[arkayn]

Here is the wisdom file that I generated on my eMac1.25 GHz

http://www.arkayn.us/mf/wisdom_emac_125.zip

[liebsmaschine]

Originally Posted by alexkan

Timed runs of the reference work unit are always appreciated, so please post them if you do them.

Tried to synthesize directions for finding and completing the reference WU. I downloaded boog's reference WU .zip from earlier in this thread file and expanded it. Then I copied your v3 worker into the folder and ran it with ./seti_enhanced-ppc-v3 -verbose -standalone (also sort of per boog's directions), but I couldn't get it to work. (An icon with the seti worker appeared briefly in my Dock and then disappeared, and my terminal prompt reappeared.) Also, on second thought, in order to time it, should I have used "time seti_enhanced-ppc-v3 -verbose -standalone"?

Any suggestions? Am I even using the right reference WU?

[Gecko_r7]

Alex, a few weeks ago you commented in the main testing thread:

"The current code blindly switches between FFTW and vDSP at a certain cutoff point without taking into account which is actually faster, but it does this at a point which makes sense for almost all machines I've looked at so far.

Does the transition between VDSP and FFTW take place between 4096 and 8912 on G4? I noticed on the FFT test2 from a couple months ago that VDSP maxes-out at 4096 benchmark and FFTW3 is larger starting at 8912.

"I was considering augmenting this FFTW/vDSP thing with a program that benchmarks to find the optimal combination of FFTs. Problem is, I'm already at least one release behind on FFTW, which claims to have made minor performance improvements. Also, it seems like generally bad form to be releasing new versions of the cruncher linked against old libraries."

FFTW3.1.1? Is this the release you are referring?

BTW, I should have a benchmark for G4 v3 w/ the new wisdom.sah file by tomorrow am. Takes a night to run the WU.

Also of note: Jackel's wisdom results were surprising being better using multi-user, console log in. I ran the same and compared against generation in single-user mode (s.u.m.). There is actually very little difference overall, though single user mode is 2-3% greater in the majority of points w/ the exception of a handfull of benchmarks, where console log in was greater. The only variable between the two tests was use of external USB plug-in for remote keyboard and mouse during generation w/ console log-in. In s.u.m., I had to use a std keyboard and mouse since USB wasn't running. Whatever problem froze Jackel's run in s.u.m. must have also affected the significantly lower output vs. what he generated w/ console log-in. At least from what I've seen, doesn't appear to be much difference in generation method, so use what is easiest. Short of Rick's upcoming GUI method, running using console log in is quicker and less hassle than s.u.m.

Cheers!

[alexkan]

Originally Posted by Gecko_r7

Does the transition between VDSP and FFTW take place between 4096 and 8912 on G4? I noticed on the FFT test2 from a couple months ago that VDSP maxes-out at 4096 benchmark and FFTW3 is larger starting at 8912.

None of the optimized Enhanced workers I've put out use vDSP for FFTs at the moment. If you were to do the math to see how much time is spent doing FFTs of size 4K and below, you would see that the amount of time spent in those functions currently doesn't justify the amount of functions that will need to be rewritten to allow me to mix FFTW and vDSP. I might take a crack at it later, but I probably have bigger fish to fry--it's nothing I haven't done before, but it just takes time.

FFTW3.1.1? Is this the release you are referring?

Yes. Most Enhanced workers are linked against FFTW 3.1.1, as is fft_test3.

BTW, I should have a benchmark for G4 v3 w/ the new wisdom.sah file by tomorrow am. Takes a night to run the WU.

Looking forward to seeing it.

[halimedia]

Now it's getting interesting! Here are the latest results from my Quad crunching the ref-wu w/ Alex' v3:

real 159m34.287s
user 159m1.948s
sys 0m17.783s
wu_cpu_time: 9525.018427

Thanx a bunch, Alex! Result-files available on request (PM me for that). Here's my quad crunching on v3 since 10:40 UTC today:

http://setiweb.ssl.berkeley.edu/show...hostid=2402169

[gulliver]

Here is the wisdom.sah for the

G5 Dual 2.7GHz: <a href="http://gulliver.macbay.de/wisdom.sah.zip">wisdom.sah.zip</a>

[sdubz]

Here is the wisdom.sah from my g4 mini http://boog.is-a-geek.org/seti/g4_mini_wisdom.sah.zip

[lepetitmartien]

If someone can host it (no time to bother about it really) or needs it, I've got the wisdom file for the G5 1.8 monocpu first version handy. Just PM me your email.

