How big is the measured performance difference?
Each chart bar shows how many times more Time or how many times more Memory one unidentified ↓ chameneos-redux program used, compared to the program that used least Time or the program that used least Memory.
chameneos-redux benchmark N=6,000,000
This table shows 5 measurements - CPU Time, Elapsed Time, Memory, Code and ~ CPU Load.
Compare how much Memory the chameneos-redux programs used - sort Memory KB. Compare how much Code the programs used - sort Code B
Column × shows how many times more each program used compared to the program that used least.
| sort | sort | sort | sort | |||
|---|---|---|---|---|---|---|
| × | Program Source Code | CPU secs | Elapsed secs | Memory KB | Code B | ~ CPU Load |
| 1.0 | C GNU gcc #5 | 3.14 | 0.79 | 684 | 2879 | 99% 99% 99% 99% |
| 4.4 | C++ GNU g++ #5 | 13.05 | 3.47 | 1,332 | 1994 | 98% 100% 89% 89% |
| 4.5 | C++ GNU g++ #4 | 13.05 | 3.57 | 1,328 | 1800 | 82% 77% 100% 100% |
| 5.7 | C++ GNU g++ #3 | 15.57 | 4.54 | 1,324 | 1845 | 71% 64% 100% 99% |
| 8.8 | Haskell GHC | 7.02 | 7.02 | 2,108 | 789 | 0% 0% 0% 100% |
| 9.2 | Java 6 steady state #4 | 21.42 | 7.27 | 19,080 | 1464 | 57% 50% 81% 81% |
| 9.5 | Haskell GHC #2 | 14.83 | 7.55 | 2,728 | 832 | 1% 84% 85% 0% |
| 10 | Go 6g 8g #4 | 8.10 | 8.11 | 836 | 996 | 0% 50% 49% 0% |
| 11 | C++ GNU g++ | 29.24 | 8.45 | 1,416 | 1798 | 72% 67% 100% 100% |
| 11 | Java 6 -server #3 | 12.57 | 8.64 | 41,288 | 1756 | 37% 45% 15% 40% |
| 11 | Java 6 -server #4 | 25.61 | 8.81 | 15,908 | 1377 | 55% 79% 53% 83% |
| 13 | Go 6g 8g #3 | 10.11 | 10.11 | 832 | 983 | 2% 0% 100% 2% |
| 17 | Java 6 -server #2 | 40.82 | 13.30 | 36,136 | 1429 | 61% 79% 62% 78% |
| 17 | C GNU gcc #2 | 44.66 | 13.63 | 664 | 1663 | 74% 72% 86% 85% |
| 18 | Java 6 -Xint #4 | 43.58 | 14.37 | 12,532 | 1377 | 62% 89% 53% 88% |
| 18 | Java 6 steady state #2 | 46.73 | 14.61 | 36,700 | 1513 | 64% 65% 81% 80% |
| 19 | Scala #2 | 44.51 | 14.75 | 37,384 | 1171 | 62% 60% 76% 75% |
| 25 | Go 6g 8g | 19.66 | 19.67 | 872 | 1142 | 49% 0% 0% 51% |
| 39 | C# Mono | 109.38 | 30.96 | 7,548 | 1409 | 82% 85% 80% 84% |
| 46 | C++ GNU g++ #2 | 86.83 | 36.23 | 1,428 | 1606 | 58% 56% 53% 53% |
| 46 | Java 6 -Xint #2 | 96.48 | 36.43 | 33,160 | 1429 | 54% 66% 51% 67% |
| 57 | Java 6 -server | 126.22 | 45.05 | 85,824 | 1462 | 60% 67% 53% 68% |
| 76 | Smalltalk VisualWorks #2 | 60.07 | 60.06 | 22,856 | 2515 | 0% 100% 0% 0% |
| 115 | Scala | 87.62 | 90.93 | 145,532 | 1234 | 23% 31% 23% 17% |
| 120 | Java 6 -Xint | 248.22 | 95.19 | 94,680 | 1462 | 54% 64% 52% 65% |
| 138 | Erlang HiPE | 198.65 | 109.96 | 9,144 | 734 | 47% 45% 30% 33% |
| 155 | C GNU gcc | 167.00 | 122.78 | 740 | 1769 | 33% 25% 38% 31% |
| 186 | Ruby JRuby | 5 min | 147.34 | 95,944 | 928 | 45% 44% 56% 54% |
| 199 | Ruby MRI | 158.27 | 158.26 | 2,656 | 928 | 0% 0% 100% 0% |
| 267 | Scheme PLT | 211.77 | 211.93 | 36,364 | 792 | 1% 0% 100% 0% |
| 273 | Python CPython #2 | 286.54 | 216.63 | 3,816 | 869 | 49% 24% 28% 22% |
| 302 | OCaml #2 | 6 min | 239.77 | 2,412 | 1286 | 37% 33% 37% 33% |
| 317 | OCaml | 6 min | 252.01 | 2,400 | 1279 | 36% 35% 30% 38% |
| 342 | Go 6g 8g #2 | 9 min | 271.65 | 920 | 1170 | 54% 49% 47% 47% |
| 377 | Java 6 -Xint #3 | 7 min | 299.59 | 33,292 | 1756 | 33% 41% 20% 40% |
| Ada 2005 GNAT | Bad Output | 1615 | ||||
| missing programs | ||||||
| ATS | No program | |||||
| Clean | No program | |||||
| F# Mono | No program | |||||
| Fortran Intel | No program | |||||
| JavaScript TraceMonkey | No program | |||||
| JavaScript V8 | No program | |||||
| Lisp SBCL | No program | |||||
| Lua | No program | |||||
| Pascal Free Pascal | No program | |||||
| Perl | No program | |||||
| PHP | No program | |||||
chameneos-redux benchmark : Symmetrical thread rendezvous requests
ndiff program output N = 600 with this output file to check your program output has the correct format before contributing.
The text-part and spelled-out numbers in the program output should match the expected output exactly. Thread scheduling may cause small differences in the other numbers, so program output is checked with ndiff -fields 2-10
Each program should
- create differently coloured (blue, red, yellow), differently named, concurrent chameneos creatures
- each creature will repeatedly go to the meeting place and meet, or wait to meet, another chameneos "(at the request the caller does not know whether another chameneos is already present or not, neither if there will be one in some future)"
- both creatures will change colour to complement the colour of the chameneos that they met - don't use arithmetic to complement the colour, use if-else or switch/case or pattern-match
- write all the colour changes for blue red and yellow creatures, using the colour complement function
- for rendezvouses with an odd number of creatures (blue red yellow) and with an even number of creatures (blue red yellow red yellow blue red yellow red blue)
- write the colours the creatures start with
- after N meetings have taken place, for each creature write the number of creatures met and spell out the number of times the creature met a creature with the same name (should be zero)
- spell out the sum of the number of creatures met (should be 2N)
The chameneos benchmark is an adaptation of "Chameneos, a Concurrency Game for Java, Ada and Others" 100KB pdf (which includes example implementations in Java, Ada and C).
Programs may use kernel threads, lightweight threads; but coroutines, cooperative threads and other programs with custom schedulers will be listed as interesting alternative implementations. Briefly say what concurrency technique is used in the program header comment.