S38 Observation thread

M. Schmitt
M. Schmitt
Joined: 27 Jun 05
Posts: 478
Credit: 15872262
RAC: 0

RE: I prefer polinomials

Message 26129 in response to message 26128

Quote:
I prefer polinomials than look-up table, because my Durons have just 256kB cache, altogether. :-)

Maybe You should collect money for a X2 with 2 · 1MB Cachè. ;-))

My Results so far, without any validation problems:

Athlon XP 2400+@2230MHz: -24,7%

Athlon X2 4400+@2616MHz: -25.5%

I changed from S38a to S39.

Excellent work!
I wish I could help to port it to Linux, but my last ASM-Coding was about 20 years ago. ;)

DanNeely
DanNeely
Joined: 4 Sep 05
Posts: 1364
Credit: 3562358667
RAC: 0

A64x2 at 2.4gig 1:35-1:05

A64x2 at 2.4gig

1:35-1:05 on big WUs 31% gain.
28-20m on small ones. 28% gain.

about 60% of my new WUs are still pending but no validation errors yet.

DanNeely
DanNeely
Joined: 4 Sep 05
Posts: 1364
Credit: 3562358667
RAC: 0

I just tried running s38 on

I just tried running s38 on my Athlon Classic. WUs errored out in a second or two with computation errors. I reverted to the old version I was running on it. Not sure which it was, dated 3/2/06.

M. Schmitt
M. Schmitt
Joined: 27 Jun 05
Posts: 478
Credit: 15872262
RAC: 0

RE: I just tried running

Message 26132 in response to message 26131

Quote:
I just tried running s38 on my Athlon Classic. WUs errored out in a second or two with computation errors. I reverted to the old version I was running on it. Not sure which it was, dated 3/2/06.

Athlon T-Birds do not support SSE, witch is necessary to run S38.
On Durons there seems to be problems too, but all Athlon XP should do the job.

My Athlon T-Bird is successfully runnig C37.

Ulrich Metzner
Ulrich Metzner
Joined: 22 Jan 05
Posts: 113
Credit: 963370
RAC: 0

Just a short question: For

Just a short question:
For the replacement of the client no app_info.xml is needed here, as opposed to Seti?

Aloha, Uli

DanNeely
DanNeely
Joined: 4 Sep 05
Posts: 1364
Credit: 3562358667
RAC: 0

correct. Just drop the new

correct. Just drop the new executable into place.

Ulrich Metzner
Ulrich Metzner
Joined: 22 Jan 05
Posts: 113
Credit: 963370
RAC: 0

Thanks a lot! Will do it

Thanks a lot! Will do it right now :D

Aloha, Uli

archae86
archae86
Joined: 6 Dec 05
Posts: 3160
Credit: 7259845334
RAC: 1522305

RE: So far my S-38/C-37

Message 26136 in response to message 26125

Quote:
So far my S-38/C-37 reports are based on a single "pure" S-38 result per CPU.
By "Dist" I mean performance on the unmodified Albert 4.37 science application as distributed by the project.


Here is an update, all four CPUs have at least one pure S-38 result completed.
[pre]CPU S-38/C-37 C-37/Dist S-38/Dist
P4 EE HT 0.862 0.516 0.445
Pentium M (Banias) 0.616 0.448 0.276
Pentium III 0.677 0.383 0.259
Pentium II 0.673 0.385 0.259[/pre]
While someone has suggested an AMD vs. Intel speedup difference, I think it may be a better fit to the data to say that older chips with (relatively) poorer floating point are speeding up by bigger percentages than the relatively more floating point capable ones.

Perhaps akosf managed to econimize even more on the overhead of the central routine than on the actual numeric content.

Misfit
Misfit
Joined: 11 Feb 05
Posts: 470
Credit: 100000
RAC: 0

S38 took 2200 seconds off of

S38 took 2200 seconds off of C37 on my 3.4 Prescott (HT enabled). Definately a winner!

me-[at]-rescam.org

B52
B52
Joined: 19 Feb 05
Posts: 45
Credit: 273899
RAC: 0

Can only concour with all the

Can only concour with all the above posts, amazing results on my first pc running S38 compared to C37.

WOW FANTASTIC

Comment viewing options

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