Seti@Home optimized science apps and information
Optimized Seti@Home apps => Linux => Topic started by: sunu on 30 Apr 2008, 04:09:58 pm
-
I'm seeing very positive feedback for the new optimized clients over at Berkeley's forums.
Will we see a linux port of them? Especially 64bit?
-
I'm seeing very positive feedback for the new optimized clients over at Berkeley's forums.
Will we see a linux port of them? Especially 64bit?
I expect someone will use the sources, available shortly, and make a Linux port long before I can relearn Linux and it's development tools (Last used Linux for that sort of thing ~1996 :o), Though if nobody beats me to it I'll probably give it a go.
-
I expect someone will use the sources, available shortly, and make a Linux port long before I can relearn Linux and it's development tools (Last used Linux for that sort of thing ~1996 :o), Though if nobody beats me to it I'll probably give it a go.
Thanks!
Since Mac Os X is Unix based maybe it's easier to port in Linux than it is in Windows. Maybe what I'm saying is totally wrong.
Keep up the good work guys!
-
If anyone needs a beta Tester for Linux port Apps, I have *cough* sufficient 64bit Linux Systems available ;)
-
Please add me to the list too.
I've got a Debian 64bit Q6600 box here that would be more than happy to partake in some testing.
UncleVom
-
Thanks both, You're on the list and the Gatekeeper (Gecko_R7) will be in touch in due course.
-
I also have a 64bit Linux box (AMD 64 X2 , SUSIE 10.3) if you need another tester.
great work on the ports by the way. thanks
-
Just updated some of my Windows boxes that run SETI to AW, but I've got a Q6600 running Ubuntu (latest) that would be glad to run the new code on Linux when you guys get it working....
Been running SETI since the classic days, so I understand the value of testing new code before it's released!
Mark
-
If that helps anyhow, i could offer one of my linux boxes (T7200, 64bit openSuse 10.3) for testing. It's a dedicated cruncher anyway. Also, if that is necessary, i could put a fresh windows on a spare HDD (for dual boot).
-
My ubuntu 8.04 32bit running an AMD XP 3000 (sse2) is available for testing if you need it.
-
My Q6600 with 64bit Ubuntu and Crunch3r's 6.1.0.64 Boinc client would like to help too :)
-
Thanks all,
More data is needed, and Crunch3r is working on more test builds, as are Andy and Raistmer. There's no rush though, so hang tight. When Gecko_R7 comes online He'll hopefully see this thread full of willing testers and consider each situation in turn against what we need, make adjustments, and give some guidance where needed.
Thanks, and glad you could make it. & all that goes for others that come along to this thread after this post.
Jason
-
Hi Everyone,
THANK YOU for your willingness to volunteer for testing duty ;D
We're still in the process of seeing what we have available & what is still needed.
Rather than open the floodgates all at once, I'd like to gradually add a few testers at a time over the next few days.
This will give us a chance to round-out a representative group w/ different hardware & OS distributions.
A main goal of testing to see how variances/choices in developmental builds respond to different platforms etc.
This will help towards builds that work well for a wider range of platforms and OS combos.
As we add folks to the group, I'll notify you by PM. Please check back at the boards over the next few days.
If you don't receive a PM, please don't be discouraged ;). We may be covered for the moment.
As everyone has seen, these types of things are processes...not events, and you may still yet be asked to help.
Thanks again for answering the call! :D
Cheers,
Gecko
-
I've got a bunch of Debian boxes here, 32bit or 64bit. T7200/T2300 or Athlon64
Regards Hans
-
Hi Hans,
Sounds like some heavy hitters are coming along, with yourself & crunch3r involved :). I hope to see you 'inside'. Looks like there'll be plenty of kinks to iron out. Wouldn't fun without that part!
Jason
-
Hi,
I don't mean to drag anybody away from the important stuff.
I was just wondering how things were progressing with this port?
