World Championship Winning Computer Chess Software Program & Downloads for Chess Databases, Analysis and Play on PC, Mac, iPad and iPhone — Visit: Hiarcs.com
Put 9 nine blind chess computers enthusiasts in a room and you get 9 different games...or another analogy equally nonsensical and befuddling.
I have two philosophies regarding my particular chess computer tournaments:
1. It is what it is
2. "Let 'em play"
I remember on SSDF, quite a while back, the moderator had a similar philosophy and stated something to the effect that:
If a chess computer program "chooses" to play the same game and lose in the same way...well then that is the programs problem".
My final note after the 20 game match (probably will post by Sunday) deals with the (disappointing) very limited opening repetoire of Fidelity...which as you likely already know:
Plays ONLY 1.e4 or 1.d4 as White!
Let's see, we are on game 12 now....unfortunately there will be several more repeat games in the upcoming 8 remaining games, but it is what it is.
So, I just "let 'em play" whatever the programs choose.
I do believe that my methodolgy skews the strength of the Fidelty Mach4-33 upwards unfairly. Fidelity has cherry picked a group of the strongest lines. Novag has lots more variety (and thus, some perhaps "suspect" openings) and even in the last 8 games you will see some suprise openings from Novag.
But, overall the tournament was a lot of fun and I learned something about the two combatants "style", programming and there was a bit of a surprise result at the end.
C99e: Ruy Lopez: Closed in book for 13 full moves. SD 14. Re3 is a little premature; book continuations are Nf1 and d5. SD would have had its biggest lead (and ONLY its second one) with 16. d5, but calculates 16. dxe5? and is behind -0.8. Next, another small mistake: 17. Nxe5? (Qe2) and M4 is up -1.6. M4 fails to capitalize with 17. .. Nxe4! and calculates weaker Bc5. But, SD is having a bad day and 18. Ng4? (Qe1) gives Black a -2.6 lead. M4-33 is in the midst of an attack when it calculates strange King “tuck” 24. .. Kh8? (Be5!). I guess M4 “saw” something here, but all that SM sees is an eval drop to -1.6. SD also “sees” something: a b4 pawn that is vulnerable and plays 29. Ba3? (Be3); the M4 lead is back up to -2.5! And here is where M4-33 starts moving like SmallFish: 29. .. Re8, 30. .. Re6, 31. .. Rce8, 32. .. Bf6! and 33. .. Bxe4! SM says M4 is now leading by -4.1! Game Over, except continuation 34. .. c3 was a little weak; admirable, but not as strong as 34. .. Qd8! At this point SD panics with 35. Re2?? GAME OVER!!!!! (Maybe that’s what M4 “saw” with 34. .. c3)? There was a period of garbage time, but M4 promotion 46. .. c1=Q puts this issue to bed: SD must lose a Rook. 10 moves later, Star Diamond resigns. SD should have resigned at move 47 (BUT, AT LEAST SD HAS THE COURTESY TO RESIGN). Great attacking game by M4-33, taking advantage of some weak SD moves!
Game 14 An absolute, 100% EXACT repeat of Game 8 flat line draw
B99u: Sicilian: Najdorf in book THROUGH WHITE MOVE 20!!!!!!.....
At move 75, I had had enough, as a B+P vs. B+P draw was a certainty: the two morons would have played another 50 moves to do the same!
After 8 games, M4-33 has reached 5 wins and will at least draw the first section of 10 games. The elo calculator says that M4-33 is playing 89 points higher than Star Diamond!
In this game, I had enough after Queen trade 69.Qxd4+ Kxd4. Game was a 100% EXACT copy of Game 8 and a clear draw. M4-33 leads now by the befuddling result of 8.5-5.5!! That calculates to an elo +76 higher than Star Diamond! M4-33 is maintaining its big elo lead (since Game 8 calculation of +89) over 2186 rated SD…presently M4-33 has a performance rating of 2262, higher than Mach4-50!!!!!!!!!!!!!
