it to work with. I can't tell which one you have since your pc's are hidden. If you do have 6.4.5 and Vista with it installed as a service, that could be the cause too. You would have to unistall it and re-install it as anything but a service.

If this doesn't apply to your case, it would be helpfull to include the version and OS with the drivers you're using, or even a link to the pc itself.

---

Explained here

I see. But if http redirects were the cause, I should experience download errors. I mean, Boinc should at least *try* and download the CUDA client.
In my case, Boinc does not even try to fetch the CUDA application. Exactly as if I had no CUDA-capable hardware.
Also, since I *did* get 6.05_cuda version of the MB application (that was before I uninstalled everything), I think one can rule out http redirects.

Depends what version you re-installed. 6.4.5 is the version recommended for it to work with.

I am running version 6.4.5 under WindowsXP. My system has two GPUs: a 8600GT with 512MB and a 8300GS(I guess) with 256 MB. The latter is built in my motherboard.
Driver version is 178.24, freshly reinstalled as well.

This is the link to my PC. I hope it works now.

http://setiathome.berkeley.edu/show_host_detail.php?hostid=3259123

Thank you for any help you may give me.

---

Explained here

I see. But if http redirects were the cause, I should experience download errors. I mean, Boinc should at least *try* and download the CUDA client.
In my case, Boinc does not even try to fetch the CUDA application. Exactly as if I had no CUDA-capable hardware.
Also, since I *did* get 6.05_cuda version of the MB application (that was before I uninstalled everything), I think one can rule out http redirects.

Depends what version you re-installed. 6.4.5 is the version recommended for it to work with.

I am running version 6.4.5 under WindowsXP. My system has two GPUs: a 8600GT with 512MB and a 8300GS(I guess) with 256 MB. The latter is built in my motherboard.
Driver version is 178.24, freshly reinstalled as well.

This is the link to my PC. I hope it works now.

http://setiathome.berkeley.edu/show_host_detail.php?hostid=3259123

Thank you for any help you may give me.

Not necessarily. We went through this before in '07 the last time the project resorted to using Coral Cache.

Basically, it boils down to if whomever is controlling your border gateway is blocking cross site redirects, your attempt to DL the application will die if the project redirects you.

In this case the best way to work around it is to DL all the apps you need directly from the project download directory (the Coop or wherever for the others) and install everything manually.

Try using this URL as the target to get the app:

http://boinc2.ssl.berkeley.edu/sah/download_fanout/coral/

<edit> Needless to say, as long as the project is using the Coral 'crutch', projects resets and detaches are something to be avoided if this is an issue for you. ;-)

HTH,

Alinator

It may be a new version, but it still stalls when the screensaver kicks in. Which to me defeats the point of the project. Stops the project from working when machine is idle. I wonder who thought that was a good idea.

---

It may be a new version, but it still stalls when the screensaver kicks in. Which to me defeats the point of the project. Stops the project from working when machine is idle. I wonder who thought that was a good idea.

LOL...

Agreed, but masquerading science as screensaver eyecandy right from the start is sort of like bringing a knife to a gunfight. :-D

You know what they say about what you have to do after one craps in the nest, so to speak! ;-)

Alinator

I don't use the screensaver function. I discovered in the Classic days that running the screensaver would increase the crunch times by about 10 times (Celeron 633). With the screensaver on.. 154 hours, off..14 hours. Ever since then, I run it with no screensaver. My screensaver is "turn off monitor after x minutes".

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

Depends what version you re-installed. 6.4.5 is the version recommended for it to work with.

I am running version 6.4.5 under WindowsXP. My system has two GPUs: a 8600GT with 512MB and a 8300GS(I guess) with 256 MB. The latter is built in my motherboard.
Driver version is 178.24, freshly reinstalled as well.

This is the link to my PC. I hope it works now.

http://setiathome.berkeley.edu/show_host_detail.php?hostid=3259123

Thank you for any help you may give me.[/quote]

You have to have:

* The right version of BOINC
* The right types of cards (not sure if yours qualifies)
* The right NVidia drivers (180.48 seems to be the best at this time)
* If I understand it correctly there are some issues with optimized apps and the App Info files so if you are running those that file may need editing. there are other threads that talk to that (over my head)

If all this is correct, when you start BOINC you will see a message indicating CUDA devices detected. If you don't, you won't get CUDA work.

---

I've had some wierd results with the CUDA crunchers (lack of better word).

System:
CPU: Intel E6850 Duo @ 3ghz
Grafix: Nvidia GeForce 8600 GTS (CUDA), version 178.24
RAM: 2gb DDR2
Windows XP SP3
BOINC manager: 6.4.5
BOINCview: 1.4.2
Running: SETI, Einstein, Spinhenge.

