Hi,


I found this post done by Mr. HA-Soft. Anybody who might help me twist it a little.

Need it for SETI@home on my Linux 64bit system.

I'm will be running cuda 5.5 on my test system 3 using a haswell cpu and 4x Nvidia GPU's. Therefore I would like to try this AVX as well. But I'm not sure about the files I need and where to find them. Please point me in the right direction and maybe fill in the blanks in the app_info.xml below. If I'm wrong about something, then I'm sorry. Its all pretty new to me ;):

Here is a complete app_info.xml for Windows 64 bit (SSE2, SSE3, AVX, CUDA55):

<app_info>

<app>
<name>period_search</name>
<user_friendly_name>Asteroids</user_friendly_name>
</app>

<!-- SSE2 -->
<file_info>
<name>period_search_10210_windows_x86_64__sse2.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>period_search</app_name>
<version_num>1021</version_num>
<avg_ncpus>1.00</avg_ncpus>
<max_ncpus>1.00</max_ncpus>
<plan_class>sse2</plan_class>
<file_ref>
<file_name>period_search_10210_windows_x86_64__sse2.exe</file_name>
<main_program/>
</file_ref>
</app_version>


<!-- SSE3 -->
<file_info>
<name>period_search_10210_windows_x86_64__sse3.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>period_search</app_name>
<version_num>1021</version_num>
<avg_ncpus>1.00</avg_ncpus>
<max_ncpus>1.00</max_ncpus>
<plan_class>sse3</plan_class>
<file_ref>
<file_name>period_search_10210_windows_x86_64__sse3.exe</file_name>
<main_program/>
</file_ref>
</app_version>


<!-- AVX -->
<file_info>
<name>period_search_10210_windows_x86_64__avx.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>period_search</app_name>
<version_num>1021</version_num>
<avg_ncpus>1.00</avg_ncpus>
<max_ncpus>1.00</max_ncpus>
<plan_class>avx</plan_class>
<file_ref>
<file_name>period_search_10210_windows_x86_64__avx.exe</file_name>
<main_program/>
</file_ref>
</app_version>

<!-- CUDA -->
<file_info>
<name>period_search_10100_windows_x86_64__cuda55.exe</name>
<executable/>
</file_info>
<file_info>
<name>cudart64_55.dll</name>
</file_info>
<app_version>
<app_name>period_search</app_name>
<version_num>1010</version_num>
<avg_ncpus>0.05</avg_ncpus>
<max_ncpus>1.00</max_ncpus>
<plan_class>cuda55</plan_class>
<file_ref>
<file_name>period_search_10100_windows_x86_64__cuda55.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>cudart64_55.dll</file_name>
<copy_file/>
</file_ref>
<coproc>
<type>CUDA</type>
<count>1.000000</count>
</coproc>
</app_version>

</app_info>

.

---

Project Headless CLI Linux Multiple GPU Boinc Servers
Ubuntu Server 14.04.1 64bit
Kernel 3.13.0-32-generic
CPU's i5-4690K
GPU's GT640/GTX750TI
Nvidia v.340.29
BOINC v.7.2.42

1) I'm almost sure that the lines <version_num>1021</version_num> have to be:
<version_num>10210</version_num>

(this is how it appears in client_state.xml)


2) With such app_info.xml I think you will receive only SSE2 CPU tasks (and CUDA tasks) as that is the first section.
Just delete unneeded CPU sections and keep only SSE3 or only AVX if AVX is supported by CPU and OS


3) Look here:
http://asteroidsathome.net/boinc/download/

... where you can:
- Download apps needed (if you don't have them already)
- See and copy the Linux names of executables to replace the '.exe' strings in app_info.xml with them, e.g.:
period_search_10210_windows_x86_64__sse3.exe --> period_search_10210_x86_64-pc-linux-gnu__sse3

---

- ALF - "Find out what you don't do well ..... then don't do it!" :)

Not sure why you want an app_info but then I'm app_info challenged.

If it helps... I'm getting the AVX app on my Haswell system without an app_info. All I did was plug it in and turn it on. HA_Soft says it's because Linux detects AVX properly but Windows can't find its own ass with both hands and a mirror. As usual.

You will get a mix of SSE2 and AVX tasks for a while until the server detects that your system does AVX way faster than SSE2 then from that point on you'll receive only AVX.

As for crunching here with your GPUs, unless you have Titans or Teslas or something with decent DP float power you'll find tasks run faster on Haswell than on GPU, I think. I tried my 670 here a while back and the times were not what I would call impressive and it's because DP is crippled on GTX 670. They have improved the GPU app at least once since then but if I understand correctly it's still slower than AVX on anything other than Titan and Tesla which have full (uncrippled) DP float capability.

Still, I didn't need an app_info for GPU either, just had to turn GPU on in website prefs.

---

BOINC FAQ Service
Official BOINC wiki
Installing BOINC on Linux

Not sure why you want an app_info but then I'm app_info challenged.