A09j: Reti: Advance in book through White move 7. M4-33 had nothing internal after 4. b4. This opening involves a pawn sac (4.b4 cxb4 5.a3 bxa3 6.Bxa3), kind of like a reverse Benko! Initially, it looked like this really backfired on Star Diamond as 10. c5? (Bxf8) gives Black a quick -1.3 eval advantage. But M4-33 misses 11. .. Be7, so after 11. .. Rb8?, the position is actually even! Later, M4-33 misses another one, 18. .. Ke7 and after 18. .. Ne7?, White is up +1.2. Looks like the gambit is working! SD gets back the gambit pawn at 24. Nxe5. M4 makes another bad move with that same Knight: 24. .. Ned5? (Kg7) and now the eval is +2.3. That Knight is a real problem as 25. .. Nc3? (Rg8) makes it +3.2. SD wins a second pawn at 28. Rxe6 and a third pawn at 35. Nxd4. The game was virtually over at move 26, but 38. .. Na4? makes sure. Excellent tactical calculations by Star Diamond: from move 12 to move 39, SD calculated 17 SmallFish recommendations or about 60%, which is enough for a win! The bad: the SD finish was forfettable and with Mate # 6 at move 64, Star Diamond needed 10 moves and two Queens to mate: Malodorous.
Game 16 MONOTONOUS and Flaccid. This is the THIRD “B99u: Sicilian”! ALSO, in 8 games as White, Mach 4 has played THREE “C82o: Ruy Lopez: Open” AS WELL!!! Even Game 10 is a repeat of a previously played “A43z: Old Benoni: Schmidt” (Premier game). The meager Mach4 28K opening book library quickly becomes tedious (SD, in contrast has 123K!!!). In fact, I wrote somewhere in a Word doc previously that ALL Fidelity programs, as White, ONLY play 1. e4 or 1. d4.
ONLY:​1. e4 or 1. d4…..THAT’s IT!!! Unsatisfactory.
YET AGAIN! FROM GAME 8:
B99u: Sicilian: Najdorf in book THROUGH WHITE MOVE 20!!!!!! As I recall, the deepest book line my chess computers have played. SD was in internal book until move 19 (as I recall), but M4-33 was already calculating about the last 3 or 4 moves (update: at move; 14 Game 14)! The very first non-book move was SD calculation 20. .. Qa7? (Qd8 is book) and SD is IMMEDIATELY down +1.5! But, 2 moves later, M4 calculates weak 22. Bg2? (h4! is the SM recommendation) and the eval goes to 0.0! And that was pretty much the game! SD moves 26. .. Qg1+ and 27. .. Bg7? (Kg8) give M4 one more chance (eval +1.2), but 28. Qh3? (Qh4) ends that. From that point on the eval advantage plot was basically FLATLINE. At move 40, material was Q+B+3P on both sides and it was a 100% draw, but I let ‘em play.
I was all set to call a Moron’s draw as in the previous 2 games (at 75 in Game 8 and 69 in game 14). Probably even sooner this time, when all of a sudden Moron4-33 rejects previous calculation 51. Qh8+ for 51. Bc2 (Initially, C8 H8 was definitely in the M4 display!). It was 1 AM and I knew I was in for a long dreary night: this changed the whole game. Fidelity calculates (THIS TIME) that it can eventually obtain a passed pawn with sequence 52.Qg4 Kf6 53.e5+ Kxe5 54.Qe4+ Kf6 55.Qxg6+; and so it did. The position (as always) was Q+B+2P, both sides, and M4 has a passed pawn on h3. After SD 59. .. Qb4, Black has a CONSTANT Mate in 1 threat on square b2. I thought the game was ended there as M4 would just perpetual check. BUT, OH NO!! SD foolishly moves the King to the protection of the Queenside pawns (70. .. Kd6?): BIG MISTAKE! 70. .. Kf8 keeping all options open is correct. Long story short:
SD 81. .. Ka5? loses the Black Bishop (on move 87. Qxh8). Now, this tedious POS is a Q+B+P vs. Q+2P endgame study nightmare (and I found nothing on the internet about the outcome or strategies). M4 at move 94 is up +4.7 and actually can win a pawn! Unfortunately, that’s just what M4 does: 94. Bxb5???????????? DRAWN! SD correctly calculates Queen checks until 99. .. Qxb2…and now material is White Q+B vs. Q+P and NO WAY for White to win! Star Dipshit loses its pawn immediately with 100. .. Kb3? Why not Kc3, moron? At 3 AM (THREE AM!) and move 110, I had enough and called a MORON’s DRAW. Just as I was writing up the email title on the iPad, these two f#@!ing imbeciles calculated the following: 110.Kf3 Qh5+ 111.Kg3 Qxe2 1/2-1/2!!!!!!! Now Q vs Q: PROBLEM SOLVED.