WU with the V2 app:
still pending right now
passed
ditto

I have a fourth one in the pipeline, once done I'll switch to V3

[Gecko_r7]

Results of v3 w/ new wisdom compared to stock worker:
G4 MDD w/ Dual Giga 1.33

Stock 5.13 Worker (Test1 5.24.06)
real 499m12.054s ( = 29952.054s)
user 498m11.342s
sys 0m57.878s
wu_cpu_time = 29915.352618

Alex v3
real 381m23.393s (22883.393s)
user 380m38.200s (22838.200s)
sys 0m42.508s
wu_cpu_time>23798.013826

Improvement vs. real = 24.6%
Improvement vs. wu_cpu_time = 20.5%

Now we're cooking w/ gas !
All things being equal, at this point, we're at the same @ level of Opt that x86 fellas are using Crunch3r's 5.12 ap (according to his estimate of improvement) Thanks Alex and Boog for getting us this far!

From here on, it's going to get interesting!
Anyone compared stock vs. v3 w/ Quad?
BTW, if anyone wants Wisdom for G4 Dual 1.33 let me know. Not a popular combo, so I can e-mail on request.

Cheers!

[alexkan]

Here's that new version of the source code that I promised you all earlier (and that I'm obligated to provided because of the GPL):

http://tbp.berkeley.edu/~alexkan/seti/src-v3a.tar.bz2

It's actually slightly improved over v3, since I forgot to make the tarball before I made a few new changes. From here on out, it looks like improvement will slow down, since all the low-hanging fruit is pretty much picked, to the extent that there's no one huge bottleneck. However, if you have Shark and a G5 that's not a 1.8, I would love to have you profile the reference work unit in Shark. I'll even send you the config I'm using, for consistency's sake.

[liebsmaschine]

It's hard to compare WU results and times directly, as the new Enhanced WUs vary greatly in size/depth/whatever. However, a result done in my stock worker gave 58.7 points of credit for 47,980 seconds of work. Under Alex's worker, I got 32.17 points for 17,463 seconds. Not a bad improvement! That's 1.5x as much credit per FFT (I think--if I did the math right)!
Gotta run, so I'll just quickly post my results page:
http://setiathome.berkeley.edu/resul...hostid=2225838
The bottom one (result 330929542) was done with the stock worker. Everything else (another one is almost done) is on Alex's v3 worker.

[halimedia]

Since running v3, I've noticed that my quad is now consistently claiming less credit per WU than the rest of the quorum. Question to Alex: is this to be expected?

Here are my quad's results: http://setiweb.ssl.berkeley.edu/resu...hostid=2402169

Still have to wait and see if the same is true for other machines of mine...

[halimedia]

For the first time since switching to Enhanced, my quad's RAC is climbing again! Thanx much, Alex!

[alexkan]

Originally Posted by halimedia

Since running v3, I've noticed that my quad is now consistently claiming less credit per WU than the rest of the quorum. Question to Alex: is this to be expected?

Whoops, that's a bug. I forgot to increment the FLOP counter in one of the vectorized function replacements. I'll put in the fix, but you won't actually see it until the next release, unless this is seriously affecting your granted credit.

[lepetitmartien]

Originally Posted by halimedia

For the first time since switching to Enhanced, my quad's RAC is climbing again! Thanx much, Alex!

I second one, just found it out too!

V3 seems to crunch like a good trooper for now (first WU still in the pipeline), we'll see tomorrow.

[Todd Madson]

Here's my wisdom file for my 2.5 dual:
http://pod.ath.cx/nuwisdom/

I just switched over to v3, no news at this point. It appeared that when I used
Boog's enhanced system I received 11k average blocks. Under Alex's V2 it was
around 17k-18k work units. But much of it is hard to tell because work unit
size fluctuates so much in this new system. I just can't tell. Not to mention
that it apparently can decide on what work units to work on first versus what
would seem logical.

Stay tuned.

