Difference between revisions of "Computer Composition Searches"

From Changeringing Wiki
Jump to: navigation, search
(New Full Monty Benchmark (Intel))
Line 5: Line 5:
 
! Method                    || Search Constraints      || Results || Time/<br>Computer/<br>Program || Date || Run by||Notes
 
! Method                    || Search Constraints      || Results || Time/<br>Computer/<br>Program || Date || Run by||Notes
 
|-
 
|-
| Cambridge Surprise Major || >=5000 changes<br>Tenors together<br>Round blocks<br>Bob = 14, Single = 1234 || 9,997 compositions<br>255,138 rotations || 9 days<br>P90 PC<br>SMC<br><br>1:42:30<br>2.2GHz Athlon X2<br>SMC32<br><br>1:03:04<br>3.2GHz Phenom X4<br>SMC32||1995<br><br><br><br>2005<br><br><br><br>2010||Graham John<br><br><br><br>Mark Davies<br><br><br><br>Simon Humphrey||See [http://www.changeringing.co.uk/cambridgefm.htm Cambridge - The Full Monty]
+
| Cambridge Surprise Major || >=5000 changes<br>Tenors together<br>Round blocks<br>Bob = 14, Single = 1234 || 9,997 compositions<br>255,138 rotations || 9 days<br>P90 PC<br>SMC<br><br>
 +
1:42:30<br>2.2GHz Athlon X2<br>SMC32<br><br>
 +
1:03:04<br>3.2GHz Phenom X4<br>SMC32<br><br>
 +
1:03:00<br>4GHz Core i7 930*<br>SMC32
 +
||1995<br><br><br><br>2005<br><br><br><br>2010<br><br><br><br>2010||<br>Graham John<br><br><br><br>Mark Davies<br><br><br><br>Simon Humphrey<br><br><br><br>Graham John<br>&nbsp;||See [http://www.changeringing.co.uk/cambridgefm.htm Cambridge - The Full Monty]<br><br><br><br><br><br><br><br><br><br><br><br>*Leaving 7 processor threads idle!
 
|-
 
|-
 
| Cambridge Surprise Major || >=5000 changes<br>Tenors together<br>Snap finishes<br>Bob = 14, Single = 1234 || 15,416 compositions || 13 days<br>PC<br>SMC32||1998 || Graham John||See [http://www.changeringing.co.uk/cambridgefm.htm Cambridge - The Full Monty]
 
| Cambridge Surprise Major || >=5000 changes<br>Tenors together<br>Snap finishes<br>Bob = 14, Single = 1234 || 15,416 compositions || 13 days<br>PC<br>SMC32||1998 || Graham John||See [http://www.changeringing.co.uk/cambridgefm.htm Cambridge - The Full Monty]

Revision as of 09:24, 3 December 2010

A computer can be a useful tool to search for compositions, yet even with today's fast processors exhausting many composition searches is too big a task. This page has been created to keep a record of searches people have successfully completed in popular methods. If you have completely exhausted a computer search for compositions, please add an entry to the table below, stating any constraints you applied to the search, when you first ran it, and any other information you think would be of interest.

Exhausted Composition Searches

Method Search Constraints Results Time/
Computer/
Program
Date Run by Notes
Cambridge Surprise Major >=5000 changes
Tenors together
Round blocks
Bob = 14, Single = 1234
9,997 compositions
255,138 rotations
9 days
P90 PC
SMC

1:42:30
2.2GHz Athlon X2
SMC32

1:03:04
3.2GHz Phenom X4
SMC32

1:03:00
4GHz Core i7 930*
SMC32

1995



2005



2010



2010

Graham John



Mark Davies



Simon Humphrey



Graham John
 
See Cambridge - The Full Monty











*Leaving 7 processor threads idle!
Cambridge Surprise Major >=5000 changes
Tenors together
Snap finishes
Bob = 14, Single = 1234
15,416 compositions 13 days
PC
SMC32
1998 Graham John See Cambridge - The Full Monty
Cambridge Surprise Major >=5000 changes
Tenors together
Round blocks
Bob = 14, Single = 1256
152,915 compositions
3,739,182 rotations
47 days
2.8GHz i7-860
SMC32
2010 Mark Davies
Bristol Surprise Maximus 5000-5100 changes
Tenors together
Round blocks
Bob = 14, Single = 1234
202,384,361 compositions 47 hours
PC
SMC32
1998 Mark Davies
Yorkshire Surprise Maximus 5000-5100 changes
Tenors together
Round blocks
Bob = 14, Single = 1234, Big Bob=18
202,927,179 compositions 5 days, 23 hours
PC
SMC32
1998 Mark Davies
Plain Bob Minor 720 changes
Bob = 14, Single = 1234
1,440,339,152 compositions 18:16:25
2.8GHz i7-860
SMC32
2010 Mark Davies

Composing Program Benchmark

The "Full Monty" search for compositions of Cambridge Surprise Major is a useful benchmark both for measuring the efficiency of composing programs and the computers they run on. As shown in the table above, the first time this was completed was in 1995 using SMC on a Pentium 90, taking 9 days. The current record for this search is one hour, three minutes, and four seconds run by Simon Humphrey using SMC32 on a Phenom X4 3.2 GHz in 2010. It should be possible to improve on this significantly using a multi-threaded program on a modern multi-processor PC. If you beat it, please update this page.