U41.xx Observation Thread

Nothing But Idle Time
Nothing But Idl...
Joined: 24 Aug 05
Posts: 158
Credit: 289204
RAC: 0

RE: btw: Om my fathers

Message 30217 in response to message 30216

Quote:

btw: Om my fathers computer (P4 3,2 HT) my times have decreased dramatically:
S40.12
7068
7302

S41.07HT
5238
5285
5187
AndyK

Interesting numbers AndyK, now compare them to my numbers yesterday on my P4 Prescott 3.0 GHz w/HT on and these files starting and running simultaneously:

S39L Z1_0995.0_202... 1:57:09 (7029 secs)
S39L Z1_0995.0_201... 1:57:09

S39L Z1_0995.0_200... 1:57:10
S39L Z1_0995.0_198... 1:57:10

S41.07HT Z1_0995.0_199... 2:29:23
S41.07HT Z1_0995.0_197... 2:28:37

32 minutes longer? I quickly switched back to S39L. I've tried a few other S and U versions of the science app and I always switch back to S39L which has been the fastest version on my machine.

AndyK
AndyK
Joined: 5 Jan 06
Posts: 21
Credit: 44767
RAC: 0

RE: RE: btw: Om my

Message 30218 in response to message 30217

Quote:
Quote:

btw: Om my fathers computer (P4 3,2 HT) my times have decreased dramatically:
S40.12
7068
7302

S41.07HT
5238
5285
5187
AndyK

Interesting numbers AndyK, now compare them to my numbers yesterday on my P4 Prescott 3.0 GHz w/HT on and these files starting and running simultaneously:

S39L Z1_0995.0_202... 1:57:09 (7029 secs)
S39L Z1_0995.0_201... 1:57:09

S39L Z1_0995.0_200... 1:57:10
S39L Z1_0995.0_198... 1:57:10

S41.07HT Z1_0995.0_199... 2:29:23
S41.07HT Z1_0995.0_197... 2:28:37

32 minutes longer? I quickly switched back to S39L. I've tried a few other S and U versions of the science app and I always switch back to S39L which has been the fastest version on my machine.

hmmm...

some more numbers:
S41.07HT p2_0550.0__74_... 9847
U41.04 p2_0550.0__75_... 9053

Have to wait a little until some "normal" wus (z1_*) are coming to get comparable results, but U41.04 seems to be faster than S41.07HT.

Maybe I'll catch one normal WU and test that one with all akosf versions on the P4, AmdXP and AMD64...
Will throw me down in RAC, but for heavens sake: We don't have one table yet for comparison. In fact we should have one reference wu to do testing with, like it is done with seti from the begining on.

With the P4 the testing will take a lot time:
- HT off
- HT on, that albert crunching only
- HT on, that albert with another albert crunching
- HT on, that albert and seti crunching

AndyK

Want to know your pending credit?

[img]http://tinyurl.com/438v3"[/img]
The biggest bug is sitting 10 inch in front of the screen.

Vorik
Vorik
Joined: 10 Nov 04
Posts: 9
Credit: 52839612
RAC: 464598

I persuaded a friend of mine

I persuaded a friend of mine to install BOINC and crunch for einstein on his sempron3000+(palermo core), using U41.04. (http://einstein.phys.uwm.edu/hosts_user.php?userid=193802)

The first result had a validation error though, it says "Validate state-Checked, but no consensus yet". (http://einsteinathome.org/task/28597819)

Don't know if it has something to do with U41.04, but I can't remember having seen this before.

AndyK
AndyK
Joined: 5 Jan 06
Posts: 21
Credit: 44767
RAC: 0

Vorik, that wu was started

Vorik, that wu was started using the standard app and then continued with the U41.04.
Wait until the next wu completely done with U41.04 is done and see then.

AndyK

Want to know your pending credit?

[img]http://tinyurl.com/438v3"[/img]
The biggest bug is sitting 10 inch in front of the screen.

Bob Guy
Bob Guy
Joined: 9 Feb 05
Posts: 12
Credit: 58059
RAC: 0

S39L is faster or as fast as

S39L is faster or as fast as any of the other versions of the app. The latest I've run is U41.04 - it is about as fast as s39L but I don't have a saved WU to test it with. I'm just guessing. Ran all of the apps on a P4 3.2 Prescott with HT enabled but with only 1 app allowed to run at a time. I had one invalid WU a while ago but I don't remember which of the apps did that - I think it was an isolated incident.

Nothing But Idle Time
Nothing But Idl...
Joined: 24 Aug 05
Posts: 158
Credit: 289204
RAC: 0

RE: ... Maybe I'll catch

Message 30222 in response to message 30218

Quote:

...
Maybe I'll catch one normal WU and test that one with all akosf versions on the P4, AmdXP and AMD64...Will throw me down in RAC, but for heavens sake: We don't have one table yet for comparison. In fact we should have one reference wu to do testing with, like it is done with seti from the begining on.