[lepetitmartien]

WU are processed by order of limit date. At least for the one with a short delay to send them back.

first WU done with V3
There's a slight difference in the credits claimed, it must be the little problem already mentioned. For the record, the first 8% of the WU were processed with V2, the rest with V3. (I'm 2397079)

[dwaring]

Originally Posted by lepetitmartien

WU are processed by order of limit date. At least for the one with a short delay to send them back.

first WU done with V3
There's a slight difference in the credits claimed, it must be the little problem already mentioned. For the record, the first 8% of the WU were processed with V2, the rest with V3. (I'm 2397079)

From your link, I noticed the one sharing your WU is using a MacBook Pro.... Are the stock client for the Intel counterpart a lot faster than the PPCs....?

[halimedia]

Originally Posted by alexkan

I'll put in the fix, but you won't actually see it until the next release, unless this is seriously affecting your granted credit.

No problem! The difference is marginal (approx. 2-4%), and in the majority of cases, I'm granted more credit than I'm claiming.

[halimedia]

Originally Posted by Elphidieus

From your link, I noticed the one sharing your WU is using a MacBook Pro.... Are the stock client for the Intel counterpart a lot faster than the PPCs....?

All things being equal, the current CoreDuo-Macs are about as fast as a dual 2GHz G5 (PPC 970FX-based) in CPU-intensive tasks. It's not by accident that Apple decided to jump the PPC-ship on the notebook side...

Edit: that said, it appears that the stock-worker is more thoroughly optimized on the i386-side of things. It has always been that way, AFAIK.

[chboss]

The change to Seti enhanced changed the production per day for many people.
It seems that Mac users (PPC) suffer the most at this point. It is good to see Alex is working on the Seti client to bring us back up to speed.

I have V3 running on my machines... Alex keep the improvments comming.

@ Halimedia

...this said I am happy to have some x86 machines working for me so I can increase my lead in our country stats. Lately I felt your breath lin my Neck...

[halimedia]

Originally Posted by chboss

@ Halimedia ...this said I am happy to have some x86 machines working for me so I can increase my lead in our country stats. Lately I felt your breath lin my Neck...

Hello, arch-rival Can't wait to encroach further on your territory!

[TiloProbst]

hm now I've got worker-v3 + boinc_5.4.9ExtraCreditUniversal (non superbench) installed on my Single 1,8. how is Alex' v3 worker supposed to show up in Activity Monitor? "seti_enhanced-pp" or am I using the default one?

btw wisdom-files for Single 1,8 and Dual 2,0 will have to wait, our university webserver seemingly is down since saturday

[dwaring]

I'm running on Alex's v3 with one of my own wisdom file, but I didn't notice any speed gains....?

Anyone with a 1.9GHz G5 here....?

And it seems that v3 takes up as much as 66MB of real memory. Is this normal....?

[alexkan]

Originally Posted by Elphidieus

I'm running on Alex's v3 with one of my own wisdom file, but I didn't notice any speed gains....?

Anyone with a 1.9GHz G5 here....?

Since Enhanced does a variable amount of work per unit, you'll have to look at the seconds it takes per unit of credit requested/granted. That, or benchmark the reference work unit, since I like cold, hard numbers.

And it seems that v3 takes up as much as 66MB of real memory. Is this normal....?

Yes. To my knowledge, there aren't any memory leaks, but that's what I said last time. I do know, however, that this client uses more memory than the stock client, because it transforms certain data to make accessing it faster.

Speaking of faster, here's v4 and its related sources. This may be the last version I make without separate G4 and G5 compiles. I can't point you to any specific files to look at, because I changed a bunch this time.

http://tbp.berkeley.edu/~alexkan/set...ced-ppc-v4.zip
http://tbp.berkeley.edu/~alexkan/seti/src-v4.tar.bz2

Please post your reference work unit benchmarks if you do them!

[Gecko_r7]

Originally Posted by alexkan

Speaking of faster, here's v4 and its related sources. This may be the last version I make without separate G4 and G5 compiles. I can't point you to any specific files to look at, because I changed a bunch this time.

http://tbp.berkeley.edu/~alexkan/set...ced-ppc-v4.zip
http://tbp.berkeley.edu/~alexkan/seti/src-v4.tar.bz2

Please post your reference work unit benchmarks if you do them!