If it helps... I'm getting the AVX app on my Haswell system without an app_info. All I did was plug it in and turn it on. HA_Soft says it's because Linux detects AVX properly but Windows can't find its own ass with both hands and a mirror. As usual.

You will get a mix of SSE2 and AVX tasks for a while until the server detects that your system does AVX way faster than SSE2 then from that point on you'll receive only AVX.

As for crunching here with your GPUs, unless you have Titans or Teslas or something with decent DP float power you'll find tasks run faster on Haswell than on GPU, I think. I tried my 670 here a while back and the times were not what I would call impressive and it's because DP is crippled on GTX 670. They have improved the GPU app at least once since then but if I understand correctly it's still slower than AVX on anything other than Titan and Tesla which have full (uncrippled) DP float capability.

Still, I didn't need an app_info for GPU either, just had to turn GPU on in website prefs.

You're 100 % right, except that I don't think Asteroids uses DP. (see this post : http://asteroidsathome.net/boinc/forum_thread.php?id=268)
Like your point of vue about Windows :D
Ran some tests yesterday, and AVX "better" than GPU.
http://asteroidsathome.net/boinc/forum_thread.php?id=268

You're 100 % right, except that I don't think Asteroids uses DP.

Yes it does.
See this thread: http://asteroidsathome.net/boinc/forum_thread.php?id=131#1237

i am running 6 of these projects at a time. is there an app_info or app_config that where i can limit that from 6 to 4 ?

i am running 5 of these projects at a time. is there an app_info or app_config that where i can limit that from 5 to 4 ?

An app file for one project will not help you with another project. Simply choose the project you do not wish to crunch, select No new tasks, abort any you wish and problem solved.

i want to run 4 of these projects continuously. no more than 4.

i want to run 4 of these projects continuously. no more than 4.

Pick the one you do not want and select No new tasks, abort any remaining from that project, and you're done.

i want to see if my 'puter and boinc will run more efficiently if i set it to 4 with "on multiprocessor systems use at most 100% of the processors" setting

i want to see if my 'puter and boinc will run more efficiently if i set it to 4 with "on multiprocessor systems use at most 100% of the processors" setting

It won't.

It sounds like maybe you're thinking that if you have tasks from 5 projects then your computer crunches all 5 simultaneously. It doesn't. If you have a 4 core CPU and have "on multiprocessor systems use at most 100% of the processors" set then it will crunch only 4 tasks simultaneously.

---

BOINC FAQ Service
Official BOINC wiki
Installing BOINC on Linux

i want to see if my 'puter and boinc will run more efficiently if i set it to 4 with "on multiprocessor systems use at most 100% of the processors" setting

I see you're using an 8 core machine. If you want ro run only 4 WUs simultanously on it then use "on multiprocessor systems use at most 50% of the processors".

thank you

thank you

As Dagorath said though you will NOT be able to have 1 unit from each project running at the same time, Boinc does NOT work that way. Alot of people have asked for that over the years but the programmers say that doesn't fit with their 'master plan'. Now VERY RARELY it could work out that way, but normally you will see units from one project running while another projects units are either paused and waiting to run or just in the queue. It all depends on how long it takes the units to run and your settings.

i didn't think it would be a big deal to suggest running 4 at the same time for one project and running 2 of the another at the same time.
will there be improvement made for the "in queue" process in the future? seems i noticed a delay in some projects in the past, mainly gpu projects. maybe it's fine.

app_config.xml for cpu wus on this project. It gets placed in the project folder. This version will limit the cpu tasks to 4 concurrent.


<app_config>
<app>
<name>period_search</name>
<max_concurrent>4</max_concurrent>
</app>
</app_config>

If you want to control cpu and gpu tasks, you will need to expand the above to include the gpu setup. I haven't tried them on this project so I don't have the setup for the gpu handy.

Forgot to mention you need to be on one of the newer boinc clients for it to work, 7.0.40 or newer I think (might be 42).

i didn't think it would be a big deal to suggest running 4 at the same time for one project and running 2 of the another at the same time.
will there be improvement made for the "in queue" process in the future? seems i noticed a delay in some projects in the past, mainly gpu projects. maybe it's fine.

No, it is apparently not on the drawing board right now.

app_config.xml for cpu wus on this project. It gets placed in the project folder. This version will limit the cpu tasks to 4 concurrent.


<app_config>
<app>
<name>period_search</name>
<max_concurrent>4</max_concurrent>
</app>
</app_config>

If you want to control cpu and gpu tasks, you will need to expand the above to include the gpu setup. I haven't tried them on this project so I don't have the setup for the gpu handy.

I think we are talking about running 2 Seti unit and 2 Asteroid units, for example, all at the same time endlessly, not 4 units from any project at once. So in effect you would ALWAYS run 2 Seti units AND also ALWAYS run 2 Asteroid units. What I am saying is that this is NOT possible right now, unless it just happens by pure luck and happenstance.

thanks for all of the feedback if you understood what i was talking about or not. i am satisfied with all of the feedback.