hi guys!

I have retired my HD4850 now for the moment and want to bring MW from a mainproject to a backupproject on one of the POEM Cruncher (so it will run MW often enough :D ) with a HD7950. The problem is, when POEM has a resourceshare of 100% and MW 0% to get it run only when POEM is really without Work, i normaly must use the backupproject without Appinfo when the mainproject on the machine uses an appinfo. Because otherwise its mixing the wus from both projects with 0.33 or 0.25 GPU Values as example. BUT the card is that fast, it crunches a single MW unit faster then the sheduler allowes to report the workunit, so the use of the card is 50% computing and 50% waiting before the 60seconds shedulerwait is over to report and get a new single workunit. So i have now running POEM GPUs with a appinfo with 4 Units (Quadcore cpu) and MW as limited backupproject without appinfo. Hope it is understandable what my problem is, my english is not the best for such explaination ^^ Are there any ways to get MW running to with an appinfo without mixing with POEM?

---

DSKAG Austria Research Team: http://www.research.dskag.at

You don't use an app_info.xml for this project, but an app_config.xml. First I would disable n-body as it will use every CPU core to run a work unit and it does not use the graphics card to compute yet.

app_config.xml goes in the same position as app_info.xml would. Make sure you have the latest BOINC (7.065).

<app_config>
<app>
<name>milkyway</name>
<max_concurrent>4</max_concurrent>
<gpu_versions>
<gpu_usage>.25</gpu_usage>
<cpu_usage>.1</cpu_usage>
</gpu_versions>
</app>
</app_config>

Make sure it's properly formatted in your text editor or else boinc will misread it.

I'm doing this from Linux with my 7950.

I posted this thread to both projects forums. It seems i miswrote something :/ The problem was more the transistion between both projects where it seems to mix something up.

I deactivaded nbody, but i dont think i got anytime these to compute and changed the config to 300% POEM and 2% MW with reactivaed appinfo for MW. I wish POEM would ask more often the sheduler for tasks, so MW takes almost the most computingtime.

---

DSKAG Austria Research Team: http://www.research.dskag.at

I posted this thread to both projects forums. It seems i miswrote something :/ The problem was more the transistion between both projects where it seems to mix something up.

I deactivaded nbody, but i dont think i got anytime these to compute and changed the config to 300% POEM and 2% MW with reactivaed appinfo for MW. I wish POEM would ask more often the sheduler for tasks, so MW takes almost the most computingtime.

I think the problems lie in the 300% and the 2%, 2 out of 300 is NOT a very big number and is only a little better then 0 out of 300. On my laptop I am using 4 cpu cores and have one project set at 50% and one at 10%. Normally one project crunches 3 units while the other crunches 1 unit, but sometimes all 4 cores are crunching for the 50% project.

When I was at POEM, my monstrous array of GPUs would devour all work units I'd be given in under 30 minutes and be left for hours without units, so I ditched POEM at the time. I spent the last year bitcoin mining with my 7950, which earned me A good 60 coins ($6,600+ on current exchange rate). Now I'm back to running them on BOINC.

I think the problems lie in the 300% and the 2%, 2 out of 300 is NOT a very big number and is only a little better then 0 out of 300. On my laptop I am using 4 cpu cores and have one project set at 50% and one at 10%. Normally one project crunches 3 units while the other crunches 1 unit, but sometimes all 4 cores are crunching for the 50% project.

It was 100/0
Now it is 300/2

I wish i would understand the exact! meaning of the values ^^ on cpu side i know i can control a bit the usage of a multicore cpu how much cores every project get. But on GPU O.o

---

DSKAG Austria Research Team: http://www.research.dskag.at

I think the problems lie in the 300% and the 2%, 2 out of 300 is NOT a very big number and is only a little better then 0 out of 300. On my laptop I am using 4 cpu cores and have one project set at 50% and one at 10%. Normally one project crunches 3 units while the other crunches 1 unit, but sometimes all 4 cores are crunching for the 50% project.

