Linux port of Alex v8 code

Message boards : Number crunching : Linux port of Alex v8 code
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 · Next

AuthorMessage
Profile ML1
Volunteer moderator
Volunteer tester

Send message
Joined: 25 Nov 01
Posts: 20329
Credit: 7,508,002
RAC: 20
United Kingdom
Message 749246 - Posted: 7 May 2008, 0:01:36 UTC - in response to Message 748679.  
Last modified: 7 May 2008, 0:03:43 UTC

Hi all!

Just peeking in, at the right moment as it seems :o)

Props for porting Alex' code, that stuff is seriously fast!

If you want to check out the new code under Linux, there's some kind of band-aid available:

- Install a recent version of wine. (I used 0.9.61 under Debian)
- Use wine to run the windows ...

Good to see you on the forums again Hans.

And good for documenting the old WINE trick.


Known Problems:

- reported CPU times are always 0 seconds, because wine doesn't support the required system call (yet)

No problem for s@h or other fpops counting applications. That would be a killer for other projects if they count the elapsed time!

- crunching isn't done at nice level 19 (background priority)

Is that the case even if you run WINE itself from "nice"?

- You can't continue crunching your current linux WUs because BOINC detects a platform change and would discard them. No use in copying over files from your linux install.

It is all using a different science client after all and on a different API.

- No automatic startup of BOINC at boot time. I'll fix that shortly.

Easily fixed in various ways.

Happy Crunching indeed!

Regards,
Martin
See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)
ID: 749246 · Report as offensive
Hans Dorn
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 3 Apr 99
Posts: 2262
Credit: 26,448,570
RAC: 0
Germany
Message 749248 - Posted: 7 May 2008, 0:01:50 UTC
Last modified: 7 May 2008, 0:03:19 UTC

@Michael - Yup:

yonah1:~# cat /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 14
model name      : Genuine Intel(R) CPU           T2400  @ 1.83GHz
stepping        : 8
cpu MHz         : 1833.412
cache size      : 2048 KB
physical id     : 0
siblings        : 2
core id         : 0
cpu cores       : 2
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 10
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx constant_tsc arch_perfmon bts [color=red]pni[/color] monitor vmx est tm2 xtpr
bogomips        : 3670.42
clflush size    : 64


Hi Martin :o)
ID: 749248 · Report as offensive
Michael Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 21 Aug 99
Posts: 4608
Credit: 7,427,891
RAC: 18
United States
Message 749618 - Posted: 7 May 2008, 21:16:49 UTC - in response to Message 749248.  

@Michael - Yup:

yonah1:~# cat /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 14
model name      : Genuine Intel(R) CPU           T2400  @ 1.83GHz
stepping        : 8
cpu MHz         : 1833.412
cache size      : 2048 KB
physical id     : 0
siblings        : 2
core id         : 0
cpu cores       : 2
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 10
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx constant_tsc arch_perfmon bts [color=red]pni[/color] monitor vmx est tm2 xtpr
bogomips        : 3670.42
clflush size    : 64


Hi Martin :o)


and ht = hyperthreading?

ID: 749618 · Report as offensive
Hans Dorn
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 3 Apr 99
Posts: 2262
Credit: 26,448,570
RAC: 0
Germany
Message 749626 - Posted: 7 May 2008, 21:37:39 UTC - in response to Message 749618.  


and ht = hyperthreading?


Guess so. Obviously doesn't work properly - there's no hyperthreading on the T2400....


Regards Hans
ID: 749626 · Report as offensive
Michael Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 21 Aug 99
Posts: 4608
Credit: 7,427,891
RAC: 18
United States
Message 749632 - Posted: 7 May 2008, 21:45:25 UTC - in response to Message 749626.  


and ht = hyperthreading?


Guess so. Obviously doesn't work properly - there's no hyperthreading on the T2400....


Regards Hans


Same with my Pentium-4 Mobile 2Ghz...says ht however, I don't have any hyperthreading.

