Are there any sites providing optimized clients? -- PART II

Message boards : Number crunching : Are there any sites providing optimized clients? -- PART II
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 . . . 19 · Next

AuthorMessage
Profile SargeD@SETI.USA
Volunteer tester
Avatar

Send message
Joined: 24 Nov 02
Posts: 957
Credit: 3,848,754
RAC: 0
United States
Message 347902 - Posted: 24 Jun 2006, 22:13:34 UTC - in response to Message 347896.  

Thanks for the clarification Sarge!

No problem. :)

ID: 347902 · Report as offensive
Profile KWSN - Chicken of Angnor
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 9 Jul 99
Posts: 1199
Credit: 6,615,780
RAC: 0
Austria
Message 347967 - Posted: 25 Jun 2006, 0:18:48 UTC - in response to Message 347806.  
Last modified: 25 Jun 2006, 0:25:22 UTC

Stats for second WU:

testWU-2 (AR: 0.4437317022742)

Windows 32-Bit

Default 5.15 with graphics
27m 37s (1657 seconds)

Crunch3r 5.12 SSE2
13m 04s (784 seconds) - 52.7% quicker

My 5.15 SSE2
13m 10s (790 seconds) - 52.6% quicker

Those 6 seconds difference are again well within standard result variance, so I'd calll it a draw at this AR on Windows (between optimized clients), too.

Linux 32-Bit

Default 5.12 (no X-Win installed, so no graphics? not sure)
23m 04s (1384 seconds)

Crunch3r 5.12 SSE2
16m 34s (994 seconds) - 28.2% quicker

My 5.15 SSE2
14m 52s (892 seconds) - 35.6% quicker

--------------------------------

Linux vs. Windows

Default client
Linux is 16.5% quicker.

Crunch3rs 5.12
Windows is 21.1% quicker.

My 5.15 builds
Windows is 11.4% quicker.

So the results seem to be pretty repeatable for the first two.
Next one coming right up.

Okay, here is the first round of test results. Bear in mind a test run takes me around 1-2 hours all in all (with collating results and rebooting twice).
The chirp_limit has been adjusted (to 1/2.5) so the units take comparatively less time, or one run would take 12-20 hours.

testWU-1 (AR: 0.60488405578199)

Windows 32-Bit

Default 5.15 with graphics
23m 33s (1411 seconds)

Crunch3r 5.12 SSE2
11m 30s (690 seconds) - 51.1% quicker

My 5.15 SSE2
11m 25s (685 seconds) - 51.5% quicker

Those 5 seconds difference are well within standard result variance, so I'd calll it a draw at this AR on Windows.

Linux 32-Bit

Default 5.12 (no X-Win installed, so no graphics? not sure)
19m 30s (1150 seconds)

Crunch3r 5.12 SSE2
14m 16s (856 seconds) - 25.57% quicker
My 5.15 SSE2
12m 49s (769 seconds) - 33.13% quicker

--------------------------------

Linux vs. Windows

Default client
Linux is 18.5% quicker.

Crunch3rs 5.12
Windows is 19.4% quicker.

My 5.15 builds
Windows is 12.2% quicker.

So to sum things up, Windows does seem quicker at this AR with optimized clients, but slower with the default (due to the graphics? probably).

Next unit I'll test has an AR of ~0.44, so scaling should be different there.

Regards,
Simon.


Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information
ID: 347967 · Report as offensive
Profile KWSN - Chicken of Angnor
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 9 Jul 99
Posts: 1199
Credit: 6,615,780
RAC: 0
Austria
Message 347982 - Posted: 25 Jun 2006, 0:33:03 UTC

Stats for third WU:

testWU-4 (AR: 1.2796485198966)

Windows 32-Bit

Default 5.15 with graphics
10m 46s (646 seconds)

Crunch3r 5.12 SSE2
4m 19s (259 seconds) - 59.9% quicker

My 5.15 SSE2
4m 17s (257 seconds) - 60.0% quicker

Those 2 seconds difference are again well within standard result variance, so I'd call it a draw at this AR on Windows (between optimized clients), too.

Linux 32-Bit

Default 5.12 (no X-Win installed, so no graphics? not sure)
8m 00s (480 seconds)

Crunch3r 5.12 SSE2
5m 02s (302 seconds) - 37.1% quicker

My 5.15 SSE2
4m 45s (285 seconds) - 40.6% quicker

--------------------------------

Linux vs. Windows

Default client
Linux is 25.7% quicker.

Crunch3rs 5.12
Windows is 14.2% quicker.

My 5.15 builds
Windows is 9.8% quicker.

The same sort of scaling seems to apply, roughly.

