Category Archives: Chess Engines

When the Tesla Roadster was revealed, it was quite a surprise. People thought the Tesla Semi reveal was over and that everyone was going to head home soon. Suddenly, the lights changed and the prototype came out of the back of a Semi like the DeLorean in Back To The Future came out of Doc Browns van.

It turned out that Tesla had not only the showmanship for the reveal down, but a vehicle built to back it. The fastest acceleration times, great top speeds, and over 600 miles of range (obviously not at top speed, of course) made for a big change in the supercar world. The companies making cars that sell for millions found much of their performance looking weak next to a car that would sell for far, far less. It indeed could give a hardcore smackdown to gas-powered cars.

If we compare the Roadster reveal to the destruction of the Death Star in Star Wars Episode IV, what were seeing now is Act II of the play, where the baddies strike back (like, you know, The Empire Strikes Back). Thats the part of the play where the heroes often end up in a bind again. Obviously, ICE cars dont have The Force (in the auto world, we call this massive low-end torque) that Tesla does in terms of raw acceleration, which puts them at a big disadvantage, but they still have some tricks up their sleeve.

Lamborghini says they arent even going to focus on getting the best 0-60 times, and will instead get a competitive advantage through better handling. As I pointed out in this article, this is a real threat to any dominance Tesla may gain with the 2G Roadster. Sure, the Roadster will beat almost anything the supercar makers can throw at the drag strip, but on other types of tracks, handling will matter a lot more.

While we dont yet know the curb weight, it seems very unlikely that it will weigh less than 3000 pounds. The battery alone will be around 2000, and with everything else the vehicle needs, 35004500 pounds is a reasonable range.

Contrast this with cars like the Gordon Murray T50. Not only will the car weigh around half what the 2G Roadster will, but it will also have an active fan system that suctions the vehicle down to the road like the legendary Chaparral 2J. With this weight advantage, plus technology that was banned from most forms of racing because the vehicle was too good, the advantages of more modern engine technology is going to give that car a shocking handling advantage over the 2G roadster.

This really shows us where the supercar industry is going. Everyone is making their vehicles lighter, with better handling, and more top-end power than ever before.

In other words, they cant beat the Roadster on acceleration, so theyre stepping up their game in places where the Roadster is currently the weakest. But will that last?

In response to all these recent supercar developments, Mercedes-Benzs performance division, AMG, is going to do something truly wild. Theyre basically taking a Formula 1 racer, Making the minimum changes it takes to make it road legal, changing it a bit more to make it street drivable, and selling it. Theyre calling this beast the Mercedes-AMG ONE.

Image by Mercedes-Benz AMG.

The plug-in hybrid car has a modified F1 engine, plus 4 electric motors. Three of the cars motors drive the wheels directly, with two in the front (one for each wheel), and one in the rear tied to the gas engine to drive the rear wheels. The fourth electric motor is part of a turbocharger system, and can either use electricity to push more air into the engine or generate electricity from exhaust gases. While these motors would serve in a support role during racing, the vehicle can drive for 16 miles on electric only.

The gas engine itself is about as wild as they come. Its displacement is only 1.6 liters, spread out over six cylinders. It wont rev up to 11,000 RPM like a Formula 1 car, because they couldnt sell a vehicle with an engine that only lasts a couple thousand miles, but it will still have impressive abilities for the track.

Final figures arent out yet, but expect superior handling and loads of top-end power for the straight sections of the track.

At this point, the reasonable readers are wondering who would win, while the Stans are 100% sure the Roadster would still beat it. As it turns out, youre both right.

At a drag strip? Tesla all the way. The F1 engine and limited electric power wont propel the AMG ONE from 060 mph or through a straight-line quarter mile the way Teslas monster EV system will. For racing without sharp turns, the Tesla will be hard to beat with anything but other upcoming electric supercars (and those are coming, too).

On a track with lots of curves, though, the ONE will either run very competitively or be the one doing the smackdowns. Its all going to come down to the layout of the track.

On something like the Nordschleife, expect the Tesla to initially pull ahead because of its superior low-end acceleration. Once they get to the turns, though, the Roadster will lose ground and start to fall behind. A vehicle thats 1,000 to 1,500 pounds lighter and still has the benefits of electric torque vectoring just wont be beat by a heavy EV here unless they completely botch the suspension design. Given that its a modified F1 car, expect this to not be the case, and expect it to romp all over in the curvy sections.

What happens in the longer straightaways is still an open question that will decide the race first. Tesla says the Roadsters top speed is over 250 mph, and AMG says its car will go more than 217 mph. Those are both open-ended figures, and not actual top-speed figures, so its possible that either car will have a greater top speed. Either way, wed see the Tesla take back ground in the straights initially, at least until the F1 engine can get to screaming and possibly give the AMG the advantage.

The Tesla will have to start braking before the other supercars, though, when the straightaway ends. Thats going to take away part of the gains it made (if any) in the straight sections.

The above comparison also leaves out any potential magic from SpaceX thrusters.

Bottom line here: we dont know who will win here.

For the Mercedes-AMG ONE, there are still going to be some major downsides, though.

The biggest one is the price, which will approach $3 million. Thats 1015 the price of the Roadster, so that extra little edge that could beat the Roadster on some tracks comes at quite a price. On top of that, the engine will only be designed to run for 30,000 miles. This is impressive, because the real F1 race engines only last 2,000 miles at most, but thats going to add to the expense of owning one.

Expect some owners to try to do a lot of casual driving on electric so they dont wear out the racing engine.

The price advantage probably wont hold in the long run, though. The Gordon Murray T50 will be almost a million bucks cheaper than the Mercedes-AMG ONE, and we can expect supercar technology to slowly trickle down to the cheaper sports cars.

One likely contender that could give the Roadster a challenge for a similar price is the upcoming Corvette E-Ray. Specs are all rumor at this point, but some expect to see a hot turbocharged version of the cars mid-mounted V8 gas engine that produces 900 horsepower alone. Add two front electric motors, and youd have a machine that will make buyers think about their options.

Another indicator that hybrid supercars will continue to compete comes from Honda. The NSX has been out a few years, and many people were initially disappointed with its performance. When people started giving them upgrades and tunes, though, the NSXs got to the point where they werent far behind a Raven Tesla on the drag strip. We can expect to see more affordable sporty hybrids come out in the same price range as the Roadster with even better performance.