ID: 749632 · Report as offensive
Cygnus X-1
Volunteer tester

Send message
Joined: 15 Feb 04
Posts: 75
Credit: 3,732,505
RAC: 175
Canada
Message 760900 - Posted: 30 May 2008, 22:47:39 UTC - in response to Message 748679.  


- crunching isn't done at nice level 19 (background priority)


Well you can set it to nice level 19 manually but you have to do it for every WU because it will go back to 0 after a new one starts.

Another issue I'd like to point out: Not very important but my AMD CPU is reported as being Intel
ID: 760900 · Report as offensive
brinx

Send message
Joined: 13 Dec 05
Posts: 11
Credit: 856,060
RAC: 0
Australia
Message 761031 - Posted: 31 May 2008, 1:33:33 UTC

I'm running AKV8 under wine 1.0 on my Q6600 with Linux (Sidux) 2.6.24 kernel and it seems that I cannot get more than 1 extra WU. The msg keeps coming back that I wouldn't have enough time to complete extra work. Is this related to the 0.0 CPU time feature of wine or something else. BTW THANKS for AKV8. it really romps along on my setup taking about 1 hour 12 mins to complete a WU on each processor. Fantastic!
ID: 761031 · Report as offensive
UncleVom

Send message
Joined: 25 Dec 99
Posts: 123
Credit: 5,734,294
RAC: 0
Canada
Message 762573 - Posted: 4 Jun 2008, 0:25:53 UTC
Last modified: 4 Jun 2008, 1:11:36 UTC

It looks like Crunch3r comes through again :-D

http://calbe.dw70.de/index.html

Special thanks to Alex Kan, Crunch3r, Jason G, Raistmer, JDWhale and anybody else I missed.

UncleVom
ID: 762573 · Report as offensive
Stephen R

Send message
Joined: 25 Aug 99
Posts: 56
Credit: 3,736,096
RAC: 0
United States
Message 762668 - Posted: 4 Jun 2008, 3:13:07 UTC - in response to Message 762573.  

It looks like Crunch3r comes through again :-D

http://calbe.dw70.de/index.html

Special thanks to Alex Kan, Crunch3r, Jason G, Raistmer, JDWhale and anybody else I missed.

UncleVom


Not sure that url works too get to the apps very well.

Perhaps may be of use

AK8 Linux 64bit
AK8 Linux 32bit

Regards
ID: 762668 · Report as offensive
UncleVom

Send message
Joined: 25 Dec 99
Posts: 123
Credit: 5,734,294
RAC: 0
Canada
Message 762684 - Posted: 4 Jun 2008, 3:54:26 UTC - in response to Message 762668.  



Not sure that url works too get to the apps very well.

Perhaps may be of use

AK8 Linux 64bit
AK8 Linux 32bit

Regards


Thanks for the fix up.

This port really makes the penguins dance.
Looks like a huge improvement on the few work units I watched so far, I don't want to do the % comparison for the small sample, but wow.
Cores are running a degree or two C hotter.

UncleVom



ID: 762684 · Report as offensive
Stephen R

Send message
Joined: 25 Aug 99
Posts: 56
Credit: 3,736,096
RAC: 0
United States
Message 762688 - Posted: 4 Jun 2008, 4:16:50 UTC - in response to Message 762684.  
Last modified: 4 Jun 2008, 4:35:29 UTC



Not sure that url works too get to the apps very well.

Perhaps may be of use

AK8 Linux 64bit
AK8 Linux 32bit

Regards


Thanks for the fix up.

This port really makes the penguins dance.
Looks like a huge improvement on the few work units I watched so far, I don't want to do the % comparison for the small sample, but wow.
Cores are running a degree or two C hotter.

UncleVom



No prob.

Yep certainly does seem to make them penguins hop. Only finished one mixed WU so far myself, but new port seems to have shaved a nice chunk off the crunch time. Same as you I notice ~ 2C increase in core temp.

Thanks! for keeping your eye out for the AK app. port to Linux.

Regards
ID: 762688 · Report as offensive
UncleVom

Send message
Joined: 25 Dec 99
Posts: 123
Credit: 5,734,294
RAC: 0
Canada
Message 762775 - Posted: 4 Jun 2008, 12:14:54 UTC
Last modified: 4 Jun 2008, 12:16:32 UTC

The official word is out on the Lunatics site with a more complete explanation.
http://lunatics.kwsn.net/

Thanks again to all involved!

