S38 Observation thread

RandyC
RandyC
Joined: 18 Jan 05
Posts: 6003
Credit: 111139797
RAC: 0

RE: Based on 4 results, my

Message 26149 in response to message 26143

Quote:

Based on 4 results, my AMD XP 1600+ system has dropped about 1700 secs per WU. About 20% decrease in processing time C37 to S38.

My XP 2600+ system is upgraded and has completed a single result now. It shows approximately 1400 seconds less for a 23% decrease in processing time.

Good job!!

Seti Classic Final Total: 11446 WU.

kev1701e
kev1701e
Joined: 11 Feb 05
Posts: 15
Credit: 460291
RAC: 0

My A64 3000+ @2.3GHz is now

My A64 3000+ @2.3GHz is now doing a WU in about 1200 seconds.

Fantastic Akosf!

kev

A64 X2 4400+ @2.85GHz, 2x XP 1800+ @2000+

Akos Fekete
Akos Fekete
Joined: 13 Nov 05
Posts: 561
Credit: 4527270
RAC: 0

RE: I find very interesting

Message 26151 in response to message 26146

Quote:
I find very interesting fact.

Yes, these values good for analysing.
I can determine when doesn't use somebody the computer. So, good time for hacking... ;-))

ca_grufti
ca_grufti
Joined: 9 Feb 05
Posts: 53
Credit: 4309237
RAC: 0

Akos, thank you. Half the

Akos,

thank you. Half the time I laugh out loud when I read your comments. It's hard to decide what I like better: your sense of humor or your coding skills.

Hans

DanNeely
DanNeely
Joined: 4 Sep 05
Posts: 1364
Credit: 3562358667
RAC: 1580

RE: I must've missed the

Message 26153 in response to message 26140

Quote:

I must've missed the details on how C37 improves upon performance over A36, because the difference I was seeing was negligible - sorry, no numbers to post at the moment. But I'll try the S38 over the next day or two and let you know what I find. :)

Thanks for all of your efforts, akosf. :D

C37 did give a noticeable gain for P4 systems. There was no real benefit on AMD chips. Also IIRC C37 was entirely 386 compatable and gave large gains to people using older machines (mainly amd since they lagged badly in adding intel extentions to thier architecture).

Bruce Allen
Bruce Allen
Moderator
Joined: 15 Oct 04
Posts: 1119
Credit: 172127663
RAC: 0

RE: [Bruce suggested to me

Message 26154 in response to message 26128

Quote:
[Bruce suggested to me Chebyshev polinomials instead of Taylor series a month ago. I did a fast test to compare these methods, but I found that Chebyshev approximation produced worse average by same number of coefficients. I'm working on a program that will generate more precise values)

See Section 5.8 page 192 of Numerical Recipes. You should also look at page 76 of Abramowitz and Stegun.

Director, Einstein@Home

Akos Fekete
Akos Fekete
Joined: 13 Nov 05
Posts: 561
Credit: 4527270
RAC: 0

RE: RE: [Bruce suggested

Message 26155 in response to message 26154

Quote:
Quote:
[Bruce suggested to me Chebyshev polinomials instead of Taylor series a month ago. I did a fast test to compare these methods, but I found that Chebyshev approximation produced worse average by same number of coefficients. I'm working on a program that will generate more precise values)

See Section 5.8 page 192 of Numerical Recipes. You should also look at page 76 of Abramowitz and Stegun.

THANKS Bruce!

But I found a much faster method to calculate sin/cos values, so S39 has it!
Thanks to nadamhu!

KWSN Sir Clark
KWSN Sir Clark
Joined: 26 Jun 05
Posts: 42
Credit: 1200171
RAC: 0

S38 gave me 25 minutes less

S38 gave me 25 minutes less crunch time than C37....

Off to try S39........I can't keep up.

Comment viewing options

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