UncleVom
-
Hi UncleVom,
There is testing of builds going on behind the scenes, mixed results so far in terms of compatibility with different distros. If I understand what's going on properly (which I probably don't as I'm not directly involved with that build, or that familiar with Linux development either) then it will take longer to obtain satisfactory release builds ... but it is progressing, and it is getting me more interested in Linux development myself. So you and me both are wanting to see what comes of this.
Jason
-
Hi Jason,
Thanks for the update.
Be careful with your Linux uptake, I hear it can become addictive. :-)
UncleVom
-
Just so UncleVom does not have to ask twice in a row, any word - good, bad, indifferent on the Linux port. It has been almost two weeks since the last update and inquiring, impatient minds would like to have any small crumb of information if available from time to time. ;) Thanks.
-
Hi there,
there are several builds in testing, and it seems to me (being not overly familiar with Linux) there are difficulties with library compatibility across distros at the moment. This is being ironed out and performance results look promising.
Jason
-
Thanks for the update. I am a Slackware user myself so I can appreciate them taking the time to make sure it works with more than one flavor of Linux.
-
Thanks for the update.
I think ensuring easy installation across distros is very important, there are a lot of Linux newbies out there these days.
Sorting out package dependencies and varying file locations can be trying enough for veteran users.
On the other hand I'm pretty sure there are some who would more than willing to hack the distro to fit the app. ;-)
UncleVom
-
Linux version of Alex Kan's V8 posted on Crunch3r's site http://calbe.dw70.de/ .
Running it right now :)
-
Special thanks to Alex Kan, Crunch3r, Jason G, Raistmer, JDWhale and anybody else I missed. :-D
This seems very fast.
UncleVom
-
Impressive work guys, many thanks to all involved.
-
If it's the same hardware (QX6700) and I switch from WinXP AK V8 to LINUX AK V8, I would see more crunchingspeed?
Which LINUX would be the best (performance) for only crunching rig?
-
On T2330 SSSE3/SSE3 32/64 bit linux versions were slower than corresponding x86 Windows ones.
The same picture (for SSE3) was observed on AMD Venice.
So linux versions preferable for linux guys only IMHO. No need to switch to linux only to run SETI for now.
-
On T2330 SSSE3/SSE3 32/64 bit linux versions were slower than corresponding x86 Windows ones.
The same picture (for SSE3) was observed on AMD Venice.
So linux versions preferable for linux guys only IMHO. No need to switch to linux only to run SETI for now.
I would say it's exactly the opposite, linux app seems faster than the windows version.
My machine is a Q6600 G0 OCed to 3.11GHz (345 FSB) and DDR2-1150 with ubuntu 8.04 64 bit. This is my working machine, not a dedicated cruncher and I'm always running a massive 25000 file torrent in the background.
In all workunits mentioned below, my wingman also has a Q6600.
Take a look in these workunits: 271702837 (http://setiathome.berkeley.edu/workunit.php?wuid=271702837), 272171135 (http://setiathome.berkeley.edu/workunit.php?wuid=272171135), 272059004 (http://setiathome.berkeley.edu/workunit.php?wuid=272059004), 271665803 (http://setiathome.berkeley.edu/workunit.php?wuid=271665803) (all Win32). Taking into account my overclock (even with 100% scaling), linux is faster than windows.
In this workunit, 272124897 (http://setiathome.berkeley.edu/workunit.php?wuid=272124897), with Win64, still linux is faster.
Finally in workunit 271664596 (http://setiathome.berkeley.edu/workunit.php?wuid=271664596) (Win32) my wingman has a bigger overclock than me, but still my linux machine is faster.
-
Glad to hear it's working well for you sunu. We'll keep an eye on performance figures to find out if this applies to other machines as well.
Cheers, Jason
-
It is working great! :-D
I'm finding a similar situation.
Q6600 box 9x333mhz.
I did a comparison on the same machine (sample size of one) between the Win32 ssse3 client run as a service and the Linux64 ssse3 client in user space, kde3 desktop and the Linux client seems around 3 minutes faster on a 73 credit work unit. Sadly I don't have a copy of Win64 to use for a comparison.
I'm wondering if the testing referred to on the T2330, which I take to be a laptop, took place with the default power saving and freq daemons running.
