More info on S&H v3

[Assimilator1]

Whilst visiting Overclockers.co.uk forum I 1st stumbled across their new SETI forum & their new team.Anyway I found a post about v3b,here's a link to the beta 3 page you would get.

http://www.marcmcallister.freeuk.com./beta3.html

And here's a screen grab of it in action.

http://www.marcmcallister.freeuk.com/beta3.gif

Thanks to riddlermarc_2 from Overclockers uk forum for this info I hijacked

Inccidently a few of the Matrox users are also running v3b(v2.66) and they are reporting WU times of about 3x longer than current clients.So I wouldn't go rushing to try it out just yet!(expect perhaps 1 machine)

 

[MisterM]

Got it, seems a lot slower to me too!
I´ll stick with the older client...

 

[Assimilator1]

Well ,this is odd.I'm running it for MisterM (I didn't want to upset their Beta testing group) ,so far I have done 6.6% in about 33 mins.
MisterM tells me this is about twice as fast as he managed!

My PC_____Cel 366 @ 550 MHz,192 Mb RAM CAS2 ,BX chipset
MisterM's PC PII 450 @ 475 MHz,224 Mb RAM CAS2 ,BX chipset

Unfortunatly he'd had a little too much to drink last night so we can't get more accurate than that! (he's currently making a sacrifice to the SETI gods as an apology!).

I have an early theory on this partially based on this statement sent in their E mails to the Beta testers.

>>>>The big change in 3.0 is vastly improved science - we are now searching for pulses and triplets as well as spikes and gaussians. Web pages describing these in more detail will be up soon on the main home page.
Other improvements include optimized FFTs (for faster analysis of wave data) and various bug fixes. Because of the extra analysis, processing workunits will take more time than it did with version 2.0x. If the workunits take more than twice as long as they used to, however, please let us know (include information about how you run your client, i.e. do you go to a blank screen immediately, does it run in the background, etc.).<<<<<

If this holds out for the rest of the WU than it could be that their FFT is no longer larger than 512k &amp; is in fact smaller than 128k.Also it could be that the new FFT prefers full speed cache to bigger cache ,unlike the older clients.Of course this is just speculation on my part ,anyone else ?

 

[Assimilator1]

OK ,scrap that theory ! ,MisterM just realised he was running the CLi version hidden alongside ,that's what I get for taking info from him when he was drunk!

 

[MisterM]

It was 4 AM, I had some friends at my place, we had been playing Diablo 2 for 14 hours and I had a little bit too much to drink!

I promise it won`t ever happen again! SETI is so much more important than having fun with some friends!

 

[MGallik]

I dont know about that! Diablo II has been the Grail in this household
for over a year and a half. It should show up in the weekly output stats
this weekend.

The wife and kids are on the machines all day, every day!

And NO, HackStrap, my 28th level Barbarian nore Twisto, my 22nd level
Necromancer feel the least bit guilty either.

Heheh...

 

[Assimilator1]

Hmm ,it seems to of slowed right down ,13% @ 3hrs 6 mins.

Thats nearly 3x longer than v2.0x

 

[Assimilator1]

So far its at 21.2% @ 6hrs 32mins ,thats a total time of 32.5 hrs!! ,holy sh*t! 4x times longer than current versions ,so far!!

I've stopped running it for now ,it was just taking too long &amp; was hurting my race to 1000 WU's.I will restart it once I've done 1000 WU's so we can have a better idea of how long a WU will take.

MisterM

Are you still running it?

Is anyone else?

 

[DanStp]

Does anyone have an info on the 3.0 CLI clients? If the new ones are that much slowwer, I think I will be hanging on to the 2.4 as long as possible.........?

 

[MisterM]

I am still running, but it happened again! Some friends showed up and they forced me to play Diablo II!
At least my Level 27 barbarian has killed the cow king now!

I am at 33% after 17 hours, thanks to Diablo 2, but I will stop using it (the new client)too, it seems awfully slow!

 

[Darkone]

welp, I'm running at 30% after 7.5 hours uniterupted cracking.. (this machine I'm using it on would normally have done a WU in that time.)
so i'm looking at 22 and a bit hours a WU with the v3 client....


Looks like I'll be joining Dan in staying on 2.04 as long as i can.!!

D

 

[Michael]

For me, it is not the # of work units I can crank out that count, but the science behind the work units. If the new client takes 3 times as long but give three times the information to the scientists, then I'm doing what SETI@Home intends.

Michael

 

[Assimilator1]

Michael

I agree to a point ,but if they dont do a near immediate switch over from 1 client to the other then all those will be at a disadvantage as far as the stats go!
I would love to switch everthing over to the better client now but that huge time hit is not worth it until we are all on equal ground .

I shall also hang on to the old clients for as long as they let us use it ,despite the v3 having better 'searching' abilities

 

[JHutch]

Since this is almost a &quot;new&quot; seach, I'd like to see them set up a &quot;new&quot; stats system for the v3 clients... But given the fact that they seem overwhelmed as it is, I don't think we'll see that happen.

JHutch

 

[Assimilator1]

On the other hand they will have a reduced number of WU's coming through with this new client if it stays as it is.The stats would be more manageable

 

[office boy]

Yep, there appears to be a contention problem between the threads. There's a recent change that could have caused this, and we're testing some machines with that portion altered. If that fixes it, we'll have a new beta version next week. If anyones seen this problem on anything besides Windows, let me know. The anticipated runtime difference between this and the old version should be around 25%.
Eric
From Seti Staff

 

[Assimilator1]

Interesting ,where did you get that from OB?

 

[office boy]

alt.sci.seti
heres another one
>Has anyone submitted an official bug report on this?

Consider it official. We're chasing 3 leads. Those accusing us of
Intel favoritism should note that this bug only appears to affect Windows
machines.

Eric

--
Eric Korpela | An object at rest can never be
korpela@ssl.berkeley.edu | stopped.
<a href=&quot;http://sag-www.ssl.berkeley.edu/~korpela&quot;>Click for home page.</a>