How is GIMPS' sustained throughput defined? - Page 6

kracker · 2014-02-14, 22:17

Quote:

Originally Posted by TheMawn

"sustained"

NBtarheel_33 · 2014-02-15, 00:29

I'd say 30 days sustained would be a fair definition of "sustained". On the other hand, it would be interesting to see TFLOP/s figures for the last year, or even since the inception of GIMPS.

TheMawn · 2014-02-15, 03:49

Quote:

Originally Posted by NBtarheel_33

I'd say 30 days sustained would be a fair definition of "sustained". On the other hand, it would be interesting to see TFLOP/s figures for the last year, or even since the inception of GIMPS.

*Insert Exponentional Curve Here*

NBtarheel_33 · 2014-04-26, 17:59

Just a hair under 222 TFLOP/s at the moment.

Xyzzy · 2014-04-27, 00:32

We use this system to compute the effectiveness of our diet.

http://www.fourmilab.ch/hackdiet/www...2_4_0_4_3.html

We think it could be applied to GIMPS WRT sustained output.

NBtarheel_33 · 2015-02-11, 08:56

GIMPS' potential 30-day computing power has broken the 600 TFLOP/s barrier. This is the expected computing power that would be attained if every computer registered to GIMPS were to sustain 100% efficiency for 30 days. Typically, GIMPS runs between 30-35% efficiency (i.e. 180-195 TFLOP/s) over a rolling 30-day interval.

The all-time highest sustained 30-day actual computing power remains ~222 TFLOP/s, achieved over the 30-day period preceding April 26, 2014. This figure corresponds to ~37% efficiency.

alpertron · 2015-02-11, 11:29

Maybe I'm wrong, but it appears that the potential computing power only considers CPU, while the actual computing power also include GPU output, so they are not comparable.

VictordeHolland · 2015-02-11, 12:50

Quote:

Originally Posted by alpertron

Maybe I'm wrong, but it appears that the potential computing power only considers CPU, while the actual computing power also include GPU output, so they are not comparable.

I think you might be right there, when some of the big guns upload a weeks worth of GPU(s) results the curve jumps up considerably. The real question is whether manual results are counted towards the sustained throughput?

NBtarheel_33 · 2015-04-15, 06:30

GIMPS has just set a new record of 226+ TFLOP/s for 30-day sustained throughput.

NBtarheel_33 · 2015-04-18, 22:33

228.5 TFLOP/s now.

NBtarheel_33 · 2015-04-19, 16:22

Wow! 234.857 TFLOP/s sustained. This is equivalent to 40% efficiency over the last 30 days, which is the highest that I have ever seen for that reading. Efficiency for the past week is at 101%, or 281+ TFLOP/s.

Big influx of new users or big iron in the last couple of weeks, perhaps?

2014-04-26, 17:59	#59
NBtarheel_33 "Nathan" Jul 2008 Maryland, USA 2133₈ Posts	Just a hair under 222 TFLOP/s at the moment. Last fiddled with by NBtarheel_33 on 2014-04-26 at 18:00

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
another defined sequence	MattcAnderson	Miscellaneous Math	4	2017-07-05 08:05
Budget PC Throughput	Rodrigo	Hardware	14	2011-09-26 10:16
SZPTR - whare is it defined	tichy	Software	4	2010-12-22 09:02
how is the throughput calculated?	ixfd64	PrimeNet	5	2008-05-21 13:39
My throughput does not compute...	petrw1	Hardware	9	2007-08-13 14:38

2014-02-15, 00:29	#57
NBtarheel_33 "Nathan" Jul 2008 Maryland, USA 2133₈ Posts	I'd say 30 days sustained would be a fair definition of "sustained". On the other hand, it would be interesting to see TFLOP/s figures for the last year, or even since the inception of GIMPS.

2014-04-27, 00:32	#60
Xyzzy "Mike" Aug 2002 2×11×19² Posts	We use this system to compute the effectiveness of our diet. http://www.fourmilab.ch/hackdiet/www...2_4_0_4_3.html We think it could be applied to GIMPS WRT sustained output.

2015-02-11, 08:56	#61
NBtarheel_33 "Nathan" Jul 2008 Maryland, USA 2133₈ Posts	GIMPS' potential 30-day computing power has broken the 600 TFLOP/s barrier. This is the expected computing power that would be attained if every computer registered to GIMPS were to sustain 100% efficiency for 30 days. Typically, GIMPS runs between 30-35% efficiency (i.e. 180-195 TFLOP/s) over a rolling 30-day interval. The all-time highest sustained 30-day actual computing power remains ~222 TFLOP/s, achieved over the 30-day period preceding April 26, 2014. This figure corresponds to ~37% efficiency.

2015-02-11, 11:29	#62
alpertron Aug 2002 Buenos Aires, Argentina 1341₁₀ Posts	Maybe I'm wrong, but it appears that the potential computing power only considers CPU, while the actual computing power also include GPU output, so they are not comparable.

2015-04-15, 06:30	#64
NBtarheel_33 "Nathan" Jul 2008 Maryland, USA 5×223 Posts	GIMPS has just set a new record of 226+ TFLOP/s for 30-day sustained throughput.

2015-04-18, 22:33	#65
NBtarheel_33 "Nathan" Jul 2008 Maryland, USA 5×223 Posts	228.5 TFLOP/s now.

2015-04-19, 16:22	#66
NBtarheel_33 "Nathan" Jul 2008 Maryland, USA 2133₈ Posts	Wow! 234.857 TFLOP/s sustained. This is equivalent to 40% efficiency over the last 30 days, which is the highest that I have ever seen for that reading. Efficiency for the past week is at 101%, or 281+ TFLOP/s. Big influx of new users or big iron in the last couple of weeks, perhaps?