1)
Message boards :
Number crunching :
Évariste Galois's Birthday Challenge
(Message 143919)
Posted 974 days ago by  HighTech67
Multi-threading....YES
Hyperthreading....NO
My mistake. Sorry. I often get the two confused.
I don't suppose there would be any way possible to get the short "proof" tasks to run on hyperthreads would there?
Or to only get the "proof" tasks? I've still got to read the thread about LLR2 to learn more about it.
|
2)
Message boards :
Number crunching :
Évariste Galois's Birthday Challenge
(Message 143917)
Posted 974 days ago by HighTech67
Multi-threading is supported and IS recommended.
Those looking to maximize their computer's performance during this challenge, or when running LLR in general, may find this information useful.[list]
* Your mileage may vary. Before the challenge starts, take some time and experiment and see what works best on your computer.
* If you have a CPU with hyperthreading or SMT, either turn off this feature in the BIOS, or set BOINC to use 50% of the processors.[list]
So, which is it? Use Multi-threading or not? Sounds contradictory to me.
I'm guessing you you just copied and pasted from previous challenges but don't know.
|
3)
Message boards :
Problems and Help :
Selecting which GPU to use
(Message 131411)
Posted 1413 days ago by HighTech67
Thanks John, this was super helpful. It's a little tedious but I did eventually get it to work.
No problem, glad I could help.
Being a new Ubuntu user, I also found it complicated. I reread the parts that were relevant to Linux more than half a dozen times before I even attempted it the first time. And it took me more than a couple of tries to get it all set up correctly. But it's running fine now.
John T
|
4)
Message boards :
Problems and Help :
Selecting which GPU to use
(Message 131219)
Posted 1418 days ago by HighTech67
I sort of think of the two files like this:
capitalization for effect only
CC_config.xml = Computer Config -> general configuration of BOINC for the computer.
APP_config.xml = All Project Preferences -> preferences for that particular project.
It's not perfect, but it keeps me from getting confused about the two.
Anything you change in cc_config.xml affects BOINC, everything in BOINC.
Anything you change in an app_config.xml only affects that specific project.
This is the reason for the locations of the files also.
cc_config.xml is in the C:\Program Data\BOINC folder on Windows because it affects all of BOINC.
The app_config.xml files are in specific project folders because they only affect that project and nothing else. Plus you can't have two files names app_config.xml in the same folder.
Like I said, this is how I look at it to make my life easier. I hope it helps someone else clear up any confusion.
Also, there is a great guide over on overclockers.net I believe for running multiple instances of BOINC on the same computer for various reasons. I do this and run 3 instances. One is CPU only, one is for GPU0 only and one for GPU1 only. Each GPU instance has it's own venue at each project I run and I can select not only which projects are ran by that card, but which sub-projects from that project if it offers sub-projects. I set it up on Linux after only using it for maybe a week. If anyone needs any help, PM me and I'll do what I can.
John T
|
5)
Message boards :
Sophie Germain Prime Search :
Sophie Germain Prime Search
(Message 131212)
Posted 1419 days ago by HighTech67
Thanks for the feedback.
Yes I'm aware of what HT is:D
Couldn't I just set my CPU to 50% in the BOINC settings also?
I'm now only using my old i5 3470 which doesn't have HT though. I will put my other PC's back on eventually.
I will never understand why is so hard to make 4 steps
1. restart computer
2.enter bios
3. disable HT
4. load windows and enjoy :)
Since you say you know what HT is, question is why you didnot disable it before :)
Not possible on every BIOS.
And I don't think setting BOINC to use 50% of processors will make it use the actual cores instead of threads all the time will it? Don't know, just asking. I know it helps.
If so, good.
If not, is there anything that can be done to automatically force tasks to run on the true cores only? Besides Process Lasso. For some reason it seemed to use more and more resources for itself on my main system the longer it ran until it was slowing everything down. I finally removed it.
I'm running two computers with 2nd gen Core i7 CPUs. One I can turn HT off, one I can't.
John T
|
6)
Message boards :
Number crunching :
Fastest GPU's
(Message 126642)
Posted 1576 days ago by HighTech67
If that's all you want to do it for then you could use the Exclude command and exclude each gpu from a different project and then sign up for both projects. Each gpu will run it's own project. Something like this works:
<cc_config>
<options>
<use_all_gpus>1</use_all_gpus>
<exclude_gpu>
<url>http://moowrap.net/</url>
<device_num>0</device_num>
</exclude_gpu>
<exclude_gpu>
<url>http://boinc.fzk.de/poem/</url>
<device_num>1</device_num>
</exclude_gpu>
<report_results_immediately>1</report_results_immediately>
</options>
</cc_config>
Could something similar be used to assign certain sub-projects at the same project to different GPUs?
For example, in the same computer I want to run 4 GPUs, some of which I already own, for PG it would be something like: a GTX 650 I would like to only run GFN-15, GFN-16, and both GFN-17s on, a GTX 750 I only want to run PPS-Sieve on, a GTX 970 I only want to run AP27 on, and get an RTX 2070 or higher to run only GFN-18 and above. Can this be done?