Nissan is also stepping up its attack specifically the GTR N Attack package. The price of the GTR plus all of the N Attack upgrades would give you a vehicle thats capable of approaching 7 minutes at the ring for a price thats close to that of the Roadster. Thats a little faster than the modified Plaid Model S Tesla ran on the same track.

If the emerging competition is like The Empire Strikes Back, the second act where the heroes are in peril, then something else we recently learned is going to be the equivalent of The Return of the Jedi. It turns out that a trick Elon Musk had up his sleeve wasnt just a good card, but all four aces.

As my colleague Johnna pointed out, we recently got more information about the upcoming Roadster SpaceX package. We already knew that they intended to offer one with cold gas thrusters that shoot compressed air to push the car, but what we didnt know was that it could achieve 0-60 mph in just over 1 second with that option. Thats an insane amount of acceleration, and can happen independently of speed because the thrusters are not coupled to the drivetrain.

Sure, 060 mph times arent a help against the competition, which is now focusing on the Roadsters weak spot, but it seems likely that there will be more thrusters that allow for lateral force that could cancel out inertia in corners (even if the occupants would feel queasy afterward). Plus, getting a quick burst of power when exiting turns, or a quick burst of acceleration to execute a pass, will likely more than level the playing field.

Add the possibility of using the thrusters for downforce and/or short hops, and you take the 2D racetrack and play 3D chess. Will tracks/events allow you to literally leapfrog the competition? Even if they dont, its still really intimidating to know that the vehicle could do that and that its only the rules thats saving you from utter defeat.

Im not sure what the competition could pull out to fight that, but I doubt theyll just call it quits. Its great that Tesla is still doing things to push the industry forward into new territory and keep competition strong well into the future.

Featured image by Tesla.

Here is the original post:

The Supercar/Hypercar Wars Continue, But The Roadster Still Has An Ace Up Its Sleeve - CleanTechnica

Opening Developments in Yekaterinburg

One of the most interesting aspects of a tournament at the highest level is the plethora of new and creative opening ideas displayed by each player. Each of the eight candidates spent months preparing their repertoires and created specific ideas for one another, while not being afraid to show their best analytical work.

The first part of my reflection focusses on opening ideas from the first half of the tournament (Rounds 1-7), especially ones by Giri and Caruana because these two unleashed several remarkable ideas in their games. These ideas have contributed significantly to the development of theory in the Catalan, Classical Slav, and 6Bc5 Ruy Lopez in particular. As we will see in our analysis of the games below, some of these ideas are strong enough to completely overwrite past conclusions from opening theory.

Lighting up the Symmetrical English

Right from the start of the tournament against Nepomniachtchi, Giri made it clear how much effort had been invested into his Candidates preparation. His concept began with a new idea on move 12 and the ensuing complications were filled with double-edged positions and exchange sacrifices. Nepomniachtchi demonstrated great resilience by heading straight for the critical line of Giris preparation and then successfully calculated his way out of danger. Suddenly the initiative passed to Nepomniachtchi, who eventually won a difficult endgame which was on the verge of becoming a fortress.

Anish Giri vs Ian Nepomniachtchi, Round 3 | Photo: Lennart Ootes

Caruanas Double Pawn Sacrifice

Caruanas repertoire with the Black pieces (Classical Slav and 6Bc5 against the Ruy Lopez) would have been very difficult to predict before the tournament began. In his Round 3 game against Ding Liren, Caruana not only surprised his opponent by playing the Classical Slav, but also unleashed a shocking novelty on move 9 based on a double pawn sacrifice. Similarly to the game between Giri and Nepomniachtchi, Ding responded by heading for one of the most critical lines in Caruanas preparation and successfully calculated his way out of danger. Unfortunately for Caruana, he mixed up the move order in his preparation and ended up with insufficient compensation for the two pawns after Ding found a way to consolidate.

The result doesnt diminish the impact of Caruanas new idea, and I am especially interested in seeing how the theory develops in this line after 10.e4, where the resulting positions are extremely messy and full of sharp ideas.

Ding Liren | Photo: Lennart Ootes

Giris h5-h4 in the Catalan

The Catalan is a staple in the repertoire of many elite players, especially Ding Liren, who plays closed openings almost exclusively with the White pieces. Facing Ding Lirens Catalan in Round 4, Giri employed a rare system with 4a5!? and reached a position that looks favourable for White because of Blacks hanging pawns. However, Giri had prepared some very deep ideas involving h5-h4 which not only make this system fully playable, but also give Black excellent chances to fight for the initiative in some lines (see Vidit Giri in the annotations below).

Considering that Giri had enough confidence in the 4a5 system to employ it against one of the worlds leading experts on the Catalan, I would not be surprised if it eventually becomes as mainstream as some of the other, more established ways of meeting the Catalan.

Anish Giri fighting the Catalan | Photo: Lennart Ootes

AlphaZero-Inspired Play Against the Grunfeld

I should mention that each of the ideas discussed in this article were certainly analyzed extensively by the players and their neural network engines (Leela Chess Zero and recent versions of Stockfish, among others). However, this game in particular contains clear parallels to AlphaZeros famous h-pawn advance in the Grunfeld. Caruana plays the Classical Exchange Variation with 7.Bc4 against Nepomniachtchis Grunfeld and goes for a modern idea involving h4-h5 to put pressure on Blacks kingside dark squares. Nepomniachtchi reacts well and finds a way to simplify the game, leading to a slightly worse endgame which he holds comfortably.

New Ideas in the 6Bc5 Ruy Lopez

Caruana defended one of the most critical lines (10.a5) in the 6Bc5 Ruy Lopez twice in the first half of the Candidates. In both games, Caruana showed deep knowledge of the arising positions and had several new ideas in store. Especially in the game we are about to examine, Caruanas 12th move likely came as a complete shock to Grischuk because this move was previously thought to be tactically flawed. In fact, exploiting the tactical flaw could easily backfire on White as Black seizes a powerful initiative and obtains dangerous kingside attacking chances (see the note to Whites 13th move).