Looks like the penguins will finally be closing the gap on those cats and dogs.

UncleVom
ID: 762775 · Report as offensive
Cygnus X-1
Volunteer tester

Send message
Joined: 15 Feb 04
Posts: 75
Credit: 3,732,505
RAC: 175
Canada
Message 762786 - Posted: 4 Jun 2008, 13:05:44 UTC - in response to Message 762573.  

It looks like Crunch3r comes through again :-D

http://calbe.dw70.de/index.html

Special thanks to Alex Kan, Crunch3r, Jason G, Raistmer, JDWhale and anybody else I missed.

UncleVom


This is great! I'm making the switch as soon as I get back home after work.

Thanks to all involved in making this happen.
ID: 762786 · Report as offensive
Profile Sutaru Tsureku
Volunteer tester

Send message
Joined: 6 Apr 07
Posts: 7105
Credit: 147,663,825
RAC: 5
Germany
Message 762796 - Posted: 4 Jun 2008, 13:33:07 UTC


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?

ID: 762796 · Report as offensive
Josef W. Segur
Volunteer developer
Volunteer tester

Send message
Joined: 30 Oct 99
Posts: 4504
Credit: 1,414,761
RAC: 0
United States
Message 762838 - Posted: 4 Jun 2008, 15:37:51 UTC - in response to Message 762796.  


If it's the same hardware (QX6700) and I switch from WinXP AK V8 to LINUX AK V8, I would see more crunchingspeed?

Unknown, but possible. The Windows build should be a little faster overall, but on any particular system the only way to know for sure is to try it.
Which LINUX would be the best (performance) for only crunching rig?

The best OS is the least OS. Every CPU cycle the OS uses is one which cannot be used for crunching.

Someone skilled with Linux could probably trim any distro down to the bare minimum, just as most of the background tasks of Windows can be tamed or eliminated. But in both cases, once the OS is using less than 0.5% of CPU it is probably not worth trying for further improvement.
                                                                 Joe
ID: 762838 · Report as offensive
Terror Australis
Volunteer tester

Send message
Joined: 14 Feb 04
Posts: 1817
Credit: 262,693,308
RAC: 44
Australia
Message 762907 - Posted: 4 Jun 2008, 17:57:42 UTC - in response to Message 762684.  

This port really makes the penguins dance.
Looks like a huge improvement on the few work units I watched so far, I don't want to do the % comparison for the small sample, but wow.
Cores are running a degree or two C hotter.

UncleVom


Would be very interested in your times. I'm currently running AK-V8 under Wine on 2 rigs, A Q6600 and an E6700, both with a 3GHz O/C.

Both are currently taking about 1hr:10Min to crunch a 73 credit unit. What times are you getting with the Linux version ? I've had reports back suggesting that running AK-V8 under Wine is a smidgen slower (Around 5 mins per unit) than running under Windows. It will be interesting to get some numbers on the Linux port.

Regards
Brodo
ID: 762907 · Report as offensive
UncleVom

Send message
Joined: 25 Dec 99
Posts: 123
Credit: 5,734,294
RAC: 0
Canada
Message 762930 - Posted: 4 Jun 2008, 19:48:12 UTC - in response to Message 762907.  

This port really makes the penguins dance.
Looks like a huge improvement on the few work units I watched so far, I don't want to do the % comparison for the small sample, but wow.
Cores are running a degree or two C hotter.

UncleVom


Would be very interested in your times. I'm currently running AK-V8 under Wine on 2 rigs, A Q6600 and an E6700, both with a 3GHz O/C.

Both are currently taking about 1hr:10Min to crunch a 73 credit unit. What times are you getting with the Linux version ? I've had reports back suggesting that running AK-V8 under Wine is a smidgen slower (Around 5 mins per unit) than running under Windows. It will be interesting to get some numbers on the Linux port.

