If you argue by calling everyone who's much more intelligent than you "pathetic", you aren't going to win friends and influence people. The only people who may applaud you are trolls, who are people of no account.
Better to be a correct douche than a polite moron
Matter of opinion.
Of course... being and douche and clueless while not realizing it - that's the most common case.
Saying somebody's post is pathetic is within the rules.
As for popularity issues among people who are fragile and delicate and cannot take criticism or disagreement and cannot take being blocked ...
Is there anything that matters less?
Let us ask the smartest AI in the world Grok 3 about chess, and being finite.
Is chess an example of something to big to calculate?
Yes, chess is a prime example of something finite but too big to calculate fully, at least in the sense of exhaustively computing all possible games or definitively "solving" it. Let’s break it down to see why.
Chess has a finite number of possible positions and games. The board is 8x8, pieces follow strict rules, and mechanisms like the 50-move rule or threefold repetition ensure games don’t loop forever. Claude Shannon estimated the number of possible chess games at around 10^120 (the Shannon number), a figure derived from approximating 30-35 legal moves per position and an average game length of 40 moves, with adjustments for draws and repetitions. The number of legal positions is smaller—estimated between 10^43 and 10^50 by later studies (., Victor Allis’s work). Both numbers are huge but finite. In theory, you could list every position or game, unlike, say, the infinite decimals of pi.
Now, calculating all that? Impossible in practice. Let’s say you wanted to compute every possible game (10^120). A supercomputer like El Capitan, at 2 exaflops (2 × 10^18 operations per second), could hypothetically evaluate game states crazy fast. If it checked a billion games per second (10^9), you’d need 10^111 seconds. The universe is billion years old ( × 10^17 seconds)—that’s 94 orders of magnitude short. Even if each operation evaluated a full game (wildly optimistic), you’re still at 10^102 seconds. It’s finite, but the scale dwarfs any machine’s lifespan.
What about "solving" chess—finding a perfect strategy from the starting position? That’s a different beast. Chess engines like Stockfish don’t calculate every possibility; they prune the tree with alpha-beta search, evaluating maybe 10^6 to 10^8 nodes for a deep analysis. Solving chess would mean building the full game tree or proving an optimal move for every position. With 10^43 positions, storing them is a non-starter—earth’s atoms number ~10^50; you’d need a planet-sized hard drive just for the data. And computing minimax values for each? Even at a trillion positions per second (10^12), that’s 10^31 seconds—billions of times the universe’s age.
Smaller games prove the point. Checkers, with 10^20 positions, was weakly solved in 2007 after 18 years of computation (using 100+ processors). Chess is exponentially larger. Endgames with 7 pieces or fewer are solved—databases like Lomonosov (10^12 positions) fit on modern drives—but scaling to 32 pieces blows past any feasible limit.
All of that looks correct.
Adding - in addition to El Capitan there's another supercomputer called Frontier.
Who said the morons goal is understanding truth and reality?
Wow we really do have an AI vs. critical thinking problem.
You know, the AIs actually have modules that force them to give you politically correct answers. Answers informed by someone else's political and philosophical worldview.
It's also regurgitating the opinions of masses, who are quite brainwashed and dumb.
You can't possibly rely on AI to answer philosophical questions without reference to your own opinions.
The answer would be the same. As I told crazedrat1000.
It is better for chess engines to run a 64 CPU computer dye. Like mine, then a super computer.
That is why I told him chess programs are no longer run on Super Computers. It just does not make any sense to do so for little to no gain.
It is okay to ask AI philosophical questions.
How one interprets or applies the answers is a different issue.
There's also - how the questions are asked.
The answer would be the same. As I told crazedrat1000.
It is better for chess engines to run a 64 CPU computer dye. Like mine, then a super computer.
That is why I told him chess programs are no longer run on Super Computers. It just does not make any sense to do so for little to no gain.
Given the amount of time / effort I've put into explaining how computers actually work to you, and how little you've written in response or even acknowledged while repeating "Amdahl's law!" / claiming your laptop is stronger than El Capitan - I don't think you deserve any more of an explanation from me.
It is better for chess engines to run a 64 CPU computer dye. Like mine, then a super computer.
El Capitan, the world's fastest supercomputer, has 11,136 compute nodes
Each node of the El Capitan supercomputer contains 96 CPU cores, with 84 dedicated to user processes and 12 reserved for system and Lustre processes.
Here's a more detailed breakdown: MI300A APUs: Each El Capitan node has four AMD MI300A Accelerated Processing Units (APUs).
CPU Cores per APU: Each MI300A APU integrates 24 Zen 4 processor cores.
Total Cores per Node: With four APUs per node, a node has a total of 96 CPU cores (4 APUs * 24 cores/APU).
User Cores: 84 of these cores are available for user processes, while 12 are reserved for system and Lustre processes.
Not all CPUs or cores are created equal, btw.
Actually, in the context where he was mentioning Amdahl's law it wasn't relevant whatsoever, and I already explained why.
Your laptop is not faster at running a chess engine than a supercomputer, period the end.
You claimed earlier your laptop was getting 200,000 UCI nodes per second, right?