With the P4 the testing will take a lot time:
- HT off
- HT on, that albert crunching only
- HT on, that albert with another albert crunching
- HT on, that albert and seti crunching

AndyK


AndyK, some people propose running Einstein with HT off. But my single computer is used at home for more than distributed computing. My wife and I surf the internet and she does "homework" and desktop publishing in the evenings. Limiting myself to a single processor would slow down DC (I presume) so I leave HT on out of necessity. Doesn't anyone else have this issue? --NBIT

AndyK
AndyK
Joined: 5 Jan 06
Posts: 21
Credit: 44767
RAC: 0

RE: RE: ... Maybe I'll

Message 30223 in response to message 30222

Quote:
Quote:

...
Maybe I'll catch one normal WU and test that one with all akosf versions on the P4, AmdXP and AMD64...Will throw me down in RAC, but for heavens sake: We don't have one table yet for comparison. In fact we should have one reference wu to do testing with, like it is done with seti from the begining on.

With the P4 the testing will take a lot time:
- HT off
- HT on, that albert crunching only
- HT on, that albert with another albert crunching
- HT on, that albert and seti crunching

AndyK


AndyK, some people propose running Einstein with HT off. But my single computer is used at home for more than distributed computing. My wife and I surf the internet and she does "homework" and desktop publishing in the evenings. Limiting myself to a single processor would slow down DC (I presume) so I leave HT on out of necessity. Doesn't anyone else have this issue? --NBIT

Some people propose... yes, that's the thing: They think, but they don't know for sure!
What I noticed so far is: With HT on and crunching two einstein at the same time makes the time for one wu longer than crunching one einstein and one other project like seti simultaneous. The crunching time for einstein is about 10 to 20 percent lower than with two einstein at the same time.
But I havn't a proof yet.

AndyK

Want to know your pending credit?

[img]http://tinyurl.com/438v3"[/img]
The biggest bug is sitting 10 inch in front of the screen.

Fuzzy Duck
Fuzzy Duck
Joined: 3 Dec 05
Posts: 37
Credit: 936924
RAC: 0

U41.04 dropped my times from

U41.04 dropped my times from approx 1.5 hours to less than 50 minutes using a Centrino duo T2400 (2 cpus). However using Trux client and the relevant truxoft_prefs I cannot get S@H & E@H to run as simultaneous processes. Once and if I do, it would be interseting to see the comparisons.

FD

AndyK
AndyK
Joined: 5 Jan 06
Posts: 21
Credit: 44767
RAC: 0

RE: RE: RE: btw: Om my

Message 30225 in response to message 30218

Quote:
Quote:
Quote:

btw: Om my fathers computer (P4 3,2 HT) my times have decreased dramatically:
S40.12
7068
7302

S41.07HT
5238
5285
5187
AndyK

Interesting numbers AndyK, now compare them to my numbers yesterday on my P4 Prescott 3.0 GHz w/HT on and these files starting and running simultaneously:

S39L Z1_0995.0_202... 1:57:09 (7029 secs)
S39L Z1_0995.0_201... 1:57:09

S39L Z1_0995.0_200... 1:57:10
S39L Z1_0995.0_198... 1:57:10

S41.07HT Z1_0995.0_199... 2:29:23
S41.07HT Z1_0995.0_197... 2:28:37

32 minutes longer? I quickly switched back to S39L. I've tried a few other S and U versions of the science app and I always switch back to S39L which has been the fastest version on my machine.

hmmm...

some more numbers:
S41.07HT p2_0550.0__74_... 9847
U41.04 p2_0550.0__75_... 9053

Have to wait a little until some "normal" wus (z1_*) are coming to get comparable results, but U41.04 seems to be faster than S41.07HT.

Update: with HT on but einstein on one cpu and seti on the other, like I did already with S41.07

U41.04 z1_1063.5__120_... 4960
U41.04 z1_1063.5__121_... 4877

Quote:
Maybe I'll catch one normal WU and test that one with all akosf versions on the P4, AmdXP and AMD64...


ups... it's a bit different from seti, what files do I have to copy to anoher location to test one wu without the boinc framework?

AndyK

Want to know your pending credit?

[img]http://tinyurl.com/438v3"[/img]
The biggest bug is sitting 10 inch in front of the screen.

archae86
archae86
Joined: 6 Dec 05
Posts: 3157
Credit: 7227811589
RAC: 1104639

RE: it's a bit different

Message 30226 in response to message 30225

Quote:


it's a bit different from seti, what files do I have to copy to anoher location to test one wu without the boinc framework?

AndyK

I've been using a method posted by Stonelord for the last month.

Stonelord's captive WU test method
It seems a bit crude, but hard drive space is cheap, and it works.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.