UncleVom
-
Yes, T2330 was Lenovo notebook.
Under Windows power saving was set to "hi prformance" option and checked by CPU-Z that there is no CPU multiplier switching. And this action didn't change run time against "balance" opton. At "balance" option there was CPU multiplier switch in idle state (x8 <-> x6 as I recall correctly), but again, no changes in WU run times (probably CPU don't want change his multiplier under heavy load).
So I supposed the same behavior for Linux and took no additional actions after booting Ubuntu from live CD. If anyone can point how to disable all power saving features under this linux version I would repeat that tests :)
And about AMD - It's usual host and no power saving active at all. Results was obtained with dedicated test WUs under KWSN bench Windows/Linux version. BTW, Windows version had even more load cause this host acts as router and terminal server under Windows and no activity at all under linux (even no network).
I propose to post bench results in separate thread here to collect rintime base for different OS/CPU/Memory/App combinations. There was such database on http://www.marisan.nl/seti/ but now this url leads here.
It would be very useful to have such database IMHO.
-
Since no Ubuntu users have come to the rescue.
I'm not a Ubuntu or even a Gnome user, but I've loaded the live CD on my desktop (sorry no suitable laptop) and switched it to crunching my workunits for a comparison.
I'm kind of up to my ears in work right now, but I'll try to figure out whats going on, I have confirmed that the laptop-detect and laptop-mode-tools packages are loaded by the default live CD.
UncleVom
-
The Ubuntu live CD seems to crunch at a similar rate to my Debian sid installation any difference seems to be lost in the noise of workunit variability.
I did turn up this posting from Toby in March on the Number Crunchers Forum on Seti@home in a Windows vs. Linux thread.
"I see the same behavior on my laptop running Ubuntu. Idle processes do not cause the kernel to initiate CPU power level change. You can change this behavior with the cpufreq-selector command. To make the CPU run at full speed all the time, just type this command into a terminal:
sudo cpufreq-selector -g performance
This causes the kernel to use the "performance" governor to determine when speedstep/powernow features are enabled. The other governors that are available are powersave, conservative, ondemand and userspace. I believe the default one is "ondemand" which, as the name suggests, keeps the CPU at low speed unless more power is needed. You can also lock the CPU in at a specific frequency with this command:
sudo cpufreq-selector -f 1670000
The 1670000 means 1.67 GHz (my CPU can run at 1, 1.67 or 2 GHz).
I'm sure there is some GNOME/KDE interface to this feature that doesn't require breaking out a terminal but thats the quick and dirty way :)
You can view the speed your CPU is currently running at with this command:
cat /proc/cpuinfo
and looking for the line that starts with "cpu MHz." There is also a GNOME panel widget (or whatever they're called) under the "System & Hardware" section entitled "CPU Frequency Scaling Monitor" that will graphically tell you at what speed your CPU is running."
HTH
UncleVom
-
Thank you very much for this release. I've been waiting for that impatiently 8) I'm running that on 32 bit Fedora (32bit SSE3&SSSE3), everything seems working perfectly ;)
-
@UncleVom
Thanks for info! Will try to retest with new power settings :)
-
Hi,
Is the MB version for SSE will be out soon?
Thanks for all the other version.
-
Hi there,
This seems to be a thread which is a bit old, but it seems appropriate for my question:
For sentimental value, I'm running Linux (Debian unstable) on a P3-700MHz box at home. Spare cycles are used to process SetiBOINC.
I have the SSE-optimized Astropulse clients installed, but when heading over to Crunch3rs page (link is on page 2 of this topic) I only see AK8 clients for SSE2 and up. Are there any plans to release an SSE-enabled version of the AK8 client?
Failing that, how can I let the P3-700 machine use the regular Seti@Home client, since the app_info.xml only lists the custom astropulse clients. What do I have to put in an app_info.xml file to use the regular, non-optimized Seti@Home clients?
Thanks in advance,
Patrick.
-
You can use the 2.4L versions from Crunch3r's page. They offer pretty good speed increase over stock.