El Capital has over 1.056 million cores. It has 5x the number of CPU cores as you are getting UCI nodes per second.
i.e. if every CPU core in El Capital could merely manage to process 1 single UCI node every 4 seconds it would outstrip your laptop.
Yes, ideally chess would be solved using a grid of GPUs, that's also something I discussed earlier which you seem still unaware I ever mentioned.. you do not read much less understand our conversations.
From Grok:
"El Capitan surpasses Frontier as the world’s fastest supercomputer, boasting higher performance (1.742 vs. 1.353 exaFLOPS), a newer AMD MI300A APU architecture, and a greater core count, though it consumes more power (30 MW vs. 21 MW). Frontier, while slightly more energy-efficient in raw terms, has been outpaced due to its older MI250X GPUs and earlier design.
El Capitan is tailored for national security, particularly nuclear simulations, whereas Frontier supports a wider range of scientific endeavors. Both represent cutting-edge achievements by HPE and AMD, but El Capitan currently holds the crown in raw computational power as of March 2025."
---------------------------
Apparently Capitan passed Frontier in November last year.
"El Capitan's speed of 1.742 exaFLOPS is equivalent to 1.742 million trillion FLOPS"
That speed doesn't compare well to the number of possible positions.
Which is about 50 million trillion trillion trillion positions.
"El Capitan's speed of 1.742 exaFLOPS is equivalent to 1.742 million trillion FLOPS"
That speed doesn't compare well to the number of possible positions.
Which is about 50 million trillion trillion trillion positions.
It doesn't, correct, but nothing in existence does and I have never suggested otherwise. However, it does dwarf Dubrovnik's laptop, doesn't it?
Actually, in the context where he was mentioning Amdahl's law it wasn't relevant whatsoever, and I already explained why.
But you are just wrong....
Amdahl's law is relevant, and so is latency. As Grok 3 the smartest AI in the world explained.
And if the AI happens to mention Amdahl's law - or Euler - as an aside that does no harm either.
Neither does members doing so.
Since everyone here likes ChatGPT, and Copilot so much. I thought it would be fun to ask Elon Musk's new AI Grok. The question to this thread. Just for fun.
Here is Grok's answer.
Chess cannot be "solved" in the strict sense—like how tic-tac-toe or checkers have been—because of its immense complexity and the sheer number of possible game states. To solve a game means to determine the perfect strategy for both players, leading to a guaranteed outcome (win, loss, or draw) from the starting position, assuming perfect play. Here’s why chess resists this:
Enormous Number of Positions: The number of possible chess positions is estimated to be around 10^43 to 10^50. Even with simplifications, like accounting for legal positions only, it’s still vastly beyond current computational power. For comparison, tic-tac-toe has fewer than 10^5 possible games, and checkers, solved in 2007, has about 10^20. Chess’s scale dwarfs these.
Branching Factor: Each move in chess offers an average of about 35 legal options (the branching factor). A typical game might last 40 moves, so the number of possible game paths grows exponentially—roughly 35^40, or 10^60. Even pruning techniques used by chess engines (like alpha-beta search) can’t reduce this enough to exhaustively analyze every outcome.
Computational Limits: Modern supercomputers and AI, like Stockfish or AlphaZero, can evaluate billions of positions per second and play at superhuman levels. But "solving" requires exploring all positions, not just finding strong moves. Even if we built a computer the size of the universe, it would take longer than the age of the universe to crunch through chess’s game tree.
Unknown Optimal Outcome: We don’t even know if chess, with perfect play, is a win for White (who moves first), a win for Black, or a draw. Many experts suspect it’s a draw due to the balance of the game and the prevalence of draws at the highest levels, but there’s no proof. Solving chess would mean proving this definitively, which remains out of reach.
Contrast with Simpler Games: Tic-tac-toe is trivial to solve because its game tree is small. Checkers took decades but was solved by exhaustively mapping all 500 billion billion positions. Chess’s complexity is orders of magnitude higher, making it intractable with current or foreseeable technology.
That said, weaker forms of "solving" exist. Endgames with few pieces (up to 7 or 8) have been fully solved with tablebases—databases showing the perfect move for every position. But scaling this to the full game, starting from move one, is a different beast entirely. Chess remains a beautiful mystery, too vast for us—or our machines—to crack completely.
Pretty crappy answer, but not much crappier, if at all, than the average contents of the thread to date.
If the answer is crap. Then I guess it will be pretty simple to disprove the reasons it gave for saying.... Chess cannot be "solved" ...
Grok 3 the world's smartest AI.
Enormous Number of Positions:
Branching Factor
Computational Limits:
Unknown Optimal Outcome
Contrast with Simpler Games
Yes I will, but first say exactly what question you asked it. The title to the thread is a statement not a question.
A position has been advanced that Stockfish doesn't run much better on a supercomputer than on a laptop.
And the poster gave reasons.
I didn't see anything refuting that. Or not yet. Maybe its hard to refute.
hello
I told you I asked it the question to the thread.
Will chess every be solved.?
Which is not the title of the thread as I pointed out, but thank you.
What happened to ibrust? Wait. I get it. He changed his name.
He's here. Posting here. New name begins with c.