Latest Update: 28.03.2022
Dear Chess Friends,
As usual, I'm pleased to announce that,I've organized a new Time Control competition!
Note that from time to time, I run similar TC tourneys!And for anyone interested/missed:: 2021 / 2019 / 2014
Well, via this competition the target will be to see the Elo influenceBetween various time controls such as Bullet, Blitz, Rapid and Slow...
In short, There are Auto232 mode matches between e.g 60m+10s vs 15m+2s etc.
As Participants,I picked the ones, which are more stable under these time control conditions,Because many engines suffer or forfeits on time e.g at Ultra fast time controlsI preferred NON-NNUE engines, e.g,which are mainly average 2000-3600 Elo Because, in case of NNUE 3700+ Elo, almost all games can be ended draws..As you may know, if we test via Evalfile...then the engines will start to benefit From the previous played / learned data,I mean not so good idea as NNUEMeanwhile, for these kind of TC testings: Ponder ON is not so good idea too...The reason, if Ponder ON, then the engine will benefit during opponent's turnThat means via NNUE or Ponder ON,All the played results can lead to wrong Elo comparing, measuring...!That's why I used to play as Ponder OFF and NON-NNUE !
About Time Controls,For example, as 1st start, I used to play various TC:
60 min + 10 sec = 4250 sec (70.83 min)
15 min + 2 sec = 1030 sec (17.16 min)
3 min + 1 sec = 245 sec (4.08 min)
22 sec + 0.6 sec = 61 sec (1.01 min)
Note also that for the above Time Control calculations,As average 65 moves (per side) and x2 (two) times for doubling are usedFrankly, according to my older and current TC testings.There is no any right standard Elo calculation formula e.g in case of doubling, It all depends, some engines perform close to 40-50 Elo, where some more...But note also just in case of doubling via Ultra fast time controls...then the EloDifference is huge, not 50 Elo, minus plus 100-200 Elo, sure via Cfish 030820
Meanwhile, here are the 1st played results:
1st Test - Bullet Vs Blitz: 200 Elo
1 3m+1s +107/-1/=96 75.98% 155.0/2042 22s+06s +1/-107/=96 24.02% 49.0/204
------------------------------------------------------------------------------
2nd Test - Blitz Vs Rapid: 104 Elo 1 15m+2s +46/-4/=94 64.58% 93.0/1442 3m+1s +4/-46/=94 35.42% 51.0/144
-----------------------------------------------------------------------------
3rd Test - Rapid Vs Slow: 92 Elo 1 60m+10s +27/-1/=72 63.00% 63.0/1002 15m+2s +1/-27/=72 37.00% 37.0/100
Some Notes:The above tests are played by 4-times Champion Cfish 030820 C40The Hash-table size are increased for Rapid and Slow time controls..E.g 512 MB for Rapid /1024 MB hash for Slow time controls are usedFor Blitz: 128 MB, where the Ultra fast TC are played via 64 MB hash...
And here are the latest results, using various engines and TC:
4th. Fast Bullet vs Bullet: 106 Elo
1 1m+1s +44/-8/=70 64.75% 79.0/1222 22s+06s +8/-44/=70 35.25% 43.0/122
5th Ultra Fast vs Fast Bullet : 186 Elo
1 22s+06s +104/-5/=93 74.50% 150.5/2022 11s+03s +5/-104/=93 25.50% 51.5/202
Note: The both testings are played via Cfish 030820 C40 tooPlus, via 5th test, it seems the doubling is more than 180+ Elo-------------------------------------------------------------------------------------------------
6th. Ultra Fast TC (Cfish 030820) Vs Blitz (Rybka 3): 67 Elo
1 Rybka3 3m+1s +88/-52/=48 59.57% 112.0/1882 Cfish030820 8s+02s +52/-88/=48 40.43% 76.0/188
Note: Normally Blitz Elo difference between both vers. is approx. 570 Elo---------------------------------------------------------------------------------------------------
7th Ultra Fast TC (Cfish 030820) Vs Blitz (Chessmaster 11): 107 Elo
1 Chessmaster11 3m+1s +118/-56/=34 64.90% 135.0/2082 Cfish030820 4s+02s +56/-118/=34 35.10% 73.0/208
Note: Chessmaster 11 engine is played via Innovation setting: ResultsNormally the Blitz Elo difference between both vers. is approx. 860 Elo---------------------------------------------------------------------------------------------------
8th. Fast TC (Houdini 4) Vs Blitz (Rybka 3): 16 Elo
1 Rybka3 3m+1s +39/-33/=58 52.31% 68.0/1302 Houdini4 22s+06s +33/-39/=58 47.69% 62.0/130
Note: Normally Blitz Elo difference between both vers. is approx. 200 Elo-------------------------------------------------------------------------------------------------- 9th. Fast Bullet (Shredder 5) Vs Bullet (PikoSzachy 4): 8 Elo
1 PikoSzachy4 1m+1s +53/-50/=29 51.14% 67.5/1322 Shredder5 22s+06s +50/-53/=29 48.86% 64.5/132
Note: Normally Blitz Elo difference between both vers. is approx. 130 Elo------------------------------------------------------------------------------------------------- 10th. Fast Bullet (Shredder 5) Vs Blitz (Amateur 2.82): 47 Elo
1 Shredder5 22s+06s +79/-56/=37 56.69% 97.5/1722 Amateur282 3m+1s +56/-79/=37 43.31% 74.5/172
Note: Normally Blitz Elo difference between both vers. is approx. 240 Elo-------------------------------------------------------------------------------------------------
11th. Fast Bullet (Feuerstein0461) Vs Blitz (Tyrell 1.02): 64 Elo
1 Feuerstein0461 22s+06s +55/-34/=27 59.05% 68.5/1162 Tyrell102 3m+1s +34/-55/=27 40.95% 47.5/116
Note: Normally Blitz Elo difference between both vers. is approx. 230 Elo-------------------------------------------------------------------------------------------------
More Details,Just amazing, the Top engines even under these conditions are very strong.I tried many other TC too, but I tested, which are more close in Elo strengthThe current competition is ended, but there is a possibility to be resumed...As usual, all of the played test games available SCCT Time Control III
And I hope, All of the current new data will be useful and good indicator for comparing...And in my estimation, probably:3600+ engines via Bullet are close in strength as GM via Slow time controls!Or Top engines via Ultra fast TC are on level as many GMs via Blitz, Rapid etc.
Have fun,Sedat