Posts by Mike
log in
1) Message boards : Number crunching : Random download errors (Message 23441)
Posted 423 days ago by Mike
In my case, I had to update ca-bundle.crt after the site's certificate was replaced. I found on Seti@home instructions to get more information from the log.
In notepad, create a file called cc_config.xml in directory F:\ProgramData\Boinc with these contents:
<cc_config>
<options>
<http_1_0>1</http_1_0>
</options>
</cc_config>

That led me to that crt file.


I already have a cc_config.xml file, but do not have a ca-bundle.crt file in \programdata\Boinc.

Where should it go and what should it contain?
2) Message boards : Number crunching : Random download errors (Message 23435)
Posted 425 days ago by Mike
For the last few days all my uploads and downloads on both machines are stuck and I just get error messages like:

24/11/2016 13:00:52 | Collatz Conjecture | Temporarily failed download of collatz_sieve_2954731493023411077120_6597069766656: transient HTTP error

One machine has now run out of work completely. Any advice?
3) Message boards : Number crunching : Computation Errors (Message 20647)
Posted 947 days ago by Mike
Recently with in the last 10 days I have been seeing computations errors popping up as the last WU is finished the next one to start goes just 2 secs then goes to error. This will usually occur in pairs then stop and go to the next WU and perform flawlessly.
And i with several R3 Kalindi by Debian Jessie.
http://boinc.thesonntags.com/collatz/result.php?resultid=18364275

I have nothing changed.


This one is interesting:

http://boinc.thesonntags.com/collatz/workunit.php?wuid=15913445

Exactly the same error message for both an AMD and an NVIDIA GPU:

At offset 16777216 got 627 from the GPU when expecting 203

My guess: this is the same problem as discussed in this thread:

http://boinc.thesonntags.com/collatz/forum_thread.php?id=1272

Tom


and to complete the set I get the same error behaviour on my integrated intel GPU. Usually 2 or 3 units error out after 2 seconds, then one or more run OK.
4) Message boards : News : Variable Sized Wokunits & New CPU Applications (Message 18847)
Posted 1411 days ago by Mike
Something funny going on. My INTEL2500 is now crunching WU which take 6 minutes (estimated and actual). Assuming these are the new shortest WU, that implies a full length WU should take 24 minutes when they were actually taking 6 hours!

EDIT they are worth around 60 credits each which is about 1/60 of a full length WU and they take about 1/60 of the time so that at least is consistent.
5) Message boards : News : More Updates (Message 18727)
Posted 1427 days ago by Mike
The second batch of credits was awarded yesterday.

Any idea when the remaining credits will be awarded? My host have had no compensation for credit lost during the outage.


Credit was granted to the user, not the individual hosts.

Well then same question but at user level. My boinc stats page for collatz shows no obvious compensation on any day since the outage.
6) Message boards : News : More Updates (Message 18724)
Posted 1427 days ago by Mike
The second batch of credits was awarded yesterday.

Any idea when the remaining credits will be awarded? My host have had no compensation for credit lost during the outage.
7) Message boards : News : CUDA vs nVidia OpenCL User Poll (Message 18546)
Posted 1439 days ago by Mike
Definitely A.
8) Message boards : Number crunching : Optimizing Collatz v6.xx OpenCL and CUDA Applications (Message 17854)
Posted 1526 days ago by Mike
You have an older Nvidia card, the identical settings are not supposed to work, except by coincidence. You computer has 4 cpu cores yet you are trying to run 8 threads thru a NVIDIA GeForce GT 430 gpu. I think you need to go back to the stock settings and talk to someone, not me, who really understands these settings and can help you tweak them for your gpu.


Two things...
He has an Intel 2500 embedded GPU which is not as fast as a stand-alone nVdia or AMD GPU. The CPU with the GPU costs < $300. The high end GPUs cost $600 without a CPU (not that dollars always coincides with relative speed, but there is no such thing as a free lunch.)

Secondly, my use of the term "threads" in GPU apps is because that is what they were called in Brook+/CAL. OpenCL's term for "threads" is work group size. So, do not confuse threads on the GPU with cores on a CPU. "8" threads means 2^8 or 256 calculations done in parallel on the GPU, not the number of cores on the CPU.