Alex, In a future revision, can you make a notation of revision # & using FFTW3 wisdom in the completed WU strerr output? Might make a little easier to identify which WUs, crunched w/ what, and if crunching w/ intended revision. I'll bench v4. BTW, so far, no validation or crunching probelms w/ v3 outside of claimed credit bug previously mentioned.

Thanks for all you do!

[halimedia]

Originally Posted by Gecko_r7

Alex, In a future revision, can you make a notation of revision # & using FFTW3 wisdom in the completed WU strerr output? Might make a little easier to identify which WUs, crunched w/ what, and if crunching w/ intended revision.

I second this motion. The standard seti worker output is rather messy. A nicely formatted output à la crunch3r would be nice

I'll bench v4. BTW, so far, no validation or crunching probelms w/ v3 outside of claimed credit bug previously mentioned. Thanks for all you do!

Ditto and ditto and ditto

[dwaring]

Originally Posted by alexkan

Yes. To my knowledge, there aren't any memory leaks, but that's what I said last time. I do know, however, that this client uses more memory than the stock client, because it transforms certain data to make accessing it faster.

V4 had the real memory hovering just above 38MB... Crunching speed is very erratic....

[chboss]

Great Alex!

Your frequent updates keep me busy... love it.

[halimedia]

I'm having tons of fun here! Latest results from my Quad crunching ref-wu w/ v4:
real 136m38.388s
user 136m30.802s
sys 0m7.597s
wu_cpu_time: 8160.082663s

Now we're talkin'! Thanks so much Alex! You're the last defender of the PPC-age!

[arkayn]

I seem to be having a problem with my installation, boinc refuses to see that I have v4 in the folder and always downloads the standard worker.

[BTBlomberg]

Originally Posted by Gecko_r7

Alex, In a future revision, can you make a notation of revision # & using FFTW3 wisdom in the completed WU strerr output? Might make a little easier to identify which WUs, crunched w/ what, and if crunching w/ intended revision. I'll bench v4. BTW, so far, no validation or crunching probelms w/ v3 outside of claimed credit bug previously mentioned.

Thanks for all you do!

As of Ver 3 he has a lot in the "strerr out". Here is an example from a failed WU:

Code:
<core_client_version>5.4.9</core_client_version>
<message>
Maximum disk usage exceeded
</message>
<stderr_txt>
ar=0.422310 NumCfft=72953 NumGauss=           464507884 NumPulse=         89036768383 NumTriplet=       7691545591808
ar=0.422310 NumCfft=72953 NumGauss=           464507884 NumPulse=         89036768383 NumTriplet=       7691545591808

Crashed executable name: seti_enhanced-ppc-v3
built using BOINC library version 5.5.0
Machine type PowerPC 7400
System version: Macintosh OS 10.4.6 build 8I127
Sun May 28 22:04:06 2006
Stack frame backtrace:
 #  Flags Frame Addr  Caller PC   Return Address Symbol
===  ===  ==========  ==========  =====================
  1  ---  0xf0100560  0x000d5374  MoreBacktraceMachSelf + 0x12c
  2  ---  0xf01005e0  0x000a81e8  PrintBacktrace + 0x6c
  3  ---  0xf0101930  0x00084dd8  _Z26handle_process_control_msgv + 0x110
  4  ---  0xf0101d80  0x00084f00  _Z12worker_timeri + 0xc0
  5  ---  0xf0101de0  0x0008512c  _Z12timer_threadPv + 0x30
  6  ---  0xf0101e30  0x9002ba68  
</stderr_txt>

It does not reflect the FFT info though.

Here is a story from the bottom of the G4 pile (well a PM 350 G4 tower may be more the bottom but this is close).

I had two WU like this that had "Maximum disk usage exceeded" errors that make me look to be onlky one with error in Quarum. This is on my PowerBook G4 500 that has taken a dive with the change to Enhanced. My P3 550 running WinXP (current) is getting 54,926.24 (15.25 hours) a full WU with Crunc3r's 5.12 client but is still faster than the PowerBook that has not completed a Full WU yet but is looking at 23 hours. I had started the WU with Boog's last client (reported 31 hours to finish), switched to "seti_enhanced-ppc-v3" after spoting it last night and then finally generated a Wisdom file for it (first run took 1.5 hours and I did not realize it was made the file in the Root of the User account) the next try only took 1/2 hour. With v3 it looked like it would still be about 22-23 hours total but i just replaced it with v4 and it is still calculating time based on the progress of v3 so it is yet to be seen what the benefit would be on this old dog.

