Message boards : Number crunching : How is the SSE3 thing coming along?
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · Next

AuthorMessage
MB Atlanos

Send message
Joined: 14 Jul 05
Posts: 11
Credit: 81,274
RAC: 0
Message 24245 - Posted: 11 Jul 2012, 0:30:12 UTC

And a old lowend netbook for comparsion:

10/07/2012 09:00:43 | | Starting BOINC client version 7.0.28 for windows_intelx86
10/07/2012 09:00:43 | | Processor: 2 GenuineIntel Intel(R) Atom(TM) CPU N270 @ 1.60GHz [Family 6 Model 28 Stepping 2]
10/07/2012 09:00:43 | | Processor: 512.00 KB cache
10/07/2012 09:00:43 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 nx tm2 movebe pbe
10/07/2012 09:00:43 | | OS: Microsoft Windows XP: Home x86 Edition, Service Pack 3, (05.01.2600.00)

the gain from the faster sse3 app would be significant for this pc.
ID: 24245 · Report as offensive     Reply Quote
angler

Send message
Joined: 25 Nov 06
Posts: 25
Credit: 4,686,113
RAC: 0
Message 24263 - Posted: 11 Jul 2012, 16:09:31 UTC

looks like linux Athlon 3300+ is now using sse2 image, only one laptop still using gen image even though sse2 capable so far

have newer pc's on Windows using sse2, sse3 and pni versions

ID: 24263 · Report as offensive     Reply Quote
Mattia Verga

Send message
Joined: 27 Sep 04
Posts: 20
Credit: 23,880
RAC: 0
Message 24264 - Posted: 11 Jul 2012, 16:14:30 UTC
Last modified: 11 Jul 2012, 16:14:48 UTC

I saw that a new PNI app is available, however my PC has just downloaded a SSE2 one: http://lhcathomeclassic.cern.ch/sixtrack/show_host_detail.php?hostid=9932051

My current wingman has a linux PC with SSE3 app: http://lhcathomeclassic.cern.ch/sixtrack/show_host_detail.php?hostid=9966949

But if you look at the task list of that PC it shows that the first WU was SSE3, the second a plain one, and now another SSE3... it seems to me that hosts are just downloading random versions...
ID: 24264 · Report as offensive     Reply Quote
Keith T.
Avatar

Send message
Joined: 1 Mar 07
Posts: 47
Credit: 32,356
RAC: 0
Message 24265 - Posted: 11 Jul 2012, 16:22:13 UTC
Last modified: 11 Jul 2012, 16:26:50 UTC

Can I force my SSE3 capable computer to use SSE3 file using app_info.xml ?

I have used optimized apps at SETI@home and understand the basics of file editing, but I'm not 100% sure what detail to put in.

[edit]
I've also manually downloaded the SSE3 and PNI files from the server, but they show as file instead of .exe


11/07/2012 17:13:37 Starting BOINC client version 6.10.58 for windows_intelx86
11/07/2012 17:13:37 log flags: file_xfer, sched_ops, task
11/07/2012 17:13:37 Libraries: libcurl/7.19.7 OpenSSL/0.9.8l zlib/1.2.3
11/07/2012 17:13:37 Running as a daemon
11/07/2012 17:13:37 Data directory: C:\Documents and Settings\All Users\Application Data\BOINC
11/07/2012 17:13:37 Running under account boinc_master
11/07/2012 17:13:38 Processor: 2 GenuineIntel Intel(R) Atom(TM) CPU N270 @ 1.60GHz [Family 6 Model 28 Stepping 2]
11/07/2012 17:13:38 Processor: 512.00 KB cache
11/07/2012 17:13:38 Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 tm2 movebe pbe
11/07/2012 17:13:38 OS: Microsoft Windows XP: Home x86 Edition, Service Pack 3, (05.01.2600.00)
11/07/2012 17:13:38 Memory: 1014.36 MB physical, 2.39 GB virtual
11/07/2012 17:13:38 Disk: 71.04 GB total, 35.37 GB free
11/07/2012 17:13:38 Local time is UTC +1 hours
11/07/2012 17:13:38 No usable GPUs found
11/07/2012 17:13:38 SETI@home Found app_info.xml; using anonymous platform
11/07/2012 17:13:38 LHC@home 1.0 URL http://lhcathomeclassic.cern.ch/sixtrack/; Computer ID 9766664; resource share 100
ID: 24265 · Report as offensive     Reply Quote
Tex1954

Send message
Joined: 24 Apr 11
Posts: 37
Credit: 1,295,012
RAC: 0
Message 24268 - Posted: 11 Jul 2012, 17:32:43 UTC
Last modified: 11 Jul 2012, 17:33:46 UTC

Well, I just enabled LHC again since the new batch this morning and I am pleased to report that I received SSE2 and (pni) tasks on a linux 1090T box. Also, I received SSE3 on the i7-950 box and all SSE2 on another Windows-7 1055T box.