Alexander Grischuk and Fabiano Caruana analyse their game | Photo: Lennart Ootes

Nepomniachtchis French Defence

We have already discussed some surprising changes in Caruanas Black repertoire for the Candidates, but arguably the most shocking repertoire change in the event was Nepomniachtchis decision to deviate from his usual Najdorf and play the Winawer French in Rounds 3 and 7. In an extremely important Round 7 matchup between MVL (3.5/6 points) and Nepomniachtchi (4.5/6 points), MVL went for a relatively new and critical setup against the Winawer and obtained a very comfortable position immediately out of the opening.

This was the only game in the Candidates where Nepomniachtchis calculation abilities and resilience in defence were not enough to survive against deep, targeted preparation and strong middlegame play by his opponent.

The eventual tournament winner Ian Nepomniachtchi | Photo: Lennart Ootes

Pros and Cons of Deep Preparation

One of the key themes seen throughout this article was the use of deep, original preparation. Perhaps the most attractive reason to employ such preparation is that the opponent will be forced to think over the board while the prepared player can rely on remembering and understanding their home analysis. This leads to an advantage on the clock as well as a psychological advantage because the player on the receiving end of the preparation may be concerned with traps set by the opponent. For example, Grischuk remarked after his game with Caruana that it was very unpleasant to play half of the game against a computer (i.e., Caruanas preparation).

It may be the case that certain players react better to their opponents deep preparation. In the first two games of the article, we saw how Nepomniachtchi and Ding Liren both managed to calculate their way out of opening problems, seize the initiative, and eventually take the full point over Giri and Caruana respectively. Especially in the case of Nepomniachtchi, who usually employs a double-edged and narrow repertoire with the Black pieces (Grunfeld and Winawer or Najdorf), this ability to calmly calculate when faced with the opponents deep preparation is extremely important.

As a further warning of relying heavily on deep preparation, we saw in the Ding Liren Caruana game that it is impossible to memorize every detail (17Ng6 instead of 17Rc8!) and that the engine can sometimes provide a false sense of knowledge and security. In reality, the position may be just as dangerous (or even more so) for the side employing the preparation, but this is only realized over the board.

Clearly deep preparation will remain an important component of top-level chess, but I think that it is useful for players of all strengths to recognize its limitations and understand the risks involved for both sides, so they can be addressed and diminished before it is too late during a game.

The first game in my second article will be another piece of deep preparation by Caruana, which eventually brings him a very important victory.

The second part of this 2020-2021 Candidates reflection will focus on opening ideas from the second half of the tournament (Rounds 8-14).

See the original post here:

Reflecting on the 2020-21 Candidates Part 1 - Chessbase News

Artificial intelligence has been all over headlines for nearly a decade, as systems have made quick progress in long-standing AI challenges like image recognition, natural language processing, and games. Tech companies have sown machine learning algorithms into search and recommendation engines and facial recognition systems, and OpenAIs GPT-3 and DeepMinds AlphaFold promise even more practical applications, from writing to coding to scientific discoveries.

Indeed, were in the midst of an AI spring, with investment in the technology burgeoning and an overriding sentiment of optimism and possibility towards what it can accomplish and when.

This time may feel different than previous AI springs due to the aforementioned practical applications and the proliferation of narrow AI into technologies many of us use every daylike our smartphones, TVs, cars, and vacuum cleaners, to name just a few. But its also possible that were riding a wave of short-term progress in AI that will soon become part of the ebb and flow in advancement, funding, and sentiment that has characterized the field since its founding in 1956.

AI has fallen short of many predictions made over the last few decades; 2020, for example, was heralded by many as the year self-driving cars would start filling up roads, seamlessly ferrying passengers around as they sat back and enjoyed the ride. But the problem has been more difficult than anticipated, and instead of hordes of robot taxis, the most advanced projects remain in trials. Meanwhile, some in the field believe the dominant form of AIa kind of machine learning based on neural networksmay soon run out of steam absent a series of crucial breakthroughs.

In a paper titled Why AI Is Harder Than We Think, published last week on the arXiv preprint server, Melanie Mitchell, a computer science professor at Portland State University currently at the Santa Fe Institute, argues that AI is stuck in an ebb-and-flow cycle largely because we dont yet truly understand the nature and complexity of human intelligence. Mitchell breaks this overarching point down into four common misconceptions around AI, and discusses what they mean for the future of the field.

Impressive new achievements by AI are often accompanied by an assumption that these same achievements are getting us closer to reaching human-level machine intelligence. But not only, as Mitchell points out, are narrow and general intelligence as different as climbing a tree versus landing on the moon, but even narrow intelligence is still largely reliant on an abundance of task-specific data and human-facilitated training.

Take GPT-3, which some cited as having surpassed narrow intelligence: the algorithm was trained to write text, but learned to translate, write code, autocomplete images, and do math, among other tasks. But although GPT-3s capabilities turned out to be more extensive than its creators may have intended, all of its skills are still within the domain in which it was trained: that is, languagespoken, written, and programming.

Becoming adept at a non-language-related skill with no training would signal general intelligence, but this wasnt the case with GPT-3, nor has it been the case with any other recently-developed AI: they remain narrow in nature and, while significant in themselves, shouldnt be conflated with steps toward the thorough understanding of the world required for general intelligence.

Is AI smarter than a four-year-old? In most senses, the answer is no, and thats because skills and tasks that we perceive as being easy are in fact much more complex than we give them credit for, as Moravecs Paradox notes.

Four-year-olds are pretty good at figuring out cause and effect relationships based on their interactions with the world around them. If, for example, they touch a pot on the stove and burn a finger, theyll understand that the burn was caused by the pot being hot, not by it being round or silver. To humans this is basic common sense, but algorithms have a hard time making causal inferences, especially without a large dataset or in a different context than the one they were trained in.

The perceptions and choices that take place at a subconscious level in humans sit on a lifetimes worth of experience and learning, even at such an elementary level as touching hot things will burn you. Because we reach a point where this sort of knowledge is reflexive, not even requiring conscious thought, we see it as easy, but its quite the opposite. AI is harder than we think, Mitchell writes, because we are largely unconscious of the complexity of our own thought processes.

Humans have a tendency to anthropomorphize non-human things, from animals to inanimate objects to robots and computers. In so doing, we use the same words wed use to discuss human activities or intelligenceexcept these words dont quite fit the context, and in fact can muddle our own understanding of AI. Mitchell uses the term wishful mnemonics, coined by a computer scientist in the 1970s. Words like reading, understanding, and thinking are used to describe and evaluate AI, but these words dont give us an accurate depiction of how the AI is functioning or progressing.