Final note: The match has become so repetitive thanks to the Mach 4 woefully inadequate opening library that I have lost a lot of interest in not only the match, but in further Fidelity upgrade purchases. I suppose, Mach 4 playing as Black, still provides some diversity and some entertainment. To a lesser degree, I started to notice this same repetition in the Mach4-50 vs. Atlanta match. At that time, I didn’t get overly concerned about it, but together with this match, it is clear it is a “non-positive” inherent characteristic in the Fidelity units, that is worth mentioning. In contrast, Star Diamond in 8 games as White, has played 8 distinct opening variations: From Ruy Lopez (several lines) to Sicilian (several) to Nimzo-Larsen! to Reti! to King’s Indian, etc. Now that’s an opening book library!
Four.nine wrote:Final note: The match has become so repetitive thanks to the Mach 4 woefully inadequate opening library that I have lost a lot of interest in not only the match, but in further Fidelity upgrade purchases.
In a last ditch effort to perhaps change your mind
again...if you FORCE the Mach 4 to open with a variety of moves(as white or black)then you will see that it has a fairly extensive opening repertoire
I know I mentioned this before and you rejected it out of hand but I think that its really not so important what the computer will play on its own at move 1...
As soon as you issue an all points bulletin to the forum that you fully acknowledge that all table top chess computer hardware technology is fully valid, i. E. "Modded", "enhanced", "now dated technology", "upgraded" units,
then I will issue an all points bulletin to the forum that:
I fully acknowledge that it is acceptable that all table top chess computer software can be FORCED to open with a variety of of moves that prove that an algorithm has a fairly extensive opening repetoire, despite the fact that the algorithm was specifically written to cherry pick advantageous openings to enhance tournament elo performance.
Get the feeling that I have the same aversion to software "manipulation" that you have to hardware "manipulation"?
Maybe we aren't so different?
Best persistent regards,
4.9
Submitted with limited editing...hope it actually makes sense
As soon as you issue an all points bulletin to the forum that you fully acknowledge that all table top chess computer hardware technology is fully valid, i. E. "Modded", "enhanced", "now dated technology", "upgraded" units,
then I will issue an all points bulletin to the forum that:
I fully acknowledge that it is acceptable that all table top chess computer software can be FORCED to open with a variety of of moves that prove that an algorithm has a fairly extensive opening repetoire, despite the fact that the algorithm was specifically written to cherry pick advantageous openings to enhance tournament elo performance.
Get the feeling that I have the same aversion to software "manipulation" that you have to hardware "manipulation"?
Maybe we aren't so different?
Best persistent regards,
4.9
Submitted with limited editing...hope it actually makes sense
Four.nine wrote:Today's installment, games 15 and 16:
Where we at?
I believe 9-7 Mach4-33.
4.9 - thank you sharing your tournament with us. Your commentary is priceless ("Star Dipstick") LMAO
I have a Fidelity 2100 Display with a modded board (SteveUK) that I enjoy playing. Thinking back, I don't remember my Mach 4+ playing openings other than D1 or E1. Unlike you and Steve I am open to some manipulation, both hardware and software, if it makes the overall experience more interesting. In my next tournament I plan to adopt Steve's recommended openings for each unit and hope to see even more interesting games.
I am interested in reading more about your tournament between the Mach4-50 and the Atlanta.
From the very start you have been supportive ......and yet even you are now disrespecting me:
1. It was Star "Dipshit"
2. Table top computer chess (and turbo charging) is not meant to be an enjoyable experience
3. Do NOT in any way, shape or form adopt the heresy of certain non purists of forcing opening moves on poor unsuspecting algorithms
Thank you
4.9
Oh, one more thing:
You have a DD2100 with a moded SteveUK board? Wow, that's really neat! What frequency does it operate at? Isn't it cool??? I have 3. I hope I can get a 68060 at 60 or more MHz at some point.........wait.......