Regards,
Simon.
Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information
ID: 347982 · Report as offensive
Alex Kan
Volunteer developer

Send message
Joined: 4 Dec 03
Posts: 127
Credit: 29,269
RAC: 0
United States
Message 348003 - Posted: 25 Jun 2006, 0:56:48 UTC

Simon, are you planning on making your benchmark system available to the public at some point? I'm sure other people would be more than willing to help contribute benchmarks, not to mention that I'm kind of curious as to how a G5 running the Mac client stacks up. :P

Also, are these tests still being run on the 3.06 GHz P4 laptop you mentioned before?
ID: 348003 · Report as offensive
Ulrich Metzner
Volunteer tester
Avatar

Send message
Joined: 3 Jul 02
Posts: 1256
Credit: 13,565,513
RAC: 13
Germany
Message 348005 - Posted: 25 Jun 2006, 1:07:28 UTC

Well, this all sounds very, very interesting to us all, but where are the clients to download and test this whole things for us all? ;)
Aloha, Uli

ID: 348005 · Report as offensive
Profile KWSN - Chicken of Angnor
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 9 Jul 99
Posts: 1199
Credit: 6,615,780
RAC: 0
Austria
Message 348008 - Posted: 25 Jun 2006, 1:14:27 UTC
Last modified: 25 Jun 2006, 1:31:55 UTC

No,

these are all being run on a Pentium-D 805 right now (stock).

Also, because the computer store didn't have two Corsair modules, it's only running single channel until they can restock. Still, it makes for very comparable results.

I installed Windows XP 32 and 64 Bit, and Linux 32 and 64 Bit (so a total of 4 Operating Systems) on this computer. This should enable me to do repeatable results without too much variance, because this system is doing nothing else.

Of course you can have the WUs I'm using to crunch these tests as well as the shell scripts I made to compare clients.

The WUs can be found here along with a little benchmark shell script.

Uli, I'll put together an SSE2 Linux package after I had some more sleep. Check the announcements thread tomorrow.

Regards,
Simon.
Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information
ID: 348008 · Report as offensive
Ulrich Metzner
Volunteer tester
Avatar

Send message
Joined: 3 Jul 02
Posts: 1256
Credit: 13,565,513
RAC: 13
Germany
Message 348011 - Posted: 25 Jun 2006, 1:18:51 UTC - in response to Message 348008.  

(...) Uli, I'll put together an SSE2 package after I had some more sleep. Check the announcements thread tomorrow.
Thanks, i'll come after that on monday, cause the SSE3 machine is out of reach right now, and now i only have a very decent 3Dnow! and a retired SSE machine at hand ;)

Aloha, Uli

ID: 348011 · Report as offensive
Profile KWSN - Chicken of Angnor
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 9 Jul 99
Posts: 1199
Credit: 6,615,780
RAC: 0
Austria
Message 348659 - Posted: 25 Jun 2006, 23:50:52 UTC
Last modified: 25 Jun 2006, 23:51:08 UTC

ID: 348659 · Report as offensive
Profile KWSN - Chicken of Angnor
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 9 Jul 99
Posts: 1199
Credit: 6,615,780
RAC: 0
Austria
Message 348677 - Posted: 26 Jun 2006, 0:29:17 UTC

During testing, I've found that the Linux default client I've been using to test doesn't contain graphics and is between 10 and 30% quicker than Windows with graphics!

If it's not too much of a stretch code-wise, it would really be a cool feature to be able to turn off the graphics via the BOINC manager, for example. Right now, if they're compiled into the client, they are ALWAYS drawn, not only when you're actually looking at them (when screensaver runs or you select "view graphics" in the manager).

It really seems a waste of perfectly good cycles even if you don't use an optimized client.

Regards,
Simon.
Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information
ID: 348677 · Report as offensive
Pepperammi

Send message
Joined: 3 Apr 99
Posts: 200
Credit: 737,775
RAC: 0
United Kingdom
Message 348704 - Posted: 26 Jun 2006, 1:08:45 UTC
Last modified: 26 Jun 2006, 1:10:38 UTC

@ KWSN - Chicken of Angnor

Thanks for all yor work. Just noticed your linux app availible. I'm sure you've made a lot of happy people :)

Didn't know the graphics were always drawn. Hopefully in the future they'll be seperated from the main app then it would be simple to just delete them if you don't want them or, as you say, an option in boinc would be nice.

Thanks again. Looking forward to your windows version being ready for release.
ID: 348704 · Report as offensive
Profile Gecko
Volunteer tester
Avatar

Send message
Joined: 17 Nov 99
Posts: 454
Credit: 6,946,910
RAC: 47
United States
Message 348740 - Posted: 26 Jun 2006, 1:41:23 UTC - in response to Message 348659.  
Last modified: 26 Jun 2006, 1:42:08 UTC

Bam.