Items per kernel is also a power of two (18 = 2^18 or 262,144. In other words, each kernel will calculate 262,144 numbers before returning the data. To do that, it will calculate the steps for 256 numbers (2^8 threads) at a time until all 262,144 are done. It will repeat that 2^7 or 128 times (a.k.a. 7 kernels per reduction). So, after 262,144 * 128 numbers, the CPU will total up the steps and make note of the highest steps (hoping to find one that causes an overflow as that would potentially be one that proves the Conjecture is false).
It repeats that until done with the WU.


So how come I see no speed enhancement with the optimization? Is it a case of playing around with the parameters? If so, which ones? Or is it just that my puny little INTEL GPU cannot work any harder than the stock conditions?
9) Message boards : Number crunching : Optimizing Collatz v6.xx OpenCL and CUDA Applications (Message 17847)
Posted 1527 days ago by Mike
Have you actually completed any WUs with the file in place from the beginning?


Yes one every 13 hours for the last two days. See http://boinc.thesonntags.com/collatz/results.php?hostid=134418. the valid results from today and yesterday will have had the file in place from the start.
[/url]
10) Message boards : Number crunching : Optimizing Collatz v6.xx OpenCL and CUDA Applications (Message 17845)
Posted 1527 days ago by Mike
In do not have a Collatz directory, just BOINC.thesonntags.com_collatz


That's it, the Collatz folder.


Well that's where I put the file and it is being copied to the running slot, but no decrease in run time.
11) Message boards : Number crunching : Optimizing Collatz v6.xx OpenCL and CUDA Applications (Message 17843)
Posted 1527 days ago by Mike
In do not have a Collatz directory, just BOINC.thesonntags.com_collatz

12) Message boards : Number crunching : Optimizing Collatz v6.xx OpenCL and CUDA Applications (Message 17838)
Posted 1528 days ago by Mike


The key is optimising the output. To do that you must paste this ......


verbose=1
items_per_kernel=18
kernels_per_reduction=7
threads=8
sleep=1
build_options=-Werror


.... into the existing file "solo_collatz_4.07_windows_x86_64__opencl_ati_100" marked as a CONFIG file. Probably the directory on yours is

C:\ProgramData\BOINC\projects\boinc.thesonntags.com_collatz



I tried this on one of my INTEL 2500 systems. But it made absolutely no difference to the run times. I am running solo_collatz 4.09. I edited my solo_collatz_4.09_windows_x86_64__opencl_intel_gpu_100.config file in notepad to contain the above parameters.

verbose=1
items_per_kernel=18
kernels_per_reduction=7
threads=8
sleep=1
build_options=-Werror

This file did get copied to the collatz.config file in the slot where the app was running, but the run time has not changed from 13.5 hours.

Am I missing something?
13) Message boards : Number crunching : help please I am trashing WUs (Message 17744)
Posted 1541 days ago by Mike
It looks like this problem is now solved. the system has run fine for 24 hours with 4 CPU cores and 2 GPUs running a mixture of SETI and Collatz with no problems.

Thanks for the help.
14) Message boards : Number crunching : help please I am trashing WUs (Message 17729)
Posted 1542 days ago by Mike
well I do not understand why, but that seems to have fixed the problem. Boinc now recognises both graphics cards and has downloaded 11 new Collatz tasks, the first of which is currently running OK. I have turned off network access so that I can check it runs OK overnight and to stop the PC downloading and trashing lots more tasks if something goes wrong.
15) Message boards : Number crunching : help please I am trashing WUs (Message 17728)
Posted 1542 days ago by Mike
now I am really confused. I swapped the two monitors over and my Boinc log changed to the following:

03/11/2013 19:34:37 | | Starting BOINC client version 7.0.64 for windows_x86_64
03/11/2013 19:34:37 | | log flags: file_xfer, sched_ops, task, sched_op_debug
03/11/2013 19:34:37 | | Libraries: libcurl/7.25.0 OpenSSL/1.0.1 zlib/1.2.6
03/11/2013 19:34:37 | | Data directory: C:\ProgramData\BOINC
03/11/2013 19:34:37 | | Running under account M North
03/11/2013 19:34:37 | | Processor: 4 GenuineIntel Intel(R) Core(TM) i5-3470 CPU @ 3.20GHz [Family 6 Model 58 Stepping 9]
03/11/2013 19:34:37 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 popcnt aes syscall nx lm vmx smx tm2 pbe
03/11/2013 19:34:37 | | OS: Microsoft Windows 7: Professional x64 Edition, Service Pack 1, (06.01.7601.00)
03/11/2013 19:34:37 | | Memory: 3.90 GB physical, 7.81 GB virtual
03/11/2013 19:34:37 | | Disk: 453.47 GB total, 349.16 GB free
03/11/2013 19:34:37 | | Local time is UTC +0 hours
03/11/2013 19:34:37 | | CUDA: NVIDIA GPU 0: GeForce GT 430 (driver version 331.65, CUDA version 6.0, compute capability 2.1, 1024MB, 902MB available, 269 GFLOPS peak)
03/11/2013 19:34:37 | | OpenCL: NVIDIA GPU 0: GeForce GT 430 (driver version 331.65, device version OpenCL 1.1 CUDA, 1024MB, 902MB available, 269 GFLOPS peak)
03/11/2013 19:34:37 | | OpenCL: Intel GPU 0: Intel(R) HD Graphics 2500 (driver version 9.17.10.3040, device version OpenCL 1.1, 1624MB, 1624MB available, 17 GFLOPS peak)
03/11/2013 19:34:37 | SETI@home | Found app_info.xml; using anonymous platform
03/11/2013 19:34:37 | | Config: use all coprocessors
03/11/2013 19:34:37 | Collatz Conjecture | Config: excluded GPU. Type: NVIDIA. App: all. Device: 0
03/11/2013 19:34:37 | | Bad URL in GPU exclusion: http://setiweb.ssl.berkeley.edu/
03/11/2013 19:34:37 | Collatz Conjecture | URL http://boinc.thesonntags.com/collatz/; Computer ID 134418; resource share 100
03/11/2013 19:34:37 | Einstein@Home | URL http://einstein.phys.uwm.edu/; Computer ID 9226819; resource share 0
03/11/2013 19:34:37 | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7118766; resource share 100
03/11/2013 19:34:37 | Cosmology@Home | URL http://www.cosmologyathome.org/; Computer ID 200668; resource share 100

both GPUs are now recognised (lines 12-14), but they are both listed as GPU 0! How can this be? presumably this is why the Collatz tasks were being trashed due to a conflict with the Nvidea.
16) Message boards : Number crunching : help please I am trashing WUs (Message 17723)
Posted 1542 days ago by Mike


Did you exclude the Nvidia card from getting gpu units? If not that could be the one trashing the units not the Intel gpu. Try this file:
<cc_config>
<log_flags>
<file_xfer>1</file_xfer>
<sched_ops>1</sched_ops>
<task>1</task>
<sched_op_debug>1</sched_op_debug>
</log_flags>
<options>
<use_all_gpus>1</use_all_gpus>
<exclude_gpu>
<url>http://boinc.thesonntags.com/collatz/</url>
<device_num>0</device_num>
<type>NVIDIA</type>
</exclude_gpu>
<exclude_gpu>
<url>http://setiweb.ssl.berkeley.edu/</url>
<device_num>0</device_num>
<type>intel_gpu</type>
</exclude_gpu>
</options>
</cc_config>


I edited the cc_config to get it to show some more info as well as setting it to ignore only the Nvidia GPU on Collatz and the Intel on SETI.

Also, can you post the first 20 lines after startup of the BOINC Event log.


Mikey and Arkayn

Thanks for the advice. I used Arkayn's version of the config file. Shut down and restarted Boinc. I got an error message about the web address for seti being wrong (I think this is easy to fix), but Boinc still does not recognise that an intel gpu is present. Extract of event log below. Any other advice most welcome.