Regards
Brodo


This one should compare fairly easily, a Q6600 B3 running at 9 x 333.
Asrock 4Core1600P35-WiFi motherboard, 2- 1GB sticks of DDR2 6400 4.4.4.12 timings. All the faster looking stuff from June 4th on is with the AK Linux 64 bit client. OS Debian AMD64 lenny/sid 2.6.24 kernel.

You'll have to dig through the tasks I'd start with an offset of 400 or so.

http://setiathome.berkeley.edu/results.php?hostid=4265438&offset=400

Please let me know what you find.

Thanks,

UncleVom
ID: 762930 · Report as offensive
Urs Echternacht
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 692
Credit: 135,197,781
RAC: 211
Germany
Message 762945 - Posted: 4 Jun 2008, 20:44:27 UTC - in response to Message 762930.  
Last modified: 4 Jun 2008, 20:47:41 UTC


This one should compare fairly easily, a Q6600 B3 running at 9 x 333.
Asrock 4Core1600P35-WiFi motherboard, 2- 1GB sticks of DDR2 6400 4.4.4.12 timings. All the faster looking stuff from June 4th on is with the AK Linux 64 bit client. OS Debian AMD64 lenny/sid 2.6.24 kernel.

You'll have to dig through the tasks I'd start with an offset of 400 or so.

http://setiathome.berkeley.edu/results.php?hostid=4265438&offset=400

Please let me know what you find.

Thanks,

UncleVom

Looks like ca. 56 minutes per 73 cr. with AKV8 Linux 64 bit SSSE3 application.
Brodo, that beats yours by nearly a quarter of an hour.
_\|/_
U r s
ID: 762945 · Report as offensive
Terror Australis
Volunteer tester

Send message
Joined: 14 Feb 04
Posts: 1817
Credit: 262,693,308
RAC: 44
Australia
Message 763007 - Posted: 5 Jun 2008, 2:24:38 UTC
Last modified: 5 Jun 2008, 2:30:09 UTC

Yes, you can definitely pick the point where UncleVom changed apps, crunching time dropped by nearly 50%. Prior to that his crunching times were about the the same as I was getting before I hit the Wine :-)

Looks like I'm going to have to get all geeky and change back to the native Linux app (damn, bother, blast, it took me 2 days to get Wine running properly on the E6700). 15 minutes per unit is worth chasing.

Will report back when I have some results.

Brodo
ID: 763007 · Report as offensive
Profile Mumps [MM]
Volunteer tester
Avatar

Send message
Joined: 11 Feb 08
Posts: 4454
Credit: 100,893,853
RAC: 30
United States
Message 763012 - Posted: 5 Jun 2008, 2:45:22 UTC - in response to Message 762907.  

Would be very interested in your times. I'm currently running AK-V8 under Wine on 2 rigs, A Q6600 and an E6700, both with a 3GHz O/C.

Both are currently taking about 1hr:10Min to crunch a 73 credit unit. What times are you getting with the Linux version ? I've had reports back suggesting that running AK-V8 under Wine is a smidgen slower (Around 5 mins per unit) than running under Windows. It will be interesting to get some numbers on the Linux port.

Regards
Brodo

I apologize for going slightly off-topic, but I'm trying to understand these numbers.

If your Q6600 is overclocked to 3GHz, I'd think it should be doing better than 1hr:10Min for the average 73 credit unit. My Q6600 running WinXP 64bit with AKv8 is currently averaging 1hr:4min for similar 73 credit units. And I've so far only clocked it up to 2.52GHz. So I'd expect a 20% higher clockrate would mean a close to 20% matching decrease in crunch-time required, rather than roughly 5% more time needed.

Could the 64bit O/S mean that much difference? Or maybe it's the RAM? (Mine is PC8500 running with the FSB at 1121MHz due to the overclocking.) Is your Q6600 a G0 stepping? Or B3?

I haven't been having much luck catching wingmen with comparable systems, so I just thought I'd try asking.

Thanks
ID: 763012 · Report as offensive
Previous · 1 · 2 · 3 · 4 · 5 · Next

Message boards : Number crunching : Linux port of Alex v8 code


 
©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.