-
You can use the 2.4L versions from Crunch3r's page. They offer pretty good speed increase over stock.
Hmm, okay. I can use these together with current app_info.xml files which specify 528/603 versions of the wus?
Or will I also get plenty work when using/merging the original app_info.xml (which specifies 521)?
Regards, Patrick.
-
You can use the 2.4L versions from Crunch3r's page. They offer pretty good speed increase over stock.
Hmm, okay. I can use these together with current app_info.xml files which specify 528/603 versions of the wus?
Or will I also get plenty work when using/merging the original app_info.xml (which specifies 521)?
Regards, Patrick.
Try replacing the text with this one.
<app_info>
<app>
<name>setiathome_enhanced</name>
</app>
<file_info>
<name>setiathome-5.15.i686-pc-linux-gnu</name>
<executable/>
</file_info>
<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>528</version_num>
<file_ref>
<file_name>setiathome-5.15.i686-pc-linux-gnu</file_name>
<main_program/>
</file_ref>
</app_version>
<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>603</version_num>
<file_ref>
<file_name>setiathome-5.15.i686-pc-linux-gnu</file_name>
<main_program/>
</file_ref>
</app_version>
<app_info>
-
OK, I can handle that. ::)
One final question though: I see there are 2 variants of the SSE 2.4L binary on Crunch3rs site:
- One using FFTW
- Another not using FFTW
Which one should I pick for a P3-700?
Regards, Patrick.
-
I can't help you in this. This post ( http://lunatics.kwsn.net/linux/new-2-4l-version-for-linux-x86.msg6403.html#msg6403 ) says that the FFTW one is slower. I guess you'll have to test it yourself.
-
OK, I can handle that. ::)
One final question though: I see there are 2 variants of the SSE 2.4L binary on Crunch3rs site:
- One using FFTW
- Another not using FFTW
Which one should I pick for a P3-700?
Regards, Patrick.
If you're going to use a 2.4L version, be sure to merge its app_info rather than the app_info for stock which arkayn posted. As to which 2.4L, the FFTW version may perform slightly better on hosts with small L2 cache but for an Intel CPU I'd suggest the other one unless you want to try each for a week or so and judge for yourself.
Joe
-
OK, I can handle that. ::)
One final question though: I see there are 2 variants of the SSE 2.4L binary on Crunch3rs site:
- One using FFTW
- Another not using FFTW
Which one should I pick for a P3-700?
Regards, Patrick.
If you're going to use a 2.4L version, be sure to merge its app_info rather than the app_info for stock which arkayn posted. As to which 2.4L, the FFTW version may perform slightly better on hosts with small L2 cache but for an Intel CPU I'd suggest the other one unless you want to try each for a week or so and judge for yourself.
Joe
Yes, I did merge the app-infos, and this day, two regular Seti WUs were downloaded where the host previously only downloaded ap wus.
My app_info.xml is:
<app_info>
<app>
<name>setiathome_enhanced</name>
</app>
<file_info>
<name>setiathome-5.15.i686-pc-linux-gnu</name>
<executable/>
</file_info>
<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>528</version_num>
<file_ref>
<file_name>setiathome-5.15.i686-pc-linux-gnu</file_name>
<main_program/>
</file_ref>
</app_version>
<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>603</version_num>
<file_ref>
<file_name>setiathome-5.15.i686-pc-linux-gnu</file_name>
<main_program/>
</file_ref>
</app_version>
<app>
<name>astropulse</name>
</app>
<file_info>
<name>astropulse-5.0.i686-pc-linux-gnu</name>
<executable/>
</file_info>
<app_version>
<app_name>astropulse</app_name>
<version_num>500</version_num>
<file_ref>
<file_name>astropulse-5.0.i686-pc-linux-gnu</file_name>
<main_program/>
</file_ref>
</app_version>
<app>
<name>astropulse_v5</name>
</app>
<file_info>
<name>astropulse-5.03.i686-pc-linux-gnu</name>
<executable/>
</file_info>
<app_version>
<app_name>astropulse_v5</app_name>
<version_num>503</version_num>
<file_ref>
<file_name>astropulse-5.03.i686-pc-linux-gnu</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>
(I hope this comes out ok...)
