Hi,
the last thing i read about icc 9.1 is that will fully integate into vs 2005
( 9.0 latest release only works on cli)

I also read 9.1 officially support (is integrated into) VS2005 in the release note. It will work fine. Good luck!

Well i'm waiting for the notification mail from intel that 9.1 is relased :)
Anyhow i'm quite keen to see if all works that smooth like the promised to...

---

Join BOINC United now!

After the comments suggesting we not use crunch3r's 5.11 versions, perhaps it is worth posting to this thread that the site was updated on May 8 with 5.12 versions.

http://www.guntec.de/Crunch3r/setiew.html

---

Hi,
the last thing i read about icc 9.1 is that will fully integate into vs 2005
( 9.0 latest release only works on cli)

I also read 9.1 officially support (is integrated into) VS2005 in the release note. It will work fine. Good luck!

---

Luckiest in the world. WMD = Weapon of Mass Distraction.
Click this table.

Crunch3r,
there are some more odd things going on . See some other postings of me.
Well that may be to much asked.I may look in to it tomorrow myself and make a compilation of my findings.

What posting ?

Forget what I said, it seems I can't communicate verry well with you. Same occurred when you brought the first optimised for standard Seti to public, wich we are all thankfull for.
Besides, as then others will soon enough spot the same things and bring it to you clearer.

Sorry to hear that but i don't have the time to read all postings here on the board.

Hope you understand that.

---

Join BOINC United now!

I've crunched one WU of enhanced with crunch3r's SSE2 5.11 science ap.
my first crunch3r enhanced WU

The quorum was fulfilled much faster than expected.

Not only was my result granted credit, but it was denoted the canonical result. If I understand the terminology and process, this means it was deemed strongly similar to at least one of the other two results, both of which appear to have been computed on the distributed 5.12.

As others predicted, the "correct" credit claim would have been about 62.4, so this version's claim of 168 was considerably excessive on that scale. I'm told this is an relic of its 5.11-ness. The tx36-attempted CPU adjustment noted in stderr not only did not affect the credit claim, it did not even propagate to the posted result page.

As the 168 credit claim would have about matched the credit productivity I had running crunch3r and trux tx36 on previous SETI, the 62.4 corrected claim means machines like mine will take a major hit in RAC running enhanced. A few months ago, crunch3r-running SETI machines had a huge RAC advantage over Einstein machines. In a complete reversal, akosf-running Einstein machines will now have a huge RAC advantage over SETI enhanced machines. This may induce some resource switch, probably a good thing in this case.

The speed question is stickier yet. I've not read whether crunch3r has implemented algorithmic changes over the stock client, but assuming he has not, we should not expect 4:1+ improvement, but much less. My actual execution time on one virtual CPU of an HT 3.2 GHz Gallatin was 19270 seconds. My two quorum partners were a 3.2 GHZ Pentium D running as 2 CPUs at 15383 seconds (is this a dual core machine, in which case he was _not_ running HT?) and a 2.8 GHz Pentium 4 apparently running HT which took 29086 seconds.

My first take on the speed results is that they might be consistent with improvement on the order of extremely roughly 20%.

I'll wait to get a short execution time WU before doing direct comparisons.

You have to keep in mind that seti enhanced uses a cache like the one introduced by hans dorn in the normal seti app. (Harold Naparst sources)

fftw made a large step towards the speed of intels ipp (but ipp still is a bit faster)

and the default enhanced app. is using processor optimized routines like SSE/SSE2 etc in fftw.

It was expected that the optimized app. is faster but not like the ones we used before that were 4 to 6 times faster than the default app.

---

Join BOINC United now!

Crunch3r,
there are some more odd things going on . See some other postings of me.
Well that may be to much asked.I may look in to it tomorrow myself and make a compilation of my findings.

What posting ?

Forget what I said, it seems I can't communicate verry well with you. Same occurred when you brought the first optimised for standard Seti to public, wich we are all thankfull for.
Besides, as then others will soon enough spot the same things and bring it to you clearer.

I've crunched one WU of enhanced with crunch3r's SSE2 5.11 science ap.
my first crunch3r enhanced WU

The quorum was fulfilled much faster than expected.

Not only was my result granted credit, but it was denoted the canonical result. If I understand the terminology and process, this means it was deemed strongly similar to at least one of the other two results, both of which appear to have been computed on the distributed 5.12.

As others predicted, the "correct" credit claim would have been about 62.4, so this version's claim of 168 was considerably excessive on that scale. I'm told this is an relic of its 5.11-ness. The tx36-attempted CPU adjustment noted in stderr not only did not affect the credit claim, it did not even propagate to the posted result page.

As the 168 credit claim would have about matched the credit productivity I had running crunch3r and trux tx36 on previous SETI, the 62.4 corrected claim means machines like mine will take a major hit in RAC running enhanced. A few months ago, crunch3r-running SETI machines had a huge RAC advantage over Einstein machines. In a complete reversal, akosf-running Einstein machines will now have a huge RAC advantage over SETI enhanced machines. This may induce some resource switch, probably a good thing in this case.