Description:

---------------

First occurance: 30. Dec 2008, 19:15-19:30 GMT+1

At first I noticed I was running 3 instances of BOINC programs. Two of these were the normal 2,4 GFLOPS for each of the Duo core, 2,4 GFLOPS is normal for this CPU running at 90%. The last program was running at seemingly 304,2 GFLOPS, without moving its progress bar?

My computer was acting up a bit so I decided to restart the BOINC manager. When it came back up it was only running 2 instances, I was expecting 3 (DUO CPU + GPU). Before the restart I was seeing 4 instances of BOINC programs, now I'm finding 2. Ontop of that - normally the Task manager shows the Performance as pushing the normal 90% (9 sec work, 1 sec off) basicly having the Performance linie in the top around the 90% load for both CPUs.

Restarting the manager again gave no change.
Reboot PC gave no change

Pausing the Einstein and Spinhenge projects gave no change
Restarting the two, now it's running 4 programs again - 1 Astropulse, 2 Einstein, 1 SETI. Performance is back up in the top with the regulated 90% load as per rules.

Now I'm confused.

BoincView tells me, I'm running 1 Einstein, 1 Astropulse, 1 SETI. Oh wait now it suddenly finished the Einstein with 80% done. This is getting even better. Now I'm running 1 Astropulse, 1 Spinhenge, 1 SETI. The Astropulse and Spinhenge are now running as usual on each their cores at 2,5 GFLOPS and the SETI is using what would assumably be 0.03% CPU and at 60 MFLOPS (rest on CUDA).

Performance is as should be expected. Astro is base priority: Low, Spinhenge: Low, SETI: Normal (CUDA).

---------------

second occurance: 30. Dec 2008, 22:22 GMT+1

Now I got 1 Astro, 2 Einstein, 1 Spinhenge, 1 SETI loaded in memory. Only 1 Einstein and 1 SETI is running (seems to be the CUDA).

The CUDA seems to need some work on its Base Priority scheduling, it creates havoc trying to play games or watch movies. Suddenly it all jumps and lags and basicly can't keep up showing it in a flowing stream of images (24frames/second or more). Hmm...

Q: could it be an issue of dropping inactive programs out of memory?

Q: could it be deciding what and how much a program should use of resources? (on the loading part)

----

This morning I'm out of CUDA programs and my buffer is filled with Astropulses, hence I need to wait till I get any CUDA's to crunch. I think its nice we can see the progress of the CUDA.

Regards

DoStuffZ

---

I have updated to version 6.4.5, reset my projects and 6.0.6 Cuda was installed. The SETI units are now completing in 16 seconds with "computation error" as the result. Any idea of how to make it work or should I just let it do my other projects and not worry about it?

My system:
Vista Ultimate X64
Intel Core i7 (920) CPU
3 GB DDR3 RAM
Nvidia Gforce 9800GT 512MB RAM

Logs:
1/2/2009 4:54:21 PM||Starting BOINC client version 6.4.5 for windows_x86_64
1/2/2009 4:54:21 PM||log flags: task, file_xfer, sched_ops
1/2/2009 4:54:21 PM||Libraries: libcurl/7.19.0 OpenSSL/0.9.8i zlib/1.2.3
1/2/2009 4:54:21 PM||Data directory: C:\ProgramData\BOINC
1/2/2009 4:54:21 PM||Processor: 8 GenuineIntel Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz [Intel64 Family 6 Model 26 Stepping 4]
1/2/2009 4:54:21 PM||Processor features: fpu tsc pae nx sse sse2 pni
1/2/2009 4:54:21 PM||OS: Microsoft Windows Vista: Ultimate x64 Editon, Service Pack 1, (06.00.6001.00)
1/2/2009 4:54:21 PM||Memory: 2.99 GB physical, 6.18 GB virtual
1/2/2009 4:54:21 PM||Disk: 97.66 GB total, 44.74 GB free
1/2/2009 4:54:21 PM||Local time is UTC -8 hours
1/2/2009 4:54:21 PM||Not using a proxy
1/2/2009 4:54:21 PM||CUDA devices found
1/2/2009 4:54:21 PM||Coprocessor: GeForce 9800 GT (1)
1/2/2009 5:06:24 PM|SETI@home|Starting 01dc08ac.19772.9070.6.8.131_1
1/2/2009 5:06:24 PM|SETI@home|Starting task 01dc08ac.19772.9070.6.8.131_1 using setiathome_enhanced version 606
1/2/2009 5:06:44 PM|SETI@home|Computation for task 01dc08ac.19772.9070.6.8.131_1 finished

---

The only thing I can think of that keeps getting rehashed over and over on here is what driver version are you using? From what I've heard, 180.48 has the best results for the largest number of people.

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

I think the largest problm, is what mood the b***h who's doing the crunching is in. There is no other reason for it. Cuda must be a male thing, afterall it only does one thing at a time. My poor machine has three personalities, the Mac part doesn't know what to do anymore.

---