Different projects would have different assignments for each card and might, if possible leave the lowest card or two on PG while using the fastest two on other projects or maybe the other way around. Can this be done as well? Like if when I switch off the AP27 and GFN-18 and above at PG and turn on Seti@Home (running on only the two fastest cards using the above exable), can the two slowest cards be maybe automatically reassigned to both run say PPS-Sieve at PG to reduce heat?
I know I ask a LOT of stupid questions but I've thought about this for quite a while and figured this thread was a good opening to ask.
I'm currently looking for a MB that can support 4 PCIE slots at least at version 3.0 and X16 each. Most of them of course slow down to X8 or even X4 the more you use. Eventually I hope to end up with 4 liquid cooled 2080 TIs in the same system. CPU brand is not a huge thing for me now either, just want best bang for my buck. Would prefer to keep the MB EATX tho as I'm wanting to use the ThermalTake X9 case.
Thanx,
John T
|
7)
Message boards :
Problems and Help :
Multithreading w/ SoB doesn't work
(Message 126012)
Posted 1584 days ago by HighTech67
Hi all. After a looooong hiatus I'm getting back into PG, and noticed multithreading is now a thing. My 2018 MacBook Pro has eight cores, and I wanted to see them shred through SoB tasks. After plenty of experimentation, I haven't gotten anything to work. The result is always one task = one core.
Here is my config file, located at ~/Library/Application Support/BOINC Data/projects/www.primegrid.com/app_config.xml:
<app_config>
<app>
<name>llrSOB</name>
<fraction_done_exact/>
<max_concurrent>1</max_concurrent>
</app>
<app_version>
<app_name>llrSOB</app_name>
<cmdline>-t 4</cmdline>
<avg_ncpus>4</avg_ncpus>
<max_ncpus>4</max_ncpus>
</app_version>
</app_config>
BOINC is configured to read off web preferences from PG. It is configured to use at most 8 microprocessors.
With that in place, I explicitly request to read from config files, then quit and restart the app. I kill my existing CPU tasks and fetch new work, making sure to have only two WUs ready in the end. When BOINC starts crunching numbers, I look at OS X's Activity Monitor. I see only two processes called primegrid_llr running, each using 100% CPU and one thread.
Any thoughts about what I'm missing?
Won't the line:
<max_concurrent>X</max_concurrent>
only let the sub-project run X tasks at a time? In this case 1.
I know nothing about Apple, but it works that way on my Windows/Intel PC so I thought I would ask. This is used if you want to run multiple sub-projects from the same project concurrently, limiting how many of each sub-project runs at a time.
There is also a tag:
<project_max_concurrent>X</project_max_concurrent>
that will let a Project only run X tasks at a time. You can use it to run multiple projects at a time, limiting how many tasks of each project run concurrently.
Another use could be to download tasks in advance to prevent running out of tasks. For example, you could set this at 4 and leave 'Use CPU cores' set at 100% on a 4 core CPU with HT and have 4 tasks waiting to run, tho not recommended for PG. I do this with Enigma@Home a lot. Set <project_max_concurrent>1</project_max_concurrent> in the app_config.xml and leave my CPUs set to 100% while running the work buffer up so I get the max number of tasks I can before I hit the limit. I can run other projects this way and run Enigma for as long as possible while they have CPU tasks.
Hope some of that helps someone.
John T
|
8)
Message boards :
Number crunching :
World Cup Challenge
(Message 118547)
Posted 1821 days ago by HighTech67
I know nothing at all about multi-socket or Linux either, but I suspect this line might be the culprit:
<max_concurrent>1</max_concurrent>
This will only allow 1 task to run at a time as far as I know, at least on single CPU systems. If you change it to:
<max_concurrent>2</max_concurrent>
what happens?
|
9)
Message boards :
Number crunching :
Mathematics Awareness Month Challenge
(Message 116526)
Posted 1895 days ago by HighTech67
1) Run all cores and multithread - runs 12 PPSE tasks each taking approx 24 minutes
2) Run all cores Boinc set to 50% CPU - runs 6 PPSE tasks each taking approx 11 minutes
3) Run with Boinc set to 50% and app_config - runs 1 task taking approx 3 minutes 40 secs
So, for the challenge, do I want to do as many tasks as possible - option 2 or do I want to complete a task in the quickest time possible - option 3?
Dad,
Also look at it like this, knowing these are just PPSE tasks and as Michael said ESP is a lot longer and the results may not be the same. There should still be time for testing.
If you use option 3, how long will it take to finish the six tasks that option 2 can finish in about 11 minutes? IFF I calculated it correctly you're looking at over 20 minutes using option 3 for those six tasks, but I wouldn't trust my math any more, so double check it. It'll definitely be longer tho. Again, this is using the times you posted for PPSE tasks.
I've tested this on this computer (i7-2600) also and come up with the same results, even before I knew about multithreading. I can run certain projects faster with HT (6 at a time) than I can with HT turned off (3) at a time. Not individual tasks, but more tasks per time period (12 hours, 24 hours, 3 days...), challenge lengths.
Hope this helps. The two most important things are test, test, test. And you can't take data from one project and use on another with anything like reliable results.
Good luck,
John T
|
10)
Message boards :
[H]ard|OCP message board
(Message 87033)
Posted 2856 days ago by HighTech67
Absolutely slick that we have our own message board here.
Does this feature exist at other projects?
|
Next 10 posts
|