I hope that Alex's progress is good. He and Rick did wonders with the old SETI client.
Here is the Link to my PBG4. You may see that before the switch to enhanced my average was 10000s per WU (2:45 hours) and my RAC was about 130-140 for the machine with Alex&Rick A4 client (a bit better that the P3 550 with Chrunc3r client RAC 110-120).

I have not run a test WU but would if desired. As is it takes a Day per WU, but I appreciate all of the work to make that change. I look at the code and after a while go glassy eyed .

Thanks for all the work guys.

[rhettmaxwell]

Originally Posted by arkayn

I seem to be having a problem with my installation, boinc refuses to see that I have v4 in the folder and always downloads the standard worker.

Hmmm...

The only fix I can think of is to make sure you have the app_info.xml file in the setiathome.berkeley.edu folder.

[arkayn]

It is right there underneath the work units.

[halimedia]

Is it just me, or are other folks affected by an inability to upload results? I keep getting the following message:

Temporarily failed upload of ..... : Error 403

Anyone else seeing this? Any idea what Error 403 is?

TIA for any insights. No time for research now, unfortunately...

[adream]

Originally Posted by halimedia

Is it just me, or are other folks affected by an inability to upload results? I keep getting the following message:

Temporarily failed upload of ..... : Error 403

Anyone else seeing this? Any idea what Error 403 is?

TIA for any insights. No time for research now, unfortunately...

the server isnt accepting uploads at the mo, all my clients are backing up results too, both windoze and mac

regards

adream

[lepetitmartien]

Just switched to V4, no problems whatsoever with V3 (only one bad WU but it was a noisy one for all), tough I can't be sure 100% till the server accepts the results again (2 WU are waiting)

[BTBlomberg]

Feedback on v4 on the PowerBook G4 500, it looks like it's flying along compared to v3 and earlier versions. In the last 1.25 hours of computing time (I was working with Photoshop and Illustrator some) the "To Completion" time droped 3.33 hours. Now that is purely an observational messurement and I will have not hard numbers as I have not had the patience to let the previous slower clients complete a WU. 23-32 hours was not attractive to me.

If this observation pans out this may be an increase of speed by at least double on this machine, but we will see. This may bring this machine to under 10 hours for a WU. For those with G5s and faster G4s this may seam like too long but if it speeds it up like that for this machine it should do great for you too. This would also make this machine to the same speed as my AMD 1700+(1400) with Crunch3r's clients.

Thanks, Alex and I look forward to v5.

[Gecko_r7]

Here's V4 results on G4 MDD DP 1.33 against Ref WU

AlexKan v4 (Test3 5.29.06)
real 318m23.606s ( = 19103.606s)
user 317m44.906s
sys 0m36.372s
wu_cpu_time = 19898.524583

vs. Stock worker

real: + 36.2%
wu_cpu_time: +33.5%

vs. V3

real: +16.5%
wu_cpu-time: +16.4%

Once again, big improvement on G4. Thanks Alex!
No issues or problems so far.
Regards!

[arkayn]

Both SETI and Einstein are not accepting uploads right now and Einstein is completely ofline.

[BTBlomberg]

Gecko_r7's numbers seam to sync up with what I am seeing for v4. I have about an hour to go but statistically working based on the current WU since changing to v4 it should take 13.46 hours to do a WU with v4. I should find out for sure tomorrow. If I extrapolate based on Gecko's numbers it should take me just over 14 hours, but the PowerBook has a 1 MB L3 cache wereas the Dual he has may have a smaller one which can make a difference. Nice improvement.

[Gecko_r7]

Originally Posted by BTBlomberg

As of Ver 3 he has a lot in the "strerr out".

In Seti std 5.3, there was a brief description in the result summary similar to this (using example from Enhanced x86 reported WU)

<core_client_version>5.4.9</core_client_version
Windows optimized S@H Enhanced application by Crunch3r
Version info: Windows SSE2 V5.12 by Crunch3r
Work Unit Info.............and so forth.

Perhaps an errored WU produces additional info (which makes sense since it's describing the error)?

Halimedia: Have you seen the description we're referring in v3 or v4 WU report summary? I have not.