Looks like things are getting sorted out well...

I am very pleased ya'll are making such good (almost realtime) progress!

Keep up the good work and find out how to get warp speed so that Scotty can beam me up!

:)
ID: 24268 · Report as offensive     Reply Quote
Keith T.
Avatar

Send message
Joined: 1 Mar 07
Posts: 47
Credit: 32,356
RAC: 0
Message 24270 - Posted: 11 Jul 2012, 18:11:56 UTC - in response to Message 24265.  

Can I force my SSE3 capable computer to use SSE3 file using app_info.xml ?


Well I gave it a try, seems to be working so far !

Here is the app_info.xml file I created:

<app_info> 
    <app>
	<name>sixtrack</name>
            </app>
    <file_info>
        <name>sixtrack_windows_sse3</name>
        <executable/>
    </file_info>
    <app_version>
	<app_name>sixtrack</app_name>
        <version_num>44401</version_num>
        <file_ref>
	   <file_name>sixtrack_windows_sse3</file_name>
            <main_program/>
        </file_ref>
    </app_version>    
</app_info> 


My original file had [Tabs], but they have disappeared when I posted this. I think the file should still work though if you want to try it.

I downloaded the executable from http://lhcathomeclassic.cern.ch/sixtrack/download/ using Chrome, then copied the file into C:\Documents and Settings\All Users\Application Data\BOINC\projects\lhcathomeclassic.cern.ch_sixtrack

I exited BOINC, copied my app_info.xml into the same folder. Restarted BOINC, and the LHC tasks which had been downloaded for sixtrack_windows_gen started using sixtrack_windows_sse3.

Checked using Windows Task Manager that SSE3 app was running.
Also checked BOINC startup messages with:

11/07/2012 18:30:11 LHC@home 1.0 Found app_info.xml; using anonymous platform

Keith
ID: 24270 · Report as offensive     Reply Quote
Eric Mcintosh
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 12 Jul 11
Posts: 857
Credit: 1,619,050
RAC: 0
Message 24295 - Posted: 12 Jul 2012, 9:14:45 UTC - in response to Message 24270.  

Well that is inetersting and a fine piece of work.
You shouldn't have to that of course, but it is a
big help. This bit of BOINC requires a re-compile and re-install
and Igor is really working on it. It is important to
have at least SSE2 for performance. Now I know nearly every box
has SSE2, but a major featore of our SixTrack is that it can run
on ANY PC (I hope) even old ones and produce identical results.
I don't want to drop the generic version and just lose those
volunteers with old systems. Eric.
ID: 24295 · Report as offensive     Reply Quote
Profile Gary Roberts

Send message
Joined: 22 Jul 05
Posts: 72
Credit: 3,962,626
RAC: 0
Message 24299 - Posted: 12 Jul 2012, 11:07:10 UTC - in response to Message 24295.  

... You shouldn't have to (do) that of course ...

Sure enough, but if the project doesn't send anything but the generic executable, there is little other option :-).

In my case I've attached 15 hosts to the project, all of which are sse3 capable. They all received the generic executable and all were crunching very slowly. As soon as the versions appeared back in the download area - about 24 hrs ago - I grabbed copies of the sse3 versions. I have mostly linux hosts with a couple of Windows XP ones. Most of the hosts are quads - mainly Intel, with a couple of Phenom IIs.

To initiate AP processing, you need to construct a suitable app_info.xml file and place it with the new executable in the project directory. BOINC needs to be restarted to force it to 'see' app_info.xml and if you haven't constructed that file correctly, you will very likely trash your entire cache of tasks at that point. For that reason, people often set NNT and complete the existing cache before switching to AP mode.

I was in a hurry and I made the assumption that any version of the app would be able to read the last checkpoint saved by a different version and I also assumed (since you had said that all versions give numerically identical answers) that a part answer from one version, then completed with a different version, would hopefully give a final result that would still validate.

So all my sse3 capable hosts were transitioned from the generic version as quickly as possible. All had caches with many 'previous version' tasks in them. The _sse3 version has been running for at least the last 12 hours and all the 'two version' results I've looked at so far seem to be validating quite OK. The speedup using the sse3 version is quite impressive!