Even learning is a misnomer, Mitchell says, because if a machine truly learned a new skill, it would be able to apply that skill in different settings; finding correlations in datasets and using the patterns identified to make predictions or meet other benchmarks is something, but its not learning in the way that humans learn.

So why all the fuss over words, if theyre all we have and theyre getting the gist across? Well, Mitchell says, this inaccurate language can not only mislead the public and the media, but can influence the way AI researchers think about their systems and carry out their work.

Mitchells final point is that human intelligence is not contained solely in the brain, but requires a physical body.

This seems self-explanatory; we use our senses to absorb and process information, and we interact with and move through the world in our bodies. Yet the prevailing emphasis in AI research is on the brain: understanding it, replicating various aspects of its form or function, and making AI more like it.

If intelligence lived just in the brain, wed be able to move closer to reaching human-level AI by, say, building a neural network with the same number of parameters as the brain has synaptic connections, thereby duplicating the brains computing capacity.

Drawing this sort of parallel may apply in cases where intelligence refers to operating by a set of rules to work towards a defined goalsuch as winning a game of chess or modeling the way proteins fold, both of which computers can already do quite well. But other types of intelligence are far more shaped by and subject to emotion, bias, and individual experience.

Going back to the GPT-3 example: the algorithm produces subjective intelligence (its own writing) using a set of rules and parameters it created with a huge dataset of pre-existing subjective intelligence (writing by humans). GPT-3 is hailed as being creative, but its writing relies on associations it drew between words and phrases in human writingwhich is replete with biases, emotion, pre-existing knowledge, common sense, and the writers unique experience of the world, all experienced through the body.

Mitchell argues that the non-rational, subjective aspects of the way humans think and operate arent a hindrance to our intelligence, but are in fact its bedrock and enabler. Leading artificial general intelligence expert Ben Goertzel similarly advocates for whole-organism architecture, writing, Humans are bodies as much as minds, and so achieving human-like AGI will require embedding AI systems in physical systems capable of interacting with the everyday human world in nuanced ways.

These misconceptions leave little doubt as to what AI researchers and developers shouldnt do. Whats less clear is how to move forward. We must start, Mitchell says, with a better understanding of intelligenceno small or straightforward task. One good place AI researchers can look, though, is in other disciplines of science that study intelligence.

Why are we so intent on creating an artificial version of human intelligence, anyway? It has evolved over millions of years and is hugely complex and intricate, yet still rife with its own shortcomings too. Perhaps the answer is that were not trying to build an artificial brain thats as good as ours; were trying to build one thats better, and that will help us solve currently unsolvable problems.

Human evolution took place over the course of about six million years. Meanwhile, its been 65 years since AI became a field of study, and its writing human-like text, making fake faces, holding its own in debates, making medical diagnoses, and more. Though theres much left to learn, it seems AI is progressing pretty well in the grand scheme of thingsand the next step in taking it further is deepening our understanding of our own minds.

Image Credit:Rene Bhmer on Unsplash

Read more:

AI Is Harder Than We Think: 4 Key Fallacies in AI Research - Singularity Hub

[Note that Jon Speelman also looks at the content of the article in video format, here embedded at the end of the article.]

The 2020-21 Candidates tournament was, presumably, the most protracted over-the-board tournament in chess history (though correspondence events have gone on for many years, and there must also have been longer league seasons).

Learning from the World Champions

With famous classical examples from the works of the giants, the author talks in detail about principles of chess and methods of play that we can use during every stage of the game.

Now that its finally over, we have almost seven months until IanNepomniachtchi begins his challenge against Magnus Carlsen,supposedly starting in Dubai on November 24th. And with my face relatively egg-free following the last column about the Candidates I thought Id have another go at prediction or at least discussion today.

Carlsen and Nepo have been playing each other for over 18 years now since they first met in under-12 tournaments in 2002. Nepo won both the European and World under-12s that year, beating Carlsen in the European event in Spain and drawing with him in Greece. A year later, in the world under-14s neither won that honour went to Sergei Zhigalko but Nepo did beat Carlsen in their individual game.

Nepo also beat Carlsen the first time they played as adults in Wijk aan Zee 2011 and, overall, I believe that he leads Carlsen by 4-1 with 6 draws at classical chess,though when you include rapidplay and blitz, much of it over the internet in 2020 and 2021, Carlsen has the edgewith +21 =37 14 over 72 games.Of course the exact numbers arent that important and two of Nepos wins at classical chess were when they were kids, but the trend isdefinitely interesting: Nepo has more than matched Magnus at classical time limits but lost out as things get quicker.

Following his confirmation as challenger after the penultimate round in Yekaterinburg, Nepo gave a bilingual interview (in English and Russian he talkedin English except for answering Russian questions in Russian). Understandably, he didnt say very much, though when asked if hed like to send Magnus a message, he pointed out that his actions already had done so. He was, of course, completely frazzled after expending such an effort and I, for one, was completely unsurprised that he got splatted by Ding Liren in the final round: not only had he presumably celebrated to some extent the previous evening, but more importantly he must have been totally out of balance adrenalin-wise.

In the press conference after that final round, Nepo looked fairly unobjective about the game itself. Ding was asked what he thought about Nepos chances, and noted that the two (Nepo and Magnus) are very different players and that much will depend on what sort of positions they can impose on each other. Nepo himself rightly discounted the games they played when they were kids, but was fairly bullish about his chances saying that the champion is always the favourite, although he thought that such matches were close to 50-50.

Master Class Vol.6: Anatoly Karpov

On this DVD a team of experts looks closely at the secrets of Karpov's games. In more than 7 hours of video, the authors examine four essential aspects of Karpov's superb play.

Ian Nepomniachtchi with second Vladimir Potkin during the closing ceremony of the Candidates Tournament | Photo: Lennart Ootes / FIDE

Carlsen himself has also been interviewed, of course, and in one (which, I believe, was a few days before the end of the Candidates when it was already clear that Nepo was almost certain to win) he made some very revealing remarks. He had played some training games against Nepo at one stage and when the latter was on form the results were very close, maybe even slightly in Nepos favour,but when the Russiandropped off Carlsen used to slaughter him winning seven or eight games in a row.