03/11/2013 13:07:09 | | Starting BOINC client version 7.0.64 for windows_x86_64
03/11/2013 13:07:09 | | log flags: file_xfer, sched_ops, task, sched_op_debug
03/11/2013 13:07:09 | | Libraries: libcurl/7.25.0 OpenSSL/1.0.1 zlib/1.2.6
03/11/2013 13:07:09 | | Data directory: C:\ProgramData\BOINC
03/11/2013 13:07:09 | | Running under account M North
03/11/2013 13:07:09 | | Processor: 4 GenuineIntel Intel(R) Core(TM) i5-3470 CPU @ 3.20GHz [Family 6 Model 58 Stepping 9]
03/11/2013 13:07:09 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 popcnt aes syscall nx lm vmx smx tm2 pbe
03/11/2013 13:07:09 | | OS: Microsoft Windows 7: Professional x64 Edition, Service Pack 1, (06.01.7601.00)
03/11/2013 13:07:09 | | Memory: 3.90 GB physical, 7.81 GB virtual
03/11/2013 13:07:09 | | Disk: 453.47 GB total, 349.16 GB free
03/11/2013 13:07:09 | | Local time is UTC +0 hours
03/11/2013 13:07:09 | | CUDA: NVIDIA GPU 0: GeForce GT 430 (driver version 331.65, CUDA version 6.0, compute capability 2.1, 1024MB, 844MB available, 269 GFLOPS peak)
03/11/2013 13:07:09 | | OpenCL: NVIDIA GPU 0: GeForce GT 430 (driver version 331.65, device version OpenCL 1.1 CUDA, 1024MB, 844MB available, 269 GFLOPS peak)
03/11/2013 13:07:09 | SETI@home | Found app_info.xml; using anonymous platform
03/11/2013 13:07:09 | | app version refers to missing GPU type intel_gpu
03/11/2013 13:07:09 | Collatz Conjecture | Application uses missing intel_gpu GPU
03/11/2013 13:07:09 | | Config: use all coprocessors
03/11/2013 13:07:09 | Collatz Conjecture | Config: excluded GPU. Type: NVIDIA. App: all. Device: 0
03/11/2013 13:07:09 | | Bad URL in GPU exclusion: http://setiweb.ssl.berkeley.edu/
03/11/2013 13:07:09 | Collatz Conjecture | URL http://boinc.thesonntags.com/collatz/; Computer ID 134418; resource share 100
03/11/2013 13:07:09 | Einstein@Home | URL http://einstein.phys.uwm.edu/; Computer ID 9226819; resource share 0
03/11/2013 13:07:09 | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7118766; resource share 100
03/11/2013 13:07:09 | Cosmology@Home | URL http://www.cosmologyathome.org/; Computer ID 200668; resource share 100

17) Message boards : Number crunching : help please I am trashing WUs (Message 17717)
Posted 1543 days ago by Mike
A week ago I got a new PC with an integrated intel gpu. I set it up to run SETI on the CPUs and Collatz on the GPU all under windows 7 - no problem.

I also had a Nvidia GTS450 graphics card and a second monitor so I decided to install those as well. I read around and thought I had everything right. I installed the NVidia card and connected one monitor to the intel outlet and one to the GTS450 outlet. I set the BIOS to use all GPUs, then rebooted and downloaded the latest Nvidia driver. Rebooted again and all appeared well, both monitors worked and I could extend my desktop across them.

I then reinstalled lunatics (a SETI thing) to install the optimised apps for the Nvidia GPU. Again all appeared well and the PC has been happily crunching SETI WUs on both CPU and Nvidia GPU ever since.

However, the PC will no longer crunch Collatz WU on the Intel GPU. Yesterday it managed to trash its cash, download lots more and trash those as well (99 in total) and today it has trashed another 11. In each case the event log showed an error message that no output file was present. I rebooted the PC and noticed that the early part of the event log contains a Collatz line 'Application uses missing intel_gpu GPU).

How can the intel gpu be missing when it is clearly present and running one of the two monitors!? Any suggestions as to what I have overlooked to get the PC crunching properly will be welcome.




Main page · Your account · Message boards


Copyright © 2018 Jon Sonntag; All rights reserved.