'nuff said :o)
Regards,
Simon.



Fantastic job Simon!
Congratulations and THANKS!!!.....and not only for an excellent opt. ap, but MOST importantly, for leading the charge and pulling the great spirit in this forum back together from the chaos that was consuming it a few weeks back. It is great to see the collaboration and support from everyone, and especially the support of those who have led previous opt. efforts that today's knowledge is built upon.
Well done!
ID: 348740 · Report as offensive
Alex Kan
Volunteer developer

Send message
Joined: 4 Dec 03
Posts: 127
Credit: 29,269
RAC: 0
United States
Message 348852 - Posted: 26 Jun 2006, 2:41:05 UTC - in response to Message 348677.  
Last modified: 26 Jun 2006, 2:48:54 UTC

First of all, congratulations on releasing a client! I'm sure you've made a lot of people happy, and there will be even more happy people in the coming days... :)
If it's not too much of a stretch code-wise, it would really be a cool feature to be able to turn off the graphics via the BOINC manager, for example. Right now, if they're compiled into the client, they are ALWAYS drawn, not only when you're actually looking at them (when screensaver runs or you select "view graphics" in the manager).

It really seems a waste of perfectly good cycles even if you don't use an optimized client.

Wow, I hadn't realized this--just looked through the initialization code to confirm. Looks like I'm going to have to reissue my v6 clients to take this into account. And I thought I was being nice in addressing that tiny minority that wanted their optimized clients to have graphics as well...
ID: 348852 · Report as offensive
EricVonDaniken

Send message
Joined: 17 Apr 04
Posts: 177
Credit: 67,881
RAC: 0
United States
Message 348891 - Posted: 26 Jun 2006, 3:40:05 UTC

Good Stuff, Simon.

Any idea as to when you will have WinXP SSE2 and SSE3 clients available?
ID: 348891 · Report as offensive
Odysseus
Volunteer tester
Avatar

Send message
Joined: 26 Jul 99
Posts: 1808
Credit: 6,701,347
RAC: 6
Canada
Message 348958 - Posted: 26 Jun 2006, 4:56:52 UTC - in response to Message 348852.  

Wow, I hadn't realized this--just looked through the initialization code to confirm. Looks like I'm going to have to reissue my v6 clients to take this into account. And I thought I was being nice in addressing that tiny minority that wanted their optimized clients to have graphics as well...

Could you offer both versions, one fully optimized and one with graphics? The screensaver may be a ‘selling point’ for some, especially those who are nostalgic for Classic. ;)

What proportion of its CPU time does the app actually spend on graphics? I presume most of the ‘heavy lifting’ for rendering & animating is done by the video board.
ID: 348958 · Report as offensive
Iztok s52d (and friends)

Send message
Joined: 12 Jan 01
Posts: 136
Credit: 393,469,375
RAC: 116
Slovenia
Message 348959 - Posted: 26 Jun 2006, 4:58:23 UTC - in response to Message 348659.  
Last modified: 26 Jun 2006, 5:53:34 UTC

Bam.

'nuff said :o)
Regards,
Simon.


Hello!

Sorry, but it failed here (just copied seti + app_info.xml)

Slackware 10.1, P4:

./setiathome-5.15-sse2
Segmentation fault

and with boinc:

2006-06-26 06:44:02 [SETI@home] Unrecoverable error for result 18mr99ab.11394.5617.329836.3.246_0 (process got signal 11)

justr tried on another box, supporting SSE3. (Slackware 10.0)
Works!
Maybe some SSE3 found its way to the code?

BR
Iztok


my CPU, no sse3, does not work:

processor : 0
vendor_id : GenuineIntel
cpu family : 15
model : 3
model name : Intel(R) Pentium(R) 4 CPU 3.00GHz
stepping : 4
cpu MHz : 3001.345
cache size : 1024 KB
physical id : 0
siblings : 2
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 5
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe pni monitor ds_cpl cid xtpr
bogomips : 5931.00


another PC, sse3, works!

processor : 0
vendor_id : GenuineIntel
cpu family : 15
model : 3
model name : Intel(R) Pentium(R) 4 CPU 3.40GHz
stepping : 4
cpu MHz : 3394.635
cache size : 1024 KB
physical id : 0
siblings : 2
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 5
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe pni monitor ds_cpl cid xtpr
bogomips : 6701.05
ID: 348959 · Report as offensive
Iztok s52d (and friends)

Send message
Joined: 12 Jan 01
Posts: 136
Credit: 393,469,375
RAC: 116
Slovenia
Message 349004 - Posted: 26 Jun 2006, 7:42:22 UTC - in response to Message 348959.  

more info on different CPUs:

works on P4 with sse2 and ss.
Works on Opteron 285.

fails on Xeon (both old and new)
fails on Pentium-M (laptop)


They all run Slack, different -current versions.