World-class players (and indeed all grandmasters) are normally able to maintain a pretty ineffable mask when playing weaker opponents and may even seem not to blunder (or rather seemed not to before the era of the engines). But like everybody else their inner selves are exposed when they come under pressure, and blunders in games between top class players are perfectly normal. Nepos variability points to an emotional man inside, and his greatest task during the world championship will perhaps be not to elevate his top level of play hes already shown that its enough to beat Carlsen fairly often but to avoid too many days when hes off colour. During the interview mentioned, David Howell picked this up from Carlsenand put it very well: Nepo needs to raise his floor.

If he can do so, or at least only drop into the depths very occasionally, then I think he has a real chance against Carlsen. His last games against the champion were only internet rapidplay and blitz in the Carlsen Invitational, but they showcased his mettle and in particular his Karpovian ability to recover from a worse position and then, even against Carlsen, not be relieved but determined to go on further and try to win. Their overall score also points to this trait in both players. In a battle between two pyrotechnic attacking players you might expect White to have a plus score, but in factits 18-17 to Black with 37 draws.

Navigating the Ruy Lopez Vol.1-3

The Ruy Lopez is one of the oldest openings which continues to enjoy high popularity from club level to the absolute world top. In this video series, American super GM Fabiano Caruana, talking to IM Oliver Reeh, presents a complete repertoire for White.

Magnus Carlsen during the rapid tiebreaks of the 2018 World Championship match against Fabiano Caruana | Photo: World Chess

Over the next seven months the two warriors will have to get ready for the showdown, while also maintaining more routine chess life.

The preparation for a world championship is a huge undertaking and Carlsen has a clear advantage in that hes done it before. But the main question is how they cope with the extreme tension during the match itself.Carlsen is surely the favourite, but not, I think, by that much: no more than 60-40, and perhaps just 55-45.

It was only after I assembled this selection of possibly relevant games that I realized that Black had won every single one! Ive annotated a couple and put in some lightish notes at the critical moments in the others.

Select an entry from the list to switch between games

Master Class Vol.8: Magnus Carlsen

Scarcely any world champion has managed to captivate chess lovers to the extent Carlsen has. The enormously talented Norwegian hasn't been systematically trained within the structures of a major chess-playing nation such as Russia, the Ukraine or China.

Read this article:

Raising the floor - Chessbase News

Most games aren't built from scratch. To save from having to reinvent the wheel with every game, developers use engines and tools to give themselves a starting point. But these engines are made in different ways with different priorities, and one developer has catalogued just how different their ideas of "nothing" can be.

The Nothings Suite, by experimental game developer Pippin Barr (creator of chess monstrosities Chesses and Chogue), is an exploration of what happens when you fire up a game engine and immediately export what the editor throws upa collection of the absolute bare minimum of what each tool considers a viable piece of software.

Understandably, The Nothings Suite contains a lot of completely blank screens. But different tools have different ideas of how much of a blank canvas it should leave a budding developer, and many (like Bitsy or Ren'Py) contain a basic project that immediately introduce you to the basics of how the engine works.

It's fun to compare the big 3D engines on this list, Unity and Unreal. The former simply leaves you facing a static skybox, but that it even has a camera to view it with feels notable. In contrast, Unreal is practically extravagant, letting you fly freely around a basic scene with lighting and a solid block of geometry. It's immediately more showy and more impressive than its counterpart, but it's also more bloatedcoming in at a whopping 31MB.

The Nothing Suite opened last week with 14 entries, but has since added four more (including old tools for creating Atari 2600 and DOS games). The entire thing is free, and most will work through your browserthough if you want to be fussy, the "Print And Play" entry might require a printer for its single blank PDF.

This isn't the first time Barr has honed in on an extremely specific facet of game development, either. In 2017, the developer opened v r 3, a gallery cataloguing dozens of different ways to render water in Unityan experiment Barr later returned to in delightfully low-res form in Bitsy museum b r 3.

Here is the original post:

Here's what 'nothing' looks like in 18 different game engines - PC Gamer

All the results from round 10:

Round 11 will take place on Friday, April 23 at 4:00 p.m. local time. Pairings:

The one decisive game of round 10 finished surprisingly quickly and had a massive effect on the fight to become Magnus Carlsens next challenger. Ian Nepomniachtchi needed only 31 moves to defeat Kirill Alekseenko after the latter misplayed the opening. With four rounds to go, Nepo is now a full point ahead of the trio made up by Fabiano Caruana, Anish Giri and Maxime Vachier-Lagrave.

Notably, the leader has faced all three of the chasers with the black pieces in fact, he has already played Giri twice, scoring a win and a draw. On Friday, after the rest day, the Russian will play Caruana; on Saturday, he will have black against Wang; and,after another rest day, he will play white against Vachier-Lagrave i.e. Nepo will get a free day to prepare for the two key encounters remaining.

With so much at stake, we expect the chasers to go all-in against the leader, whilenerves might also play a big role now that Nepo probably feels his fate is in his own hands.

Regarding a potential tie for first, things do not look good for Giri, since the head-to-head score is the first tiebreak criterion, and he has already lost against the leader. MVL, on the other hand, beat Nepomniachtchi in round 7, while Caruana would actually become the favourite to take it all if he defeats the Russian on Friday since they drew their first encounter.

Nothing is yet decided, though. As we have witnessed in the past, strange things can happen when the stakes are so high.

The playing hall in Yekaterinburg | Photo: Lennart Ootes

Very rarely we see a player as strong as Alekseenko getting in trouble strategically as quickly as he did on Wednesday. Nepo played the English Opening and went for a Catalan-like structure. Alekseenko simply could not address the main problem Black has in these positions: the activation of thelight-squared bishop.

By this point, Carlsen was already talking about how Alekseenko needed to show something specific to justify his play so far, since there are lines that have been proven to be much less problematic for black. Moreover, Alekseenko had been spending over three minutes on almost every decision since move 4.

Black did not have a surprise in store. The game continued 8...Be7 (8...Qc7 was called for, planning a quick ...b6 and ...Bb7) 9.0-0 0-0 10.d4 cxd4 11.Nxd4 Qc7 12.Rd1 Rd8 13.Be3 Nb6 14.Rac1

Power Play 24: A repertoire for black against the Catalan

On this DVD Grandmaster Daniel King offers you a repertoire for Black against the Catalan, based around maintaining the rock of a pawn on d5. Keeping central control ultimately gives Black good chances to launch an attack against the enemy king.

