25W Chess, a Good Idea?
I've been toying with this idea of creating a 25W computer chess tournament with prize money. Computer programs would be limited to running at 25W, to match the human brain energy usage. I hope this would level the playing field and create a fair comparison about the strength of both humans and computers at chess. What thinkest thou?
> [...] it took until 2009 for something (really) interesting to happen, a program called ‘pocket Fritz’ achieved parity with Deep Blue in terms of strength while running on a mobile phone and evaluating only a small fraction of the number of positions evaluated by Deep Blue. In terms of power efficiency that’s an absolutely enormous leap and it brought - in my view at least - for the first time the game on a level footing. Humans and Chess computers were now competing at approximately the same level of energy consumption.
-- https://jacquesmattheij.com/another-way-of-looking-at-lee-se...
> In 2009, a chess engine running on slower hardware, a 528 MHz HTC Touch HD mobile phone, reached the grandmaster level. The mobile phone won a category 6 tournament with a performance rating of 2898. The chess engine Hiarcs 13 runs inside Pocket Fritz 4 on the mobile phone HTC Touch HD. Pocket Fritz 4 won the Copa Mercosur tournament in Buenos Aires, Argentina with 9 wins and 1 draw on 4–14 August 2009.[33] Pocket Fritz 4 searches fewer than 20,000 positions per second.[34] This is in contrast to supercomputers such as Deep Blue that searched 200 million positions per second. Pocket Fritz 4 achieves a higher performance level than Deep Blue.
-- https://en.wikipedia.org/wiki/Human%E2%80%93computer_chess_m...
Droidfish (Android application using Stockfish chess engine) will trounce humans in tournament play on hardware drawing much less than 25 watts. The power restriction might give humans a fighting chance in Go again for a few more years.
Cool idea. Would you also limit the volume of the computer to typical human skull capacity?
It was the exact same idea I had when I saw Alpha Go beating Human Champions. I thought how was that fair? A Machine using ~600kW, while AlphaGo Zero went down to something like 1-2kW. That is still 40 to 80 Times the human brain power consumption. Not to mention our brain requires to process images, smell and other surrounding senses simultaneously. And our 25W includes DRAM, SSD, Storage and all Interconnection. Not sure if that is something Google put into its initial account as it only mention TPU, GPU and CPU.
Although as others have pointed out, Chess seems to be computationally less expensive meaning 25W is fast enough to beat any human player.
It would be interesting if competitors had a high budget, to truly try to develop a great low-powered chess-playing system. Kind of like Bitcoin mining.
Without a high budget, you just have standard hardware, and essentially it's similar to playing on fancier hardware with lower time constraints and pondering turned on.
Amazing idea.
You might have to limit storage as well to stop 'storing energy' on hard disks.
But seeing what people do is part of the fun!
This already occurs currently. Most systems that beat humans already use much less than 25 watts.
With neural network computer engine like lc0, the run-time power is not so important, because it used a lot of compute during its training phase. Also there probably need to be some memory constraints.
Sounds cool! The brain connection is evocative and constraints make for interesting solutions. How would you enforce it?
I would watch if you posted it on YouTube
Yes! Energy usage should be part of AI - no, software! - benchmarks more often anyway
But 25 W for CPU doesn't mean you can't avoid GPU offload and run everything there