Test: before updating boinc, I just run ./setiathome:
iztok@az:~/ulov/setiathome-5.15-sse2$ ./setiathome-5.15-sse2
Segmentation fault
iztok@az:~/ulov/setiathome-5.15-sse2$ cat /etc/slackware-version
Slackware 10.1.0

BR, thanks
Iztok

/proc/cpuinfo for boxes:

fails:

processor : 3
vendor_id : GenuineIntel
cpu family : 15
model : 2
model name : Intel(R) XEON(TM) CPU 2.40GHz
stepping : 4
cpu MHz : 2392.263
cache size : 512 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm
bogomips : 4771.02

processor : 3
vendor_id : GenuineIntel
cpu family : 15
model : 4
model name : Intel(R) Xeon(TM) CPU 3.20GHz
stepping : 3
cpu MHz : 3192.826
cache size : 2048 KB
physical id : 3
siblings : 2
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 5
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe lm pni monitor ds_cpl cid cx16 xtpr
bogomips : 6373.37

processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 13
model name : Intel(R) Pentium(R) M processor 1.80GHz
stepping : 6
cpu MHz : 1796.547
cache size : 64 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr mce cx8 sep mtrr pge mca cmov pat clflush dts acpi mmx fxsr sse sse2 ss tm pbe tm2 est
bogomips : 3578.26

ID: 349004 · Report as offensive
Profile The Psychotic One
Volunteer tester
Avatar

Send message
Joined: 22 May 00
Posts: 50
Credit: 4,099,029
RAC: 0
United States
Message 349030 - Posted: 26 Jun 2006, 9:15:52 UTC - in response to Message 347069.  

To anyone wanting the Visual C++ Toolkit 2003 (The free command line version from Microsoft), I didn't see it on their web site anymore. Instead they are pointing to their Visual Studio 2005 Express Edition (also free). That is located at http://msdn.microsoft.com/vstudio/express/]http://msdn.microsoft.com/vstudio/express/. I believe the Visual C++ Toolkit 2003 also had some sort of conflict also. If anyone still wants it however, I will try to locate it on my backup cds. I may have thrown it though. :( If I find it, I'll probably have to give it to someone elso to host as my i-net connection is at work. I doubt my boss would appreciate it. <BIG EVIL GRIN>


No, because I don't have a free 2003 version (and the 2005 express version croaks on some not-so-clean syntax somewhere, I forget).

Someone mentioned a free GUI-less 2003 version that's around - since the command-line options are in Visual Studio, I'm pretty sure that could be used to build.

For anyone interested in having Microsoft licenses available for personal use, you should really check out Microsoft certification...hint hint. Stop pirating, start getting free stuff from MS! Exams are much less expensive than licenses, but licenses are included when you take exams. Edumacate yourself ;o)

Regards,
Simon.


William D. Gagliardi
ID: 349030 · Report as offensive
Saimek

Send message
Joined: 25 Jan 00
Posts: 121
Credit: 454,423
RAC: 0
Poland
Message 349059 - Posted: 26 Jun 2006, 10:40:00 UTC

Just installed this client on My Dual Xeon 2.8 Ghz BOX. Running fine i'll give u some news about Crunch times when it finishes =)
ID: 349059 · Report as offensive
Natsuo Tsuji

Send message
Joined: 18 May 02
Posts: 24
Credit: 1,519,328
RAC: 0
Japan
Message 349103 - Posted: 26 Jun 2006, 12:53:12 UTC

Simon, could you make an optimized client for FreeBSD?
I tried your optimized client with Linux emulator on FreeBSD, but it didn't work.
ID: 349103 · Report as offensive
Profile KWSN - Chicken of Angnor
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 9 Jul 99
Posts: 1199
Credit: 6,615,780
RAC: 0
Austria
Message 349116 - Posted: 26 Jun 2006, 13:11:17 UTC

Oops -

since some of you seem to be having trouble executing that binary on SSE2 systems, I must have mixed up some flags.

Since I recompiled what I had before (that worked on all SSE2 hosts) and thought I used the same script, it should have been okay.

But I think I switched -axWN to -xWN and that was what's wrong.
Silly mistake, nobody's perfect :o)

I'll update the binary in an hour or so when I get it recompiled.

Regards,
Simon.
Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information
ID: 349116 · Report as offensive
Previous · 1 · 2 · 3 · 4 · 5 . . . 19 · Next

Message boards : Number crunching : Are there any sites providing optimized clients? -- PART II


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.