The speed question is stickier yet. I've not read whether crunch3r has implemented algorithmic changes over the stock client, but assuming he has not, we should not expect 4:1+ improvement, but much less. My actual execution time on one virtual CPU of an HT 3.2 GHz Gallatin was 19270 seconds. My two quorum partners were a 3.2 GHZ Pentium D running as 2 CPUs at 15383 seconds (is this a dual core machine, in which case he was _not_ running HT?) and a 2.8 GHz Pentium 4 apparently running HT which took 29086 seconds.

My first take on the speed results is that they might be consistent with improvement on the order of extremely roughly 20%.

I'll wait to get a short execution time WU before doing direct comparisons.

---

Crunch3r,
there are some more odd things going on . See some other postings of me.
Well that may be to much asked.I may look in to it tomorrow myself and make a compilation of my findings.

What posting ?

---

Join BOINC United now!

Crunch3r,
there are some more odd things going on . See some other postings of me.
Well that may be to much asked.I may look in to it tomorrow myself and make a compilation of my findings.

Meanwhile my second result under Crunch3rs came in and validated.
Same claimed credit ( as I espected) different time needed.
http://setiathome.berkeley.edu/result.php?resultid=318348995

Did you overclock you a64 3000+ ???

The reason i'm asking is that

Your time needed:
5,370.43 86.31 32.03
real_cpu_time 9430

A64 3500+
24,886.64 32.03 32.03

Strange thing thoug cos:

Message 298599 - Posted 7 May 2006 12:47:43 UTC [Edit this post]
Just completed first wu with Crunch3rs 5.11. It was from the same batch of witch 5 results were already in.
Average time on those 11985 sec. all claiming 32.03 credits and granted ( exept for one that's pending).

The 5.11 did 8382 sec. and claiming 86.31 an other result of th same wu done with 5.12 claims the regular 32.03.

In other words one would expect your aplication do the thing in 0.7 the time.

is the new app for linux?

---

Scorpions - Wind Of Change

Sorry I did'nt mentioned that . I usualy do.
Yes, overclock of 10%. Newcastle core sse2

Meanwhile my second result under Crunch3rs came in and validated.
Same claimed credit ( as I espected) different time needed.
http://setiathome.berkeley.edu/result.php?resultid=318348995

Did you overclock you a64 3000+ ???

The reason i'm asking is that

Your time needed:
5,370.43 86.31 32.03
real_cpu_time 9430

A64 3500+
24,886.64 32.03 32.03

---

Join BOINC United now!

Meanwhile my second result under Crunch3rs came in and validated.
Same claimed credit ( as I expected) different time needed.
http://setiathome.berkeley.edu/result.php?resultid=318348995

.. just wait for 9.1 to be released.

regards,

-tmr

Hi TMR,

that's what i intend to do and icc 9.1 will integrate into vs 2005.

Hi Crunch3r,

It's nice, but I am not sure it really works with VS2005 (I use with VS2003 bought on eBay). icc 9.1 for Linux doesn't work well with gcc-4.1.0 which can be installed with a minor tweak on Gentoo, although Intel says it works. So I reported it as a problem report to intel with BOINC source attached :) I suspect it's just a minor c++ header dependency problem.

Its support is very kind, helpful and eager to find and fix bugs.

-tmr

Hi,
the last thing i read about icc 9.1 is that will fully integate into vs 2005
( 9.0 latest release only works on cli)

---

Join BOINC United now!

.. just wait for 9.1 to be released.

regards,

-tmr

Hi TMR,

that's what i intend to do and icc 9.1 will integrate into vs 2005.

Hi Crunch3r,

It's nice, but I am not sure it really works with VS2005 (I use with VS2003 bought on eBay). icc 9.1 for Linux doesn't work well with gcc-4.1.0 which can be installed with a minor tweak on Gentoo, although Intel says it works. So I reported it as a problem report to intel with BOINC source attached :) I suspect it's just a minor c++ header dependency problem.

Its support is very kind, helpful and eager to find and fix bugs.

-tmr

---

Luckiest in the world. WMD = Weapon of Mass Distraction.
Click this table.

.. just wait for 9.1 to be released.
regards,
-tmr

Hi TMR,

that's what i intend to do and icc 9.1 will integrate into vs 2005.

Thanks for your response Crunch3r and your web site has a vastly better layout now. Very much improved!

---

Boinc....Boinc....Boinc....Boinc....

So, I got the optimized app off of Crunch3rs website. (Very awesome, and the new design looks wonderful man, keep it up). The WU I got is a 12 hour long one, I suspended all my other projects and just letting the enhanced run. Will check on it later and post how it works on my end. Thank you Crunch3r and crew for working so hard on this.

Jeremy

[edit] Well turns out the WU I got someone else already reported and it got the -9 error. LOL So I'll have to see how the next one after that does.

---

.. just wait for 9.1 to be released.

regards,

-tmr

Hi TMR,

that's what i intend to do and icc 9.1 will integrate into vs 2005.

---

Join BOINC United now!

Today I got my first crunch3r_seti_enhanced WU and it looks like good.
It runs ~5 hours, no errors
<core_client_version>5.3.12.tx36</core_client_version>
<real_cpu_time>18219</real_cpu_time>

I used the app_info.xml from archae86, see below
no errors, it works fine
Happy crunching

Only problem I see is that Crunch3r version 5.11 seems to overclaim credit.
http://setiathome.berkeley.edu/workunit.php?wuid=76682687 for example.
My computer claimed the higher credit.

---

Boinc....Boinc....Boinc....Boinc....