This position might seem harmless at first sight, but Black is already in deep trouble with his queen unable to find a safe spot away from potential discovered attacks by the rooks and the minor pieces especially the light-squared bishop awkwardly placed on passive squares.

Alekseenko, already way behind on the clock, opted for the anti-positional 14...e5, which was duly responded by 15.Nf5 Bxf5 16.Qxf5

Black got rid of the passive bishop, but now White has a great initiative. Nepomniachtchis conversion was exemplary, cleanly showcasing the weaknesses of his opponents plan. White gave up his dark-squared bishop for the f6-knight after 16...Nc4 17.Bg5 Rxd1+ 18.Nxd1 Rd8

Grandmaster Karsten Mller analysed the second phase of the game, explaining why Nepo chose to enter an opposite-coloured bishops position. The sole leader wreaked havoc on Alekseenkos position along the light squares.

Things looked bad for Kirill Alekseenko right out of the opening| Photo: Lennart Ootes

In round 9, we mentioned how it was surprising to see Alexander Grischuk getting a considerable edge on the clock. The very next day, however, the time-trouble addict returned to his old habits. Grischuk spent no less than 72 minutes reflecting on how to play his 11th move!

The Shining Sveshnikov Sicilian

Always wanted to play like a World Champion? Search no further! With Magnus Carlsen using the Sveshnikov variation as his weapon of choice in the World Championship match against Fabiano Caruana, this DVD could not be better timed.

Indeed, it is a tense position, which has been seen in games featuring Alexander Morozevich (draw) and Ding Liren (won) playing the black pieces. Grischuk finally went for 11...cxd4 and a number of trades in the centre led to another crucial position. This time around, it was Wang who surprised the commentators with his decision.

The Chinese sacrificed his queen for two pieces and two pawns with 21.exf6 Bxc2 22.fxe7 Rfe8 23.Nf4 Nb6 24.Nxd5.

At this point, both the engines and Grischuk consider that Black had a considerable advantage, but it is not at all simple to find the most precise moves, as White can create threats against the king with hisbishop pair. In the end, an intricate skirmish finished drawn in 41 moves.

Asked about how the 72-minute think had impacted the game, Grischuk responded:

I think it impacted in two ways. One is good for me, asit made my opponent sacrifice a queen probably not very correctly; but, on the other hand, it made me unable to really think after the unexpected 12.Nfxd4.

If you want to see what kind of lines Grischuk was considering, do check out the 18-minute post-game press conference. Classical chess is hard!

Time-trouble addict Alexander Grischuk| Photo: Lennart Ootes

Playing black, Giri again chose the Sveshnikov Variation of the Sicilian he had used to draw Nepomniachtchi in round 8. Vachier-Lagrave played the line with 7.Nd5, which had been explored at the 2018 World Championship match in London. MVL deviated from the most popular continuations on move 10 and put some pressure on his opponent.

The Frenchman eventually gained a pawn.

Master Class Vol.2: Mihail Tal

On this DVD Dorian Rogozenco, Mihail Marin, Oliver Reeh and Karsten Mller present the 8. World Chess Champion in video lessons: his openings, his understanding of chess strategy, his artful endgame play, and finally his immortal combinations.

The d6-pawn looks scary, but it is not easy to push it in a position with opposite-coloured bishops on the board. Giri showed the correct way to defend and a draw was eventually signed.

In the press conference, the players concluded:

Vachier-Lagrave: Its a bit disappointing because I had a very good position, but at the same time there is nothingobvious that I missed.

Giri: Yeah, and you faced a great defence.

Vachier-Lagrave: The draw master is back at it (smiles).

Anish Giri| Photo: Lennart Ootes

Worlds numbers 2 and 3played an interesting line out of a Spanish Opening, with Ding getting things fully under control by move 17.

The Ruy Lopez Breyer Variation

Pavel Eljanov explains in depth what Gyula Breyer already saw in 1911 and what became an opening choice of the likes of Kasparov, Kramnik, Anand or Carlsen. The Breyer Variation, which is characterised by the knight retreat to b8.

Ding later commented that he was proud for having figured out that his queen was well-placed on c8, while here, after 17...Nxa5 18.Qc2, Caruana already thought there was no wayto play for a win with white.

The Chinese star kept putting pressure, but Caruanas excellent calculation abilities allowed him to hold a draw.

The ever candid Ding Liren| Photo: Lennart Ootes

Read more:

Candidates Round 10: Nepomniachtchi widens the gap - Chessbase News

Today's programs are all so strong that they seem to really differ in the details more often than in a decisive statement of strength, and there is no question that when arguing the differences at the stratosphere, it seems almost ludicrous. Engine A is 3568 Elo, while Engine B is inferior because it is only 3565 Elo. So stated by the humans all hovering under 2800 barring a small fistful.

Still, the game made such a powerful impression on Peter Graysonthat he declared,

"Considering the fast time control that was quite amazing by Fat Fritz 2 and its subtlety was of a sophistication I would associate more with the human mind than an engine particularly for the follow up that confirmed the engine can execute a long term strategy.Perhaps the Fritz network does provide a more human rather than mechanical, logical approach?"

The game starts quietly, almost innocuously. An English line that has seen proponents on bothsides at the highest echelons.

Yet by move 12 they had both left most of the known cases behind, with only an Italian correspondence game cited in Mega 2021. The key move that incited so much enthusiasm and which got Black into such a dangerous situation came here:

"On the face of it this looks to be in line with the idea of controlling an open file where the controlling side tends to have an advantage.The following moves question that idea when it is a wing file and also whether it is advisable for the queen to lead on the rank that will likely be the first piece to come under attack. White's reply may not be immediately obvious until it is seen and few other engines find it, certainly not within the context of the game."

"How important this move was to the outcome of the game should not be understated. With Black seeking to gain control of the open a-file, suddenly the queen looks cut off and potentially a liability. That is the theme of the ensuing moves. Perhaps it deserves !!! What is fascinating is that Fat Fritz 2 exhibits almost human-like qualities to threaten the snaring of the queen."

While Black does manage to avoid the outright loss of the queen, it comes at a heavy price that ultimately costs the game.