Oh wait a second,
I already forget line item 2!
It's not supposed to be fun.
Concerning #1) If my Mother were alive and she saw me typing the S word I'd be in deep s....stuff. 2) The table top chess computers I have give me quite a beating...not fun but I keep on trying, and 3) I've been abusing algorithms for decades and I don't plan to stop now!
The DD2100 has a 68030 @ 66 mhz and is smoking hot and can reasonably compete with just about any top dedicated chess computer.
How about a summary of the Mach4-50 vs Atlanta tournament?
Why don't I have a 66 MHz????????
I will have to contact that Englander immediately!
Re: 2).....I don't even bother playing my top computers. I would have a better chance in the ring with a young Mike Tyson. Excalibur Grandmaster at default is my upper limit.
Finally, no summary for M4-50 vs Atlanta. Sorry.
But you will have to suffer through a two a day slog like this 33-SD match.
Are you OK with that (and the occasionally f bombs)?
Final note: in reality the 66 MHz is probably only 20 elo points above the 50...maybe the difference between 9.5 instead of 10 points in a 20 game match...sometimes not:
A while ago, I had an Ex. GM match between overclocked 28 MHz and 29.5 MHz units....and the 28 won 10.5-9.5!!!!!!!
AND I ACTUALLY FORCED THE BOOK OPENINGS OF THE SECOND 10 "REVERSE" GAMES TO MIRROR THE FIRST 10!!!!! (Violating my own rule)!!!!
If you do add a 66 mhz to your three current units I'm guessing that would make you the king of Fidelity DD mod's...err improvements. Are your improved boards all DD2100's?
Yes, I am looking forward to the Mach4-50 vs Atlanta match as a two-a-day slog.
Concerning 66 mhz....agreed. I was guessing it might be around 2100, but your estimate sounds about right.
What? Violating your own rule? Hum.....it is good to be flexible.
12B is, if you'll excuse the profanity, overclocked, 4x From 2MHz. Please note that The SSDF figure of 4MHz is the xtal, but NOT what the processor is running at, which is divide by two from xtal or 2 MHz. This machine looks, feels and smells like the original, but calculates in the low 1900's elo as opposed to the, if you'll pardon the expression, stock machine which was about 1750. This machine now battle GK 2000/2100, Excalibur GM and RS Master on equal terms.
Estimates:
I saw in the Aktivschach 12th online tourney that the M4-50 is 2180. Based on that let's approximate:
66 MHz is. 2200
50 is. 2180
33 is. 2150
After my SD ...and Atlanta results I'll give my opinion and why. (Note to self: Fidelity programs have a nasty tendency for " nonsequitor" moves like Kg1-h1 or Kg8-h8 in the middle of an important attack or critical defense)
Flexoble: Yes, in actuality I am not as rigid about the software settings as my bluster would indicate. I've already posted that, indeed, I did exactly what SB recommennded when I forced two machines to duplicate 10'equivalent openings in reverse.
One final note about overclocking, that people should think about and I will cut and paste a game with processor info below: Going up in frequency slightly with the same processor buys you very little, therefore 50 to 66 MHz is virtually negligible. Throughput, that's what you want: you want more elo strength: then you gotta step up from 68000 to 68020 to 68030 to 68040 to 68060 to ARM etc etc...that's what makes programs stronger: more calculations, lots more.
Here's more, a game (TASC 2.2 vs M4-50) from 2016 and processor notes (that didn't xfer to HTML that well. How do you cut and paste a Word to look like a Word doc?)