Hopefully, you will be able to come up with a reliable mechanism for detecting the CPU capabilities and the manual effort of managing AP can be dispensed with.


Cheers,
Gary.
ID: 24299 · Report as offensive     Reply Quote
Uffe F

Send message
Joined: 9 Jan 08
Posts: 66
Credit: 727,923
RAC: 0
Message 24300 - Posted: 12 Jul 2012, 11:27:44 UTC
Last modified: 12 Jul 2012, 11:28:12 UTC

I got almost exclusively SSE3 WU's. I only got 1 single WU that was the normal one on this host: http://lhcathomeclassic.cern.ch/sixtrack/show_host_detail.php?hostid=9964314

So looking like the distribution works. Atleast on my 2 hosts.

Looking at the time used then i can see that the SSE3 tasks uses only 80% of the time that a normal task does. So for my setup the SSE3 tasks are 25% faster.

Thanks for the good work on optimizing it :)
ID: 24300 · Report as offensive     Reply Quote
Tex1954

Send message
Joined: 24 Apr 11
Posts: 37
Credit: 1,295,012
RAC: 0
Message 24305 - Posted: 12 Jul 2012, 15:03:15 UTC - in response to Message 24300.  

Just got 18 of the slow tasks on the 1055T box... the ones that take > 7 hours?

LOL!

Seems random to me still...

:)
ID: 24305 · Report as offensive     Reply Quote
Profile Robert Pick

Send message
Joined: 1 Dec 05
Posts: 62
Credit: 11,398,274
RAC: 261
Message 24306 - Posted: 12 Jul 2012, 15:03:37 UTC

Low and behold!!!! I got 3 tasks this morning, in it's self is not unusual but one was an sse2, two pni, oops just got two sse3's! I've done nothing to my sys. to exacerbate the execution of this upgrade. Appariantly someone at LHC has tinkered with the code. YAA!!! Thank you Pick
ID: 24306 · Report as offensive     Reply Quote
Profile Tom95134

Send message
Joined: 4 May 07
Posts: 250
Credit: 826,541
RAC: 0
Message 24310 - Posted: 12 Jul 2012, 16:36:24 UTC
Last modified: 12 Jul 2012, 16:36:53 UTC

Good news! I "pumped" the Project and got two new Six Track 444.01 (sse3) Tasks. They show long Remaining times (35+ hrs) but it has already come down from 43+ as BOINC estimates the work better.

Running Windows 7 (x64), Inteli7-2600, 16GB RAM.

Tom
ID: 24310 · Report as offensive     Reply Quote
Cash

Send message
Joined: 13 Jul 05
Posts: 21
Credit: 456,769
RAC: 0
Message 24314 - Posted: 12 Jul 2012, 18:51:43 UTC

Just adjusted my app_info.xml to get sse3 instead of pni on my Phenom X6.
Btw, noticed a strange thing: at least once my computer got both of task WUs - http://lhcathomeclassic.cern.ch/sixtrack/workunit.php?wuid=1856494. I think it should NOT work like that (this was way before tinkering with xml)

this game has no name
ID: 24314 · Report as offensive     Reply Quote
Zapped Sparky

Send message
Joined: 22 Oct 08
Posts: 26
Credit: 75,214
RAC: 0
Message 24319 - Posted: 12 Jul 2012, 21:37:44 UTC - in response to Message 24310.  

Good news! I "pumped" the Project and got two new Six Track 444.01 (sse3) Tasks. They show long Remaining times (35+ hrs) but it has already come down from 43+ as BOINC estimates the work better.

Running Windows 7 (x64), Inteli7-2600, 16GB RAM.

Tom

Just got my first sse3 tasks myself, however the estimate is in the other direction, 1hr20m to completion, looks like they'll take about to 7hrs to complete. Still glad to be crunching though :)
ID: 24319 · Report as offensive     Reply Quote
Teilchen

Send message
Joined: 8 Jun 07
Posts: 13
Credit: 250,850
RAC: 0
Message 24325 - Posted: 13 Jul 2012, 7:04:19 UTC

I have many compute errors with sse3.
ID: 24325 · Report as offensive     Reply Quote
Toby Broom
Volunteer moderator

Send message
Joined: 27 Sep 08
Posts: 798
Credit: 644,731,442
RAC: 233,853
Message 24326 - Posted: 13 Jul 2012, 12:26:12 UTC - in response to Message 24325.  

I have many compute errors with sse3.