White now threatens to win the queen in two moves with 30. b4 a4 31. a4. Black avoids this fate by giving up the exchange, but this in itself proves fatal.

Fat Fritz 2

Fat Fritz 2.0 is the successor to the revolutionary Fat Fritz, which was based on the famous AlphaZero algorithms. This new version takes chess analysis to the next level and is a must for players of all skill levels.

Here is the full game with the generous comments by Peter Grayson.

The rest is here:

Trapping the queen - Chessbase News

It is not a surprise that chess players tend to develop a natural affinity for mathematics, purely because imagination and logic play a major role in both disciplines. But unlike some other universally popular games such as poker or scrabble where the mathematics of probability plays a huge role, chess does not deal with chance or incomplete information. Hence in Game Theory, chess is called a Perfect information game because each player, when making a decision, has access to all the information.

The problem with such perfect information games is they nosedive into irrelevance if the game is solved. From Tic-tac-toe, which a schoolboy can solve, to Checkers, which took eighteen years of hard work to solve, so many strategy games have suffered this fate. The bottom line is, in this age of supercomputers with their astonishing processing power, the clock is ticking for most perfect information strategy games. What about chess then?

Why has chess not been solved yet

Chess playing computer engines are way stronger than any human. But still, we are a long way away from knowing the absolute truth in every position. Why? Because of the sheer complexity of the game.

Back in the 19th century it was found that there are more than 300 billion ways that the first four moves of a game could be played. Later, Claude Shannon, the father of information theory, argued that while it is theoretically possible to solve chess, it wont be feasible because it would take 1090 years to evaluate all the relevant positions.

The magic of geometric progression

Wheat placed on the first five squares (photo Wikipedia)

Such gigantic numbers and chess go back a long way if you go by the ancient story about the maharaja and his courtier who supposedly invented the Chaturanga, the earliest form of chess. Legend has it that the maharaja was so pleased with this new game that he offered any reward to his courtier.

The wise man asked for a single grain of wheat for the first square of the chessboard. Then two grains for the second, four for the third, and so on. Doubling each time, until the 64th and final square is reached.

The ruler has apparently laughed it off as a meager prize for such a brilliant invention. Soon to his shock and dismay, he was informed by his treasury that all the wheat grown in the world would not be enough to fulfill the courtiers request. The unbelievable final number, they correctly calculated, is 18,446,744,073,709,551,615 (264-1).

Versions differ as to whether the smart courtier was beheaded or promoted. The ruler, by all accounts, was either very offended or very impressed.

Some Mathematical tasks

As we mentioned earlier, logic and imagination goes hand in hand when it comes to chess and mathematics. Here are a couple of interesting problems to put your skills to the test.

After 16.Nxb1 Find a proof game from the starting position (Philip Lehpamer)

Logic tells us that White must make 15 captures. We also know that white cant capture anything with the first move. Therefore from moves 2-16, white must make fifteen consecutive captures.

Somehow this task reminds one of the game Peg Solitaire (more commonly known in Sri Lanka as Chuck-a-blass thanks to a local tele drama which made the game popular in the 1990s).

The second problem involves symmetry which is an important concept in many branches of mathematics, especially when it comes to geometry. One of Sam the Puzzle King Lloyds best known problems features a unique copycat problem.

Black copies each of whites first three moves. Then white delivers a checkmate with his fourth move. Find the moves.

Loyds solution was 1.d4 d5 2.Qd3 Qd6 3.Qf5 Qf4 4.Qxc8#. A similar solution can be arrived at by playing 3.Qh3 Qh6 as the third move.

Lloyds first solution

Can you find an alternative solution?

Solutions

1.Chuck-a-blass problem

1.Nc3 b5 2.Nxb5 c6 3.Nxa7 Nf6 4.Nxc6 Ne4 5.Nxb8 Nc3 6.Nxd7 Ba6 7.Nxf8 e5 8.Nxh7 Rf8 9.Nxf8 g6 10.Nxg6 Bb5 11.Nxe5 Ra3 12.Nxf7 Qd6 13.Nxd6+ Ke7 14.Nxb5 Nb1 15.Nxa3 Ke8 16.Nxb1.

2.Checkmate the copycat problem

1.c4 c5 2.Qa4 Qa5 3.Qc6 Qc3 4.Qxc8#

Original post:

Dulan Edirisinghe's Chess Corner: Logic and imagination, mathematics and chess - nation.lk - The Nation Newspaper

Sven Magnus Carlsen, the Norwegian chess grandmaster who is the current World Chess Champion, World Rapid Chess Champion, and World Blitz Chess Champion. Carlsen first reached the top of the FIDE world rankings in 2010, and trails only Garry Kasparov in time spent as the highest rated player in the world. His peak classical rating of 2,882 is the highest in history. Carlsen also holds the record for the longest unbeaten run in classical chess

Regular readers of this column will no doubt have noticed that I draw heavily from cricket and chess. The pieces, I hope, have application-value outside of cricket and chess; but anyway, its simply because I played competitive chess and still play a lot of online games and because Im a keen follower of cricket. I am your average sports fan who knows something but not much about most sports and a little more about one or two.

Anyway, I felt I needed to get that out of the way.

The column, then. The example is chess. Again. Bear with me.

Now in chess there are positions where neither of the players have any edge, not even a slight advantage. An equal position, is how it would be described, by players and engines.

Sometimes it is a blocked position with few or no openings for either player to break through into the enemys camp. Sometimes its an ending where the particular complement of pieces simply means that one player controls some squares (say the dark ones) and the other the rest (light squares), typically in an opposite coloured bishop ending.Sometimes its just that both players are short of ideas and imagination.

If neither player has an overwhelming desire to push for victory, then it could end with them agreeing to draw the game following a few cursory moves that neither threaten nor create weaknesses that the other could exploit.

Sometimes, draw or its possibility enters the mind of a player who overly depends on tactics.

If the piece-complement, their relations with one another, and the overall position do not lend towards tactical play, one often notices a kind of resignation.

That is, a resignation to a drawn result. However, as most chess players know, chess is a game of development. Its about striving to improve ones position. Relentlessly. Even in a blocked position, for example, one could decide to develop the least developed pieces. It doesnt necessarily deliver a victory but in the very least will reduce opportunities for the opponent to force a win.