Another reason that it will be helpful is to better identify and compare processing times and differences between different AR wus and ap versions. Reported crunch times can be highly variable w/ Enhanced (and confusing if using opt aps) because WUs can be of substantially different length as described by the WU's AR.
Example: (not actual WUs)
v3 WU 18,000 secs, AR= .417
v4 WU 18,000 secs AR = .324 (on v3, time = 21,000)

Other than running the Ref WU to predit what "should" happen overall, the only way to get a handle on how fast work is actually being done is to be familiar with expected times for common ARs. It takes Boinc quite a while to settle-out predicted WU times after a revision change....seemingly an eternity on anything but a DP/DC G5. This may take a few weeks (days on a Quad ) until the client has had a chance to process a consistent range of ARs using the same application. When testing the differences between versions as several of us do, Boinc prediction is worthless anyway. Ultimately, we have to dig into the WU reports and compare. A few brave souls in the Seti forum are currently cataloging and spreadsheeting WU AR examples. The general idea is to determine basic performance comparables such as "X" AR takes "Y" % more or less than "Z" AR, for the most typical ARs. As we have all seen, Enhanced is a tiger of different stripes vs. what we previously knew so well. It's going to take time, patience and some creativity to sort this out. The fun is just beginning!
Time to sleep.....
Regards.

[Todd Madson]

Woke up this morning to find a bunch of work units at 100%.
3hr 34min, 3 hr 26 min, 3hr 24min, 3hr 12min, 3hr 16min, 3hr 11min,
3hr 11min, 3 hr 12 min, 3hr 13min, etc.

So it looks like around 3:11 to 3:34 max.

Then two weird ones:
1 min 37 secs, 1 min 19 secs, 1 min 18 secs.

Haven't had the chance to try out the reference WU though yet.

[lepetitmartien]

Todd the very fast WU are noisy one usually.

V4 is noticeably faster (sorry, not serious data, the test WU mumbo jumbo is just too hairy). Final times are roughly better of about 20-30% relative to the official app.

[halimedia]

Originally Posted by lepetitmartien

V4 is noticeably faster (sorry, not serious data, the test WU mumbo jumbo is just too hairy). Final times are roughly better of about 20-30% relative to the official app.

the ref-wu crunching is just about the only way to get serious data. V4 is roughly 70% faster than the stock worker on a dual-core G5. SETI-scientists predicted that optimizations would not improve speed beyond 100% (i.e. twice as fast). Let's see what our wizards can do!

Edit: sorry, my math was a bit off! V4 is roughly 50% faster than the stock worker (202min / 136min = 1.485)

[dwaring]

first WU with v4 validated....

http://setiathome.berkeley.edu/resul...ltid=333291461

Edited:

Subsequent noisy WUs that are validated...

http://setiathome.berkeley.edu/resul...ltid=333291481
http://setiathome.berkeley.edu/resul...ltid=333291462

[arkayn]

After removing everything from my computer and then reattaching I am crunching with v4. I will attach back to einstein once it comes back up.

[Todd Madson]

G4/400 work unit crunch times with v4 = roughly 14 hours +/ - an hour.

Prior to enhanced I was getting 2 hrs 45 minutes to 3 hours 15 minutes.

I suspected enhanced would kick it in the teeth but wow.

[gregulate]

Is there any way to check does custom wisdom.sah used or not?

[BTBlomberg]

Originally Posted by Todd Madson

I suspected enhanced would kick it in the teeth but wow.

You said it.

Likewise my PB G4 had times of around 2:45 hours, but Enhanced was able to be optimized to the same level as with the previous Alex&Rick client SETI would lead one to believe that with their statement that Points should equal out between Old and Enhanced SETI that we should be able to achieve 5:50 hour turn around on new WUs on same machine. That said I have seen none of my machines, WIN PC, Linux PC or MAC, come close to this claim, even with optimized (although works in progress) clients.

Now I hope Alex and others can pull that extra optimization out, but if not it makes me wonder if the SETI Projects claims are correct or if this early code is just the beginning of a refinement process that will gradually live up to the claims.

Oh yah, Gecko you are right. My first V4 WU is validated (ok started with Boogs for and hour, then Alex V3 [reset progress to 0% but left the time crunched at just over an hour] until 50%ish and then V4) and the "stderr out" lacks the App ID that would be good to have.