1) Message boards : Graphics cards (GPUs) : GTX 750ti switching to default clock value (1058MHz) after a while (Message 46202)
Posted 7 minutes ago by Profile Retvari Zoltan
After the card has run stable on 1200 MHz for two days (crunching BNBS), a few hours ago it again switched back to default clock 1058 MHz.

So I tested the card with SETI@home.
Step by step I raised the clock to 1300 Mhz, and for several hours it's been running at this clock without any problem.
GPU load ~97% is even a little higher than it is with BNBS (~ 94%), TDP is 66 - 69%.
GPU load is misleading, it's actually the load of the units which distribute the work to the CUDA cores.
The TDP gives a more accurate reading of how much the CUDA cores are utilized. (The TDP was between 83 and 94% by the BNBS workunits)

This nourishes my suspicion that the GPU may have a problem with the BNBS WUs (in contrast to the GTX 750ti in the other PC).
These workunits present the highest stress to your card (comparing to the SETI workunits, or the other GPUGrid workunits), so it could be a slightly 'defective' card. You could test it with stress-test tools like Furmark or MSI Kombustor. Let it run for an extended period of time. While it's running, you should look for artifacts (miscolored dots, missing triangles, vertical lines that disappear in the next frame, any strange pattern) to know if your card is defective. It could take ~30 minutes (or more) for these artifacts to show up. You could use the Unigine Heaven test too, but it does not stress the card as much as the other two tools.
2) Message boards : Server and website : NO TASKS AVAILABLE (Message 46196)
Posted 4 hours ago by Profile Retvari Zoltan
3600 simulations running in parallel is pretty nuts.
That's the point in grid computing :)
That's the power of a large community.
3) Message boards : News : WU: BNBS (Message 46185)
Posted 1 day ago by Profile Retvari Zoltan
Well, even my PCIe2x16 conncected gtx1080 runs at pretty 95% per GPU-Z... which is a very pleasing figure. The BNBS accommodate the fast Pascal I presume and reduces the WDDM influence. Anyway, I am happy with the throughput.

Exactly. There's unusually little interaction between the CPU and the GPU by these workunits.
4) Message boards : Number crunching : BOINC Supercomputer (Message 46181)
Posted 1 day ago by Profile Retvari Zoltan
1.21 Jigawatts


That's funny, I call one of my computers flux-capacitor :)
5) Message boards : News : WU: BNBS (Message 46179)
Posted 1 day ago by Profile Retvari Zoltan
On my computer (1060 3GB), BNBS works take 17 hours. In general for 410.000 credits, for some 490.000.
You should set a short work queue (less than 5 hours = 0.2 days) to achieve 490.000 credits per workunit (this amount includes the bonus for returning the result within 24 hours).

First, I am happy to have a recent Pascal card...
In second: before the new server, I had 100.000 credits more by day. We don't do that just for credits but I am sad nevertheless.
The BNBS workunits are exatly as long as the BNB was, so your PPD will return to the previous level. It's probably lower than before because the project aborted one of your workunits mid-run.

And the error rate is 30%, it is huge. (always good for me I have to say)
The error rate starts at 100% every time, then it is continuously falling, as more and more successful tasks are returned. (now it's down to 29%.)

A BNB work (with a huge 50% error rate) finished with error couple days ago.
The BNB batch had an error (they are not chained correctly, it's explained here), and had to be aborted by the project which explains the huge error rate, and your lost workunit.

Between 14 and 17 hours for ... nothing; sorry: still, I will always have my electricity bill :-(
This is that workunit, it was aborted at 28,686.86 seconds (almost 8 hours).
6) Message boards : Number crunching : "Remaining (estimated)" Times Somewhat Awry (Message 46177)
Posted 1 day ago by Profile Retvari Zoltan
On Windows XP the default location of the BOINC data directory is different, so on Windows XP you should use the following command to edit/create the app_config.xml:
notepad "C:\Documents and Settings\All Users\Application Data\BOINC\projects\www.gpugrid.net\app_config.xml"
7) Message boards : Number crunching : "Remaining (estimated)" Times Somewhat Awry (Message 46176)
Posted 2 days ago by Profile Retvari Zoltan
(Sorry for not creating a new thread about this.)
Now that I've read (again) the wiki about the application configuration in BOINC, I've 'discovered' a fairly new option (introduced in v7.3.13), which makes the BOINC manager to "base estimates of remaining time solely on the fraction done reported by the app". This estimate fluctuates 0~15 seconds at the beginning, but it won't over/underestimate the remaining time by hours when a new batch shows up. This option is called "<fraction_done_exact/>" and it should be placed in the app_config.xml for each application (acemdbeta, acemdlong, acemdshort)

---------------- How to do that (on Windows): -------------------
The app_config.xml file should be placed to the project's home directory (by default it's at c:\ProgramData\BOINC\projects\www.gpugrid.net\)
Copy the following to the clipboard:
notepad c:\ProgramData\BOINC\projects\www.gpugrid.net\app_config.xml
Press Windows key + R, then paste and press [enter].
If you see an empty file then copy & paste the following text:

<app_config> <app> <name>acemdlong</name> <fraction_done_exact/> <gpu_versions> <gpu_usage>0.5</gpu_usage> <cpu_usage>1.0</cpu_usage> </gpu_versions> </app> <app> <name>acemdshort</name> <fraction_done_exact/> <gpu_versions> <gpu_usage>0.5</gpu_usage> <cpu_usage>1.0</cpu_usage> </gpu_versions> </app> <app> <name>acemdbeta</name> <fraction_done_exact/> <gpu_versions> <gpu_usage>0.5</gpu_usage> <cpu_usage>1.0</cpu_usage> </gpu_versions> </app> </app_config>

This app_config.xml is for running two GPUGrid tasks on each GPU simultaneously (recommended for fast GPUs on Windows Vista or newer), if you want to run only 1 per GPU (recommended for Windows XP, or slow GPUs) you should change the <gpu_usage> to 1.0 in all 3 applications like this:
<gpu_usage>1.0</gpu_usage>

If you already have an app_config.xml, then you should only insert the line
<fraction_done_exact/>
after each line containing the name of the application.

Click file -> save and click [save].
Open the BOINC manager, click Options -> read config files.
8) Message boards : Server and website : Welcome back Gpugrid (Message 46174)
Posted 2 days ago by Profile Retvari Zoltan
I had two BNBS2 WUs run for 100k+ seconds and had a validation error, can anyone explain this?

Perhaps your host had a power outage, and these GPUGrid tasks restarted from 0%.
In such cases it is practical to abort the workunits manually, as there's no point in spending time and electricity crunching them.

Here's two excerpts from the stderr.txt of your failed tasks:
1st:
# GPU 2 : 73C # GPU [GeForce GTX 970] Platform [Windows] Rev [3212] VERSION [65] # SWAN Device 0 : # Name : GeForce GTX 970

2nd:
# GPU 0 : 73C # GPU [GeForce GTX 690] Platform [Windows] Rev [3212] VERSION [65] # SWAN Device 1 : # Name : GeForce GTX 690
Note that there's no line explaining the reason to the exit from the application between the 1st and the 2nd line, which is usually the sign of a dirty shutdown.
9) Message boards : Graphics cards (GPUs) : Not getting jobs - Linux Boinc 7.2.42 (Message 46173)
Posted 2 days ago by Profile Retvari Zoltan
If you look closely you'll find the reason for it as your host asks GPUGrid for CPU tasks not for GPU tasks:
Sat 14 Jan 2017 09:31:39 AM EST | GPUGRID | Requesting new tasks for CPU ... Sat 14 Jan 2017 10:31:44 AM EST | GPUGRID | Requesting new tasks for CPU

Perhaps your BOINC manager did not detect your GPU, please check the first lines in the BOINC manager at its startup.
10) Message boards : Number crunching : Multiple work units? (Message 46171)
Posted 2 days ago by Profile Retvari Zoltan
GPUGrid limits the number of workunits in progress on a given host at 2 per (NVidia) GPUs.
If your host has only 1 workunit, then some setting on your side limits the number of workunits.
It could be:
- your host is attached to too many projects
- GPUGrid resource share is set to 0
- the (BOINC manager/options/) computing preferences "store at least [...] days of work" is set to 0 (or a very low value)


Next 10