Its all about strategy. In equal positions, many have learned the hard way, the player with a plan tends to win or at least improve chances of winning. Dead drawn positions do not always end in draws. One player might plan, visualise a winning position and strategise ways of getting there. He/she may or may not, but if the other player is complacent, there is always the possibility that he/she might err and provide that half-chance which makes the difference.

On the other hand, if one player banks on tactics and tactics alone, he/she might unconsciously become blind of subtle opportunities that help improve his/her position. From there to complacency is a short distance. Complacency always creates the possibility of error. Error, even if slight, can tip the scales.

Tactics are important, make no mistake. A wise man once said, strategy without tactics is the slowest route to victory; tactics without strategy is the noise before defeat.

In chess, the length of the road is about fatigue and that matters too if its a long tournament or even just a single game afterwards. Long or short, the victory, sweet or otherwise, yields just a single point. That point is important. Hence the emphasis on strategy.

In certain positions and against opponents who drop their guard and become careless, tactics alone can deliver victory, but if one is playing stronger opponents at higher levels of competition, they wont be enough. Often necessary but seldom sufficient are tactics.

And without strategy, its just noise. You win battles or rather you may win a battle or two, but you could very well lose the war. You might come up with some neat tricks but more often than not, the opponent is not caught by surprise or left without opportunities to refute.

In general, in sports like in war, those who carefully plan are more likely to gain ground in the mini battles or sub-plots which, in the end, translate into incremental advantages that add up to give an edge that is decisive.

In the end, the noise does not count. Its the noise one earns the right to make AFTER the game is over that counts. Well, the result is what matters, but bragging rights about consolation prizes secured along the way are in the end of little import.

Strategy. It matters. One has to plan. Tactical opportunities present themselves if strategy is sound. The quote, by the way, is from Sun Tzus The Art of War. He knew about tactics. He knew about strategy. He always had the endgame in his sights. In war. In chess. In all sports. Even in life, one might add.

(The writer can be reached on [emailprotected] or http://www.malindawords.blogspot.com. Malinda Seneviratne is the Director/CEO of the Hector Kobbekaduwa Agrarian Research and Training Institute. These are his personal views.)

Read more:

Tactics without strategy is the noise before defeat Malinda Seneviratne Column - nation.lk - The Nation Newspaper

The Milan Epyc 7003 processors, the third generation of AMDs revitalized server CPUs, is now in the field, and we await the entry of the Ice Lake Xeon SPs from Intel for the next jousting match in the datacenter to begin.

The stakes are high for both companies, who are vying for what seems to be a reasonably elastic demand for compute capacity in aggregate around the world, even if there are eddies where demand slows and chutes where it accelerates. The room enough for Intel and AMD in the market, but it is the technical and economic jousting between these two that is going to make this fun and help spur future competition in the years to come.

We did our announcement day first pass on the Milan SKU stack, with the salient feeds and speeds, and slots and watts, the 19 new processors, and we also covered the actual launch event for the Milan chips by chief executive officer Lisa Su and her launch crew, which included Forrest Norrod, general manager of AMDs Datacenter and Embedded Solutions Group, and Mark Papermaster, the companys chief technical officer, and Dan McNamara, general manager of AMDs server business.

Now it is time to get into the weeds for a little bit and talk about the Milan architecture and how this processors Zen 3 cores are delivering 19 percent higher instructions per clock than the Rome processors Zen 2 cores from August 2019. It is hard to squeeze more and more performance out of a core while maintaining compatibility, but Intel, AMD, IBM, and Arm Holdings are clever engineering companies and they often go back to the drawing board and rethink how the elements of a core are organized and pipelined. They always seem to find new ways to do things better, and it really is a testament to human engineering that this is true.

Someday, we presume, AI will be used to create blocks of logic and data from transistors and place them in a 2D or 3D chip layout and do a better job than people and their EDA tools; we talked about Googles research in this area last year, in fact, and IP block placement in an EDA tool makes the games of Chess and Go look like a joke. So far, people are a necessary part of the process of designing a processor, so today is not that day. Ironically, better compute engines will hasten that day, and perhaps chip designers should not be so eager for such big improvements. . . . But, if history shows anything, you cant stop progress because people just plain have faith in it. For better or worse. Often both. And we here at The Next Platform are no different in this regard, so dont think we are taking some high brow view. Consider us raised eyebrows with the occasional furrowed brows. We admire what engineers do; we worry about what people do with what they create sometimes.

Mike Clark, an AMD Fellow who cut his teeth on the single-core K5 processors, the first in-house designed AMD X86 chip from back in March 1996, and the lead architect on the Zen 3 cores, walked us through the nitty gritty detail of the Zen 3 core that is at the heart of the Milan system on chip complex. Lets dive in.

Right off the bat, this is a whole new, ground up redesign of the core similar to what Intel is doing with Ice Lake Xeon SPs and their Sunny Cove cores and what IBM will be doing with the Power10 processor and its brand new core later this year. And the reason is simple: Everyone needs to push the IPC as hard as possible to boost single threaded performance, and then make tradeoffs in the SKUs between high clock speeds across a small number of cores and lower clock speeds against a larger number of cores to hit performance targets that are better on each class of workloads and those in between than their respective Rome Epyc 7002, Cascade Lake Xeon SP, and Power9 predecessors. You cant just have more cores with a new generation, and you need to show better thermal efficiency at different performance points, too.

So how did AMD get that 19 percent better IPC with the Zen 3 cores used in the Milan server chips? By doing a whole lot of things all at the same time, as you will see. And when you contrast this with the lack of IPC improvements as the Sunny Cove cores are coming years late to market because of Intels delays with its 10 nanometer processes that these Sunny Cove cores and their Ice Lake processors were tied to, it really shows:

Here are some of the top-level performance improvements, says Clark. We improved branch prediction, and not just accuracy, but actually being able to get the correct target address out sooner and the correct target instructions out sooner and feeding them to the machine so we get more throughput, more performance. We have beefed up the width of integer throughput. We have doubled the intake floating point for inference, as we see those workloads evolving going forward and we are reacting to that. And by pulling the eight Zen 3 cores under the larger 32 MB L3 complex, we have better communication paths and we have more cache available for in lighter-threaded workloads and therefore we can reduce the effective latency to memory and provide more performance.

The Zen 3 core still has two-way simultaneous multithreading, as we pointed out in our initial coverage, and AMD has resisted the temptation to add more threads to