Regards, Patrick.
-
Are there any disadvantages in using AK v8 or 2.4L compared to the default 6.03? Are they able to use all work units that 6.03 can use and do they perform the same analysis? Or does 6.03 have a newer, better analysis and with using an older piece of software I might miss signals?
So what made the version numbers rise from 5.x to 6.x?
-
The science done by optimized apps is the same. You don't miss anything. AKv8 is the fastest one.
-
Are there any disadvantages in using AK v8 or 2.4L compared to the default 6.03? Are they able to use all work units that 6.03 can use and do they perform the same analysis? Or does 6.03 have a newer, better analysis and with using an older piece of software I might miss signals?
So what made the version numbers rise from 5.x to 6.x?
The graphics for stock builds. 6.x are built with a separate graphics app in order to match what BOINC 6 had to do to work within Windows Vista guidelines.
Note that 6.03 doesn't run on Win95, and Eric Korpela's advice was to use 5.27 with an app_info.xml file. As far as the signal analysis is concerned, any version of setiathome_enhanced back to the 5.1x versions is fine. Credit claims can be enough different to annoy other users, though.
The CUDA app (both stock and the builds included in Raistmer's packages) has some things missing so in some cases it bails out with an "unsupported function" error (though BOINC doesn't know that's what the -12 exit code means so it calls it "unknown").
Joe
-
So it's safe to use 2.4L (SSE only) and pretend having 5.27 ?
-
So it's safe to use 2.4L (SSE only) and pretend having 5.27 ?
Yes
-
Credit claims can be enough different to annoy other users, though.
It annoys me, too. Credits claimed dropped by roughly a factor of 3, computation time dropped by a factor of 2-2.5. So in the same time I do more work for less credits than with the default app. I guess one needs SSE2 or SSE3 to have enough speedup to compensate the lesser credits with the increased number of WUs.
So I have to decide if I care most about credits, than I use the default app again, or if I care most about science and helping the project, then I stay with the optimized app.
Perhaps I should just try AP, although I guess it will take weeks to figure out if that pays off.
-
Perhaps I should just try AP, although I guess it will take weeks to figure out if that pays off.
I think right now AP 5.03 gives more credits that MB, but it takes a while to pick up.
-
Credit claims can be enough different to annoy other users, though.
It annoys me, too. Credits claimed dropped by roughly a factor of 3, computation time dropped by a factor of 2-2.5. So in the same time I do more work for less credits than with the default app. I guess one needs SSE2 or SSE3 to have enough speedup to compensate the lesser credits with the increased number of WUs.
So I have to decide if I care most about credits, than I use the default app again, or if I care most about science and helping the project, then I stay with the optimized app.
Perhaps I should just try AP, although I guess it will take weeks to figure out if that pays off.
Actually, the claims from 2.4L versions are fine, in those builds Crunch3r had implemented the code to comply with the credit_rate in the WU header. The low credits you've seen so far are because you got some Very High Angle Range work (aka 'shorties').
I concur that if you're patient optimized AP_v5 work should give higher credit/time than MB work. The patience is needed not only to get the work done but also because wingmates will often abort or error out or simply fail to complete the work, so the credit grant can be delayed a lot.
Joe
-
Actually, the claims from 2.4L versions are fine, in those builds Crunch3r had implemented the code to comply with the credit_rate in the WU header. The low credits you've seen so far are because you got some Very High Angle Range work (aka 'shorties').
I now had a WU that gave nearly the credits of the WUs before (42.6). It seems as if the speed improvement for SSE-only CPUs is in a region of about 5%, which isn't even statistically significant with only one WU done. Perhaps the version with fftw is better, but these 5% are not worth the effort of finding and installing the optimized app.
I concur that if you're patient optimized AP_v5 work should give higher credit/time than MB work. The patience is needed not only to get the work done but also because wingmates will often abort or error out or simply fail to complete the work, so the credit grant can be delayed a lot.
One week to go...