Your errors are the same as mine, the maxiumium elapsed time.
ID: 24326 · Report as offensive     Reply Quote
angler

Send message
Joined: 25 Nov 06
Posts: 25
Credit: 4,686,113
RAC: 0
Message 24329 - Posted: 13 Jul 2012, 14:50:59 UTC - in response to Message 24314.  

same thing happened to me seemed to be a server side issue


btw have an app_info which will keep existing pni (identical to sse3) and sse2 jobs flowing.

<app_info>
<app>
<name>sixtrack</name>
</app>
<file_info>
<name>sixtrack_windows_pni</name>
<executable/>
</file_info>
<file_info>
<name>sixtrack_windows_sse2</name>
<executable/>
</file_info>
<file_info>
<name>sixtrack_windows_sse3</name>
<executable/>
</file_info>
<app_version>
<app_name>sixtrack</app_name>
<version_num>44401</version_num>
<plan_class>pni</plan_class>
<file_ref>
<file_name>sixtrack_windows_pni</file_name>
<main_program/>
</file_ref>
</app_version>
<app_version>
<app_name>sixtrack</app_name>
<version_num>44401</version_num>
<file_ref>
<file_name>sixtrack_windows_sse3</file_name>
<main_program/>
</file_ref>
</app_version>
<app_version>
<app_name>sixtrack</app_name>
<version_num>44401</version_num>
<plan_class>sse2</plan_class>
<file_ref>
<file_name>sixtrack_windows_sse2</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>
ID: 24329 · Report as offensive     Reply Quote
achim.huber

Send message
Joined: 13 Jul 05
Posts: 5
Credit: 451,738
RAC: 0
Message 24331 - Posted: 13 Jul 2012, 16:44:09 UTC

I received 8 SSE3 WUs. All ended in a calculation error after a runtime of 13,518 seconds.
ID: 24331 · Report as offensive     Reply Quote
achim.huber

Send message
Joined: 13 Jul 05
Posts: 5
Credit: 451,738
RAC: 0
Message 24337 - Posted: 13 Jul 2012, 20:37:27 UTC - in response to Message 24331.  

13.07.2012 13:24:49 | LHC@home 1.0 | Aborting task w1jul_niebb1d__39__s__64.28_59.31__9.8_10__6__28_1_sixvf_boinc68469_1: exceeded elapsed time limit 13518.14 (1800000.00G/133.15G)
13.07.2012 13:38:26 | LHC@home 1.0 | Aborting task w1jul_niebb1d__39__s__64.28_59.31__9.8_10__6__24_1_sixvf_boinc68465_1: exceeded elapsed time limit 13518.14 (1800000.00G/133.15G)
13.07.2012 13:49:34 | LHC@home 1.0 | Aborting task w1jul_niebb1d__39__s__64.28_59.31__9.8_10__6__26_1_sixvf_boinc68467_1: exceeded elapsed time limit 13518.14 (1800000.00G/133.15G)
13.07.2012 13:59:44 | LHC@home 1.0 | Aborting task w1jul_niebb1d__39__s__64.28_59.31__9.8_10__6__25_1_sixvf_boinc68466_1: exceeded elapsed time limit 13518.14 (1800000.00G/133.15G)
13.07.2012 17:23:46 | LHC@home 1.0 | Aborting task w1jul_niebb1d__39__s__64.28_59.31__9.8_10__6__30_1_sixvf_boinc68471_1: exceeded elapsed time limit 13518.14 (1800000.00G/133.15G)

Why is there a time limit that makes it impossible to finish a Workunit? I've set BOINC to "No new Tasks" until this gets sorted out.
ID: 24337 · Report as offensive     Reply Quote
boroda3

Send message
Joined: 13 Mar 12
Posts: 4
Credit: 205,048
RAC: 0
Message 24344 - Posted: 14 Jul 2012, 0:27:06 UTC
Last modified: 14 Jul 2012, 0:38:44 UTC

I have SSE2 and PNI WU's.

Task / Run Time / App

4148834 / 15,002.69 / SixTrack v444.01 (pni)
4148831 / 14,962.90 / SixTrack v444.01 (pni)
4148621 / 15,884.89 / SixTrack v444.01 (pni)

4168474 / 887.66 / SixTrack v444.01 (sse2)
4168472 / 1,394.16 / SixTrack v444.01 (sse2)
4168470 / 1,023.10 / SixTrack v444.01 (sse2)

BOINC 7.0.28, Win7 x86, AMD Phenom II X4 955

You say, that SSE3/PNI faster? Hmm...

Again problems with optimization for AMD processor? Don't use Intel compiler.
ID: 24344 · Report as offensive     Reply Quote
Previous · 1 · 2 · 3 · Next

Message boards : Number crunching : How is the SSE3 thing coming along?


©2024 CERN