S39 Observation Thread

J D K
J D K
Joined: 27 Aug 05
Posts: 86
Credit: 103,878
RAC: 0

RE: Hi! I'm working on S39L

Message 26259 in response to message 26258

Quote:

Hi! I'm working on S39L that uses less L1 data cache (~10kB).
Good news to Pentium-III owners.
But now, I'm sharing my measured results with you.

Pentium4 mobile 1,8 GHz (Northwood): 30200 -> 7600 -> 397%
Athlon XP 1,53 GHz (ThoroughbredB): 21200 -> 5550 -> 382%
PentiumM 1,86 GHz (Dothan): 18500 -> 4500 -> 411%
Duron 2 GHz (Applebread): 15200 -> 4200 -> 362%

S39L needed 4100 sec on one of my Durons. :-)

Have you looked at this result, Zero time, but got credit...

http://einsteinathome.org/host/106110/tasks

Akos Fekete
Akos Fekete
Joined: 13 Nov 05
Posts: 561
Credit: 4,527,270
RAC: 0

RE: Have you looked at this

Message 26260 in response to message 26259

Quote:

Have you looked at this result, Zero time, but got credit...

http://einsteinathome.org/host/106110/tasks

:-)
I think you should reinstall BOINC.

Stef
Stef
Joined: 8 Mar 05
Posts: 206
Credit: 110,568,193
RAC: 0

RE: Hi! I'm working on S39L

Message 26261 in response to message 26258

Quote:
Hi! I'm working on S39L that uses less L1 data cache (~10kB).
Good news to Pentium-III owners.


OMG! If you carry on, it will end up by the discovery of a gravitational wave using a single pocket calculator!

ca_grufti
ca_grufti
Joined: 9 Feb 05
Posts: 53
Credit: 4,309,237
RAC: 0

Just a few quick observations

Just a few quick observations on the economic impact of S39 and the contribution that Akos has made. From the current server stats:

Active hosts: 75,218
potential flops: 93.8
actual flops: 41.3

calculated number of hosts actively crunching for Einstein:
75,218*41.3/93.8 = 33,118

Power consumption per host approx. 300W+ [some are multi CPU, count fans, power supplies, hard drives, screens sometimes on]

10,000kW at US $ 0.20 per kW/h = $2,000 per hour / $48,000 per day

calculating all S4 results in approx 200 days: $9,600,000.00 total energy cost to the Einstein contributors.

calculating all S4 results in approx 50 days: $2,400,000.00 total energy cost to the Einstein contributors.

Savings $7,200,000.00 for S4 analysis.

So maybe it's just $5M ... or maybe I have a mistake in those calculations somewhere, whatever, it's huge!

Hans

KWSN-GMC-Peeper of the Castle Anthrax
KWSN-GMC-Peeper...
Joined: 6 Oct 05
Posts: 62
Credit: 54,755,621
RAC: 0

RE: 10,000kW at US $

Message 26263 in response to message 26262

Quote:

10,000kW at US $ 0.20 per kW/h = $2,000 per hour / $48,000 per day
Hans

$.20/kwh????? California boy, eh? lol.

I just popped in today to see what was going on..noticed the new S39 and installed it.
I've been doing ~ 5 hours/wu on the original, stock albert. My first WU on the S39 looks to be completeing in ~ 1 hour 15 min.

ca_grufti
ca_grufti
Joined: 9 Feb 05
Posts: 53
Credit: 4,309,237
RAC: 0

As a worldwide average price

As a worldwide average price per kWh at retail USD 0.20 is probably pretty close. Never mind CA. You have Australia, France, Germany, Hungary, ... , and yes the US in that mix. But who cares, if $0.20 is a little high. Just over 300W per host is almost certainly low, when you take all inefficiencies and components into account. 200 days for S4 before optimization is low as you can see from the stats. The way I calculated active hosts probably tends toward low.

In essence: the energy cost of doing the S4 analysis goes from roughly $10M to roughly $2.5M after optimization.

That's just the economic side to it, there are pretty obvious ecological benefits too.

Code review and optimization should become a mandatory step for all BOINC apps that want to be considered "production apps". The wastefulness of not running optimized code on hundreds of thousands of machines is plain ugly.

DanNeely
DanNeely
Joined: 4 Sep 05
Posts: 1,364
Credit: 3,562,358,667
RAC: 0

I'm seeing modest gains with

I'm seeing modest gains with S39 over S38 on my A64x2 2.4gig. 65-70m vs 75-80m

As far as 20c/kwh goes I'll file that as reason not to move to CA #392. I'm only paying 8. :)

Nightbird
Nightbird
Joined: 17 Feb 05
Posts: 79
Credit: 561,723
RAC: 0

First wu with S39 on a Barton

First wu with S39 on a Barton 2500+ :
"before", between 5,335.00 and 5,024.05 sec
now : for the first time, under the 5,000 sec : 4,720.00 sec.

[

Terry
Terry
Joined: 9 Feb 05
Posts: 55
Credit: 1,781,475
RAC: 0

RE: First wu with S39 on a

Message 26267 in response to message 26266

Quote:
First wu with S39 on a Barton 2500+ :
"before", between 5,335.00 and 5,024.05 sec
now : for the first time, under the 5,000 sec : 4,720.00 sec.

If your mb supports 400 MHz FSB, your cpu will easily run as a 3200+.

Nightbird
Nightbird
Joined: 17 Feb 05
Posts: 79
Credit: 561,723
RAC: 0

RE: RE: First wu with S39

Message 26268 in response to message 26267

Quote:
Quote:
First wu with S39 on a Barton 2500+ :
"before", between 5,335.00 and 5,024.05 sec
now : for the first time, under the 5,000 sec : 4,720.00 sec.

If your mb supports 400 MHz FSB, your cpu will easily run as a 3200+.


Yes, the mobo supports 400 mhz fsb and my 2500+ is not locked but never i tried to overclock it.

[

Comment viewing options

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