S40.03 Observations

archae86
archae86
Joined: 6 Dec 05
Posts: 3145
Credit: 7022954931
RAC: 1835422
Topic 191057

akosf has posted S-40.03. Both his comments and the chosen designation set the expectation that this version will be slight improvement over S-40.

Following StoneLord's example and methods, I've used a previous "captured" short Einstein unit so am comparing reported CPU time directly on the exactly same unit.

Gallatin (P4 EE 2M cache 3.2 GHz, WinXPPro)
hyperthread turned off for both sides of this test
r1_0265.5_2113_S4R2a_0

S-40 1159 seconds
S-40.03 1138 seconds

time ratio .982

I've installed it.
No validations yet

[AF>France] paulatreides
[AF>France] pau...
Joined: 22 Apr 05
Posts: 4
Credit: 36297
RAC: 0

S40.03 Observations

Hello, sorry for my poor english

So, what is better for my athlon64 processor with 3DNow, SSE, SSE2 capabilities?

install D40 or the new S40.03 ?

actually, i use D40 version...

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

S39L still seems to be

S39L still seems to be slightly faster on my Prescott.

me-[at]-rescam.org

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

Upgrading my a64x2 2.6gig

Upgrading my a64x2 2.6gig from D40 to S40.03.

D40 WUs ranged from 50-57m (53 average) in length. First 4 S40.03's ran in 49:10,51:03, 51:40, 51:53m. A slight gain, but only a percent or two. Significant mainly because since I upgraded to D40 I never saw a WU break the 50m mark.

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

RE: S39L still seems to be

Message 27703 in response to message 27701

Quote:
S39L still seems to be slightly faster on my Prescott.


A couple more units since have been slightly faster than S39L.
This one is too close to call. Which app's results are more accurate?

me-[at]-rescam.org

cmds
cmds
Joined: 1 Aug 05
Posts: 62
Credit: 348073
RAC: 0

With my two Pentium 4

Message 27704 in response to message 27703

With my two Pentium 4 (Northwood) SSE/SSE2 the difference between S39L / S40.3 will be only +- 60-120 s (short/long WU).
I calculate more than 20 WUs/host with S40.3

No need to change for Pentium4 Northwood from S39L to S40.3

Chris
----
Never change a running system

*Die Signatur befindet sich aus technischen Gründen auf der Rückseite dieses Beitrages!*

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

I have compared the average

I have compared the average of 9 S40 results with the average of 9 S40.03 results on an XP 2400++.
The WUs are in the range of r1_1256.5__758 to r1_1256.5__815.

Calculation time: - ~0.90%

Faktor: 0.99

On this host S40.03 will save about 12min/day.

cu,
Michael

Stick
Stick
Joined: 24 Feb 05
Posts: 790
Credit: 31192054
RAC: 415

My first result with S40.03

My first result with S40.03 finished successfully, but was not granted credit - not sure why/how this happened. Any ideas?

EDIT: That is, if my result was invalid, then why wasn't the WU sent out to another host (to get the third valid result)?

thezfunk
thezfunk
Joined: 25 Feb 05
Posts: 14
Credit: 7031224
RAC: 0

so then the general concensus

so then the general concensus is that S40 SSE is better for your AMD chips than the D40 with 3DNOW!?

J Langley
J Langley
Joined: 30 Dec 05
Posts: 50
Credit: 58338
RAC: 0

RE: so then the general

Message 27708 in response to message 27707

Quote:
so then the general concensus is that S40 SSE is better for your AMD chips than the D40 with 3DNOW!?

My Athlon XP 2800+ has 3DNow and SSE. The S40 and S40.03 versions are faster than D40 for me.

archae86
archae86
Joined: 6 Dec 05
Posts: 3145
Credit: 7022954931
RAC: 1835422

RE: EDIT: That is, if my

Message 27709 in response to message 27706

Quote:
EDIT: That is, if my result was invalid, then why wasn't the WU sent out to another host (to get the third valid result)?

The quorum must contain three results which look OK on return (i.e. the application on the volunteers machine did not detect error. But projects can set limits for how many of the results in the quorum must actually be found valid. For E@H that second requirement appears to be at most two.

So if one of the original three does not reply in time limit, or replies self-confessing error, then a replication will be sent to a fourth, but not in the case you cite.

Confession: I'm not an expert on this--those you know better please correct any error here.

Comment viewing options

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