It was 100/0
Now it is 300/2

I wish i would understand the exact! meaning of the values ^^ on cpu side i know i can control a bit the usage of a multicore cpu how much cores every project get. But on GPU O.o

I am not sure that has settled down to a real set of finalized specs yet. Gpu crunching is only a few years old and still has a lot of tweaking yet to go. In short I think the specs are in flux much more then the cpu specs are. Although to be honest even the way the cpu is being used is still being tweaked a bit. I have 'heard' that the new Boinc 8 will have more cpu control in it, but I CANNOT find the post right now!!

I guess you run appinfo to allow multiple wu's at the same time.

My suggestion would be to use app_info.xml instead.

My app_info.xml for POEM looks like this.

<app_info>

<app>
<name>poemcl</name>
<user_friendly_name>POEM++ OpenCL</user_friendly_name>
</app>

<file_info>
<name>poemcl_1.5_windows_intelx86__opencl_ati_100</name>
<executable/>
</file_info>

<app_version>
<app_name>poemcl</app_name>
<version_num>1</version_num>
<plan_class>opencl_ati_100</plan_class>
<avg_ncpus>1</avg_ncpus>
<max_ncpus>1</max_ncpus>
<flops>209876543210</flops>
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>
<cmdline></cmdline>
<file_ref>
<file_name>poemcl_1.5_windows_intelx86__opencl_ati_100</file_name>
<main_program/>
</file_ref>
</app_version>

</app_info>

And you can use the line <count>1</count> <count>0.5</count> or <count>0.25</count> to allow 1,2 and 4 wu's at the same time.

Then with Milkyway....

I have an app_info.xml here

<app_info>
<app>
<name>milkyway</name>
</app>
<file_info>
<name>milkyway_separation_1.02_windows_intelx86__opencl_amd_ati.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>milkyway</app_name>
<version_num>102</version_num>
<flops>1.0e11</flops>
<avg_ncpus>0.05</avg_ncpus>
<max_ncpus>1</max_ncpus>
<plan_class>ati14ati</plan_class>
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>
<cmdline></cmdline>
<file_ref>
<file_name>milkyway_separation_1.02_windows_intelx86__opencl_amd_ati.exe</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>

Where you have a <count>1</count> and you can do the same 1,2 or 4

If you set both projects to 0.5 you will have 1POEM task and 1 Milkyway task running at the same time.

You will have to find which numbers are the best for you.

Download Milkyway optimized app with app_info.xml from arkayns site:
http://www.arkayn.us/forum/index.php?PHPSESSID=1128fe3162aea106a804657cf44ea433&action=tpmod;dl

I think the best thing is to have the same count number for booth projects.
But that's me, maybe you can find a better combination, maybe use appconfig.xml and app_info for POEM and only app_info.xml for MW. Find what is best for you.

And if you run both projects with count 0.5 maybe resource 100/100 can be a good thing.

And you can use cc_config.xml file options to speed some things up a bit.

An example:

<cc_config>
<options>
<rec_half_life_days>0.25</rec_half_life_days>
<http_transfer_timeout>600</http_transfer_timeout>
<report_results_immediately>1</report_results_immediately>
<max_file_xfers>12</max_file_xfers>
<max_file_xfers_per_project>6</max_file_xfers_per_project>
</options>
</cc_config>


The <rec_half_life_days> are default 10
And is used to balance the resource share so boinc know what project it will download next.

The smaller number the faster balance. I choose 0.25(I think it should be default).

<http_transfer_timeout> is default 600seconds, I don't know why I keep that line...

<report_results_immediately> 1 is to report result as soon as wu is uploaded and ready to report , Then if needed boinc asks for more wu's and that can be good for filling up workcache.

<max_file_xfers>12</max_file_xfers>
<max_file_xfers_per_project>6</max_file_xfers_per_project>

Is to download more wu's at a time. Default is 2

I hope this can be of any help.

cc_config.xml is located in C:\Programdata\BOINC folder and has to be created with notepad or similiar.

//TRuEQ

lol.. For the moment thanks for the answers, i will try again in July or something again. WCG anounced suprising the end of HPF2 in few weeks so i must switch this machine to MW as mainproject back again to get the Quad CPU 100% free from POEM to run WCG ^^

---

DSKAG Austria Research Team: http://www.research.dskag.at