***********************
Game 108 Tasc makes quick work of overmatched M4-50 (3.5-0.5 against M4)
E12d: Queen’s Indian : Petrosian in book through White move 6. M4-50 didn’t know this line and was already calculating after TASC 4. a3. M4-50 is already on a slippery slope after 12. .. g5? (Re8); TASC up +1.0. M4-50 aggressive 14. .. g4!? (Nbd7) may work against other chess computers, but NOT TASC. TASC correctly responds 15. Ne5! and M4 is down +2.0. BINGO: M4 turd 16… h2+???? GAME OVER. That’s 16 moves!!! TASC proceeds to dismantle the Black position: 20. 20. Nxg6, 23. Nxc6 and 29. Rxg7 give TASC a +7.0 lead! It was just “Clean up on Aisle 7” after that. TASC dominated M4-50 and finishes 6W 2D against the two M4 68030 units. 68030 won’t cut it against TASC, I’ll have to get a 68060 in some way and see if the higher Mips will help:
68030 .33 Mips/MHz. ​11 at 33 MHz. 16.667 (16.5) MIPS at 50. MHz
68040. 1.1 mips/MHz 44 at 40 MHz. 3.33 X 68030
68060. 1.33 mips/MHz 88 at 66 MHz. 8.0 X 68030 @ 33 MHz
A. 68030. 33 MHz. X .33=. 11. Mips
B. 68030. 50 MHz. X .33=. 16.667. Mips Or 1.515 X. A
C. 68040. 50 MHz. X 1.1=. 55 Mips Or 5 x A. 3.3. X B
D. 68060. 60 MHz. X 1.33=. 80 Mips Or 7.27 x A. 4.8. X B
E. 68060. 66 MHz. X 1.33=. 88 Mips Or 8 x A. 5.28 X B. 10% faster than D
F. 68060. 72 MHz. X 1.33=. 96 Mips Or 8.727 x A. 5.76. X B. 20% faster than D
G. 68060. 76 MHz. X 1.33=. 101.33 Mips Or 9.212 x A. 6.08. X B. 26.6% faster than D
SSDF. V11 68060 @72 is ~160 elo higher than V9 68030 @ 32 for ~ 8.7 X throughput
Diminishing returns at higher frequencies. Excalibur Grandmaster doubling (12 to 24 MHz) gets you 140 elo points!
Others have reported ~80 elo for first doubling..............Almeria 68020 is 100 elo from 12 to 24 MHz.
Numbers vary greatly: Vancouver 68020 from 12 to 24 MHz : only 34 on SSDF which sounds very wrong
The V11 68060 72 MHz is 2333 elo on Aktivschach. TASC V2.2 is 2367. A 68060 at 60 MHz (only 16 2/3 % slower) should be no lower than 2300 and give TASC a real battle most games. One more bit of buffoonery on SSDF:
Elite V9 68030 32 MHz ​or 32 x .33= 10.56 Mips​​ is 2161 on SSDF
Elite V10 68040 25 MHz ​or 25 x 1.1 = 27.5 Mips​ ​ is 2162 on SSDF
Today's installment. One game only...and a repeat at that:
Game 17 Fool me twice, shame on me: SD repeats the Game 7 E76r: King’s Indian: Four Pawns loss and ….LOSES; no surprise. The surprise was SD REFUSED to Resign????
From Game 7:
E76r: King’s Indian: Four Pawns in book through White move 11. Queens trade at move 16/17, so I figure a more positional game: wrong. Some non-optimal M4 moves, 19. .. Kh8? (h5), 20. .. Rg8 (maybe h5?) and 22. Nh5? (Nc7) give SD a decent early lead of +1.7. That is, if SD calculates 23. Rxg4. SD calculates 23. Rf2? and: No more lead (+0.4). And M4-33 went on a roll: 23. .. g3!, 25. .. Be5+, 26. .. Rg4! and 27. Rag8. Bearing down on the Star Diamond King on h1 are M4-33: 2 Rooks, 2 Bishops and a Knight; something has to give. SD panics and messes up 29. Nd5? That just loses a Knight, because SD does not recapture, but calculates 30. g3?? GAME OVER. That was quick. When the dust settles, it is move 35 and SD is down the exchange B+N+4P vs. R+N+6P and -5.7 in eval!.
SD varied at move 45. Bc6 (from Ke3). Didn’t matter, SD was totally lost and I was just waiting for “RESIGN”. It never came!!! I RESET unit at the bottom hole and will see if SD gets its mind right, next loss.
M4-33, despite the repetitive White openings drawback, leads 10-7!
Update: ACL button pushed 3/18 and SD “resigned” in a quick loss against SmallFish.
***************
Final note to Dave: M4-33 is up 10-7. Unfortunately your 10-10 prediction, while admirable, is not looking too good.
Stay tuned to this channel.......as tomorrow, all is revealed
The final 3 games.