Akosf question: A36 vs C37?

josep
josep
Joined: 9 Mar 05
Posts: 63
Credit: 1,156,542
RAC: 0

RE: S37a works on my P4

Message 25862 in response to message 25854

Quote:

S37a works on my P4 Northwood 2.4Ghz, but it seems slower than C37 (8% slower in a first and not very accurate partial measure)

Update: I've taken a second measure, in a new "mixed" WU: 30% completed with S37a, and the remaining with C37.
The results: S37a is 9% slower than C37 on my P4 Northwood

Alinator
Alinator
Joined: 8 May 05
Posts: 927
Credit: 9,352,143
RAC: 0

RE: Hey Alinator! I did C37

Message 25863 in response to message 25860

Quote:

Hey Alinator!
I did C37 for your K6...

Hey Michael!
We will not do problem from a letter...
...'f' is coming from my family name.

I'm glad that good speed of C37. :)

Okee Dokie, will set them up with C37 for the next set of results.

Alinator

Crunchers For More Power
Crunchers For M...
Joined: 3 Aug 05
Posts: 69
Credit: 1,071,273
RAC: 0

Ok, finished a run with C37

Ok, finished a run with C37 vs. S37 on P4 Prescott: C37 is as fast as S37 on this machine (diffence: 10 seconds).

So I will try it on the Yonah-Nootbook in next days (and on a Opteron if my team mate is willing to spend some cpu cycles ;)) ... and finally buy an Conroe ;)

CU HiNuN

Honasss
Honasss
Joined: 20 Aug 05
Posts: 2
Credit: 411,778
RAC: 0

A36 vs. C37 on

A36 vs. C37 on AthlonXP2600+

C37 is about 15% faster (compare 3 results with A36, 1 result with C37)

Jason A. Countryman
Jason A. Countryman
Joined: 7 Mar 05
Posts: 2
Credit: 30,572
RAC: 0

Ive noticed that running

Ive noticed that running these does the same as seti@home's optimized apps, it lowers the requested credits. has anyone checked to see if using the trux optimized client will request higher credits with optimized e@home?

Peter M. Nielsen
Peter M. Nielsen
Joined: 18 Jan 05
Posts: 2
Credit: 3,336
RAC: 0

Well, it is the way it is

Message 25867 in response to message 25866

Well, it is the way it is supposed to work. You use less cpu-time therefore you should have less credit...

- Peter

Honasss
Honasss
Joined: 20 Aug 05
Posts: 2
Credit: 411,778
RAC: 0

RE: .. has anyone checked

Message 25868 in response to message 25866

Quote:
.. has anyone checked to see if using the trux optimized client will request higher credits with optimized e@home?

My client(5.3.12tx36)still blocking calibration.
- negative calibration limit ... without opt. or using A36,C37 :-(

Pepperammi
Pepperammi
Joined: 20 Feb 05
Posts: 131
Credit: 437,943
RAC: 0

Hi i'm slightly confused as

Hi i'm slightly confused as to which of these two optimised codes i'm suppose to go for. Either the C37 or the S37a?

Reading through is seems that the S37a has enhanced instructions SSE2 which my cpu supports but it seems some people are saying its slower?

Anyway i realise it would be processor/core specifis so i used cpuid to find any extra info that might help. Don't know much about the 'paths' ect. but its;

Petium D 830, Dual core 3ghz
Codeame; Smithfield
Package; LGA775
Family; F Model; 4
Revision; A0
Instructions; MMX,SSE,SSE2,SSE3,EM64T

My other computer is;
Pentium 4 3.4ghz HT
I think its either Prescott or Northwood but i'll have to have a look

Crunchers For More Power
Crunchers For M...
Joined: 3 Aug 05
Posts: 69
Credit: 1,071,273
RAC: 0

RE: has anyone checked to

Message 25870 in response to message 25866

Quote:
has anyone checked to see if using the trux optimized client will request higher credits with optimized e@home?

I do. It works well and claim about 37 credits for "standard albert"

http://einsteinathome.org/host/407768/tasks

CU HiNuN (CFMP)

Crunchers For More Power
Crunchers For M...
Joined: 3 Aug 05
Posts: 69
Credit: 1,071,273
RAC: 0

RE: Well, it is the way it

Message 25871 in response to message 25867

Quote:

Well, it is the way it is supposed to work. You use less cpu-time therefore you should have less credit...

- Peter

Yes and No. That's the way the benchmark based credit claiming work.
But the idea behind the credit system is "If you do the same work as someone else you should receive the same amount of credit" ... assuming that faster processing of real work is reflected by a higher benchmark value which will increase the claimed credit.

Therefore the future seti app for example will use a different base for claimed credit. It will count work steps instead use the benchmark.

CU HiNuN

Comment viewing options

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