Posts by Thalus
log in
1) Message boards : Number crunching : Dead Link (Message 3427)
Posted 21 Apr 2017 by Thalus
BOINC Combined Statistics is still using the "old" URL.
2) Message boards : News : new router / downtime on weekend (done) (Message 3425)
Posted 20 Apr 2017 by Thalus
Will there be work remaining before downtime or will you let the wus run out?
3) Message boards : Number crunching : Calculated Runtimes (Message 3419)
Posted 15 Apr 2017 by Thalus
I don't know how the runtimes getting calculated:
R888
- n=250-400k
- runtime 52min-2h35min (AVX@3.8Ghz)
- 250-300k = 375 credits
- 300-350k = 525 credits
- 350-400k = 700 credits
- Sierpinski / Riesel Base app -average3
- deadline 3 days

The runtime of 2h35min is definitly wrong, a 300-350k WU takes (at the moment) 2h45min on a 6700k running at 4.2GHz.
http://srbase.my-firewall.org/sr5/result.php?resultid=216830294
Just guessing now but I think the 350-400k will take about 4h then. 525 credits for a (nearly) 3h WU seems to be very low.

I recognized that before too, calculation time is set very low, real runtime is about calculated +10-50%.
I have no idea how your 3.8GHz AVX CPU can do that in 2h35min. Just take the WU above and show me I am wrong.
4) Message boards : Number crunching : LLR Version 3.8.20 released (Message 3399)
Posted 6 Apr 2017 by Thalus

Even with restricting to 4 wu at the same time my ryzen sys takes a 5th wu with 4 vcores on a different app, like 4x riesl base + 1x sierpinski/riesel, which results in 5x 19.x percent usage.


I have the same problem with win10 x64 and my skylake... so i stopped using the "-tx" option.
5) Message boards : Number crunching : Process_priority (Message 3394)
Posted 6 Apr 2017 by Thalus
Thanks! I'll give it a try. :-)
6) Message boards : Number crunching : Process_priority (Message 3391)
Posted 6 Apr 2017 by Thalus
I would have let it run in normal mode to have a higher priority than wuprop. Thats all.
7) Message boards : Number crunching : Process_priority (Message 3389)
Posted 5 Apr 2017 by Thalus
The taskmanager tells me the follwing:
wrapper - normal priority
llr.exe - low priority (lowest possible in Windows 10)
wuprop - lower than normal (so it's higher than llr.exe)

So here llr.exe doesn't have priority "normal".
8) Message boards : Number crunching : Process_priority (Message 3387)
Posted 5 Apr 2017 by Thalus
Is it possible to change the process priority of the executed llr.exe by setting the <process_priority>-tag in the cc_config.xml? Tried that several times since WUProp actually has a higher priority (lower than normal) than llr.exe has (lowest) in windows 10. Or is the priority fixed?
9) Message boards : Cafe : Work done by BOINC compared (Message 3384)
Posted 3 Apr 2017 by Thalus
Sorry, haven't seen the last update.
Thanks!
10) Message boards : Cafe : Work done by BOINC compared (Message 3382)
Posted 3 Apr 2017 by Thalus
Hi,
How is BOINC doing compared to other projects crunching the Riesel-/Sierpinski-problems? To be more precise, how much work was done by this project of all done work presented by KEP's statistics? I dont't know if this is easy to say, I am just interested. ;)
11) Message boards : Number crunching : FMA3 vs AVX (Message 3374)
Posted 1 Apr 2017 by Thalus
Stock is 4.2 turbo on 1 core. I set all cores to 4.2 so clockspeed is identically for all cores. And yes. I can see all cores at 4.2.
I use 2x8GB DDR4 @ 3GHz.
12) Message boards : Number crunching : Subprojects (Message 3367)
Posted 1 Apr 2017 by Thalus
Ihave a question about the subprojects:
Why does this project have them? On most other projects ppl cab choose between the different ones but since this project only has work for (max.) 2 different subprojects at the same time people which wanna crunch (for example) short WUs can only use SRBase as a backup project or some kind of 2nd choice. Work is not always available.
Just for clarification, this post is no offence, I just wanna ask why the subprojects are filled sporadically with work?
13) Message boards : Cafe : SRBase server stress test for potential DC-Vault inclusion (Message 3356)
Posted 28 Mar 2017 by Thalus
How are things going til now? Found some primes? How is the outcome compared to "normal" days?
14) Message boards : Number crunching : SRBase not following sub project preferences (Message 3355)
Posted 27 Mar 2017 by Thalus
Still happens sometimes, randomly i get WUs i have not checked in the menu. Even tough I am just getting around 1-2 WUs each request... not what I wanted but it's ok.
Keep crunching! ;)
15) Message boards : Number crunching : SRBase not following sub project preferences (Message 3354)
Posted 27 Mar 2017 by Thalus
Thats kinda disappointing... I want to get work for a specifi app, but dont want to turn off the "Get anything if nothing is available for the selected" cause then it could happen, that I dont have any work at all...
16) Message boards : Number crunching : SRBase not following sub project preferences (Message 3352)
Posted 27 Mar 2017 by Thalus
Sorry for "grave-digging" but I got the same problem. Randomly I keep getting Sierpinski Base WUs. I enabled "get work for other applications if none of the selected are available" but still it's strange. There are definitly WUs available for "Sierpinski Base - short".
I noticed that kind of behavior a few days ago but thought I accidently checked a wrong box.
My current preferences are

Sierpinski / Riesel Base: no
Riesel Base: no
Sierpinski Base: no
Sierpinski / Riesel Base - short: no
Sierpinski / Riesel Base - long: no
Sierpinski / Riesel Base - average: yes
Riesel Base - short: yes
Sierpinski Base - short: yes
Sierpinski / Riesel Base - average2: yes
Sierpinski / Riesel Base - average3: yes
Sierpinski / Riesel Base - long2: no
Sierpinski / Riesel Base - long3: no

Last changes made ~5hrs ago, no Sierpinski Base where in cache at this point of time. Now I got 59 WUs here.

Edit:
This "erro" just occures when a bigger amount of WUs is downloaded. So after switching from, lets say, long WUsto sgort ones. A bigger amount is downloaded, here around 200, in multiple junks. After the first junk boinc thinks there are no more available and downloads some others, next junks are the "right" WUs again, next one again some I have not checked. And so on.
I guess there is no real workaround since this is some kind of error handling of boinc?
17) Message boards : Number crunching : Dead Link (Message 3339)
Posted 24 Mar 2017 by Thalus
The "badges"-link on the main page is dead now too. Same issue.
18) Message boards : Number crunching : Zero Padding (Message 3323)
Posted 21 Mar 2017 by Thalus
Found the solution having a closer look at the GWNUM library. Reason is, that for larger FFTs (I think it 50k or higher) another accuracy is needed, so the calculation is using the zero padded methods.
19) Message boards : Number crunching : Zero Padding (Message 3322)
Posted 21 Mar 2017 by Thalus
Why is zero padding used for some WUs and not for others? Is there any advantage or just no possibility to calculate the zero padded WUs?
20) Message boards : Number crunching : LLR Version 3.8.20 released (Message 3320)
Posted 20 Mar 2017 by Thalus
How do i reload the "stock"-settings after editing/adding a app_conf.xml? I deleted the file, reloaded the settings in Boinc but it´s still using the preferences I added.

Edit:
Nevermind... deleting app_config.xml and resetting the project.


Next 20

Main page · Your account · Message boards


Copyright © 2014-2018 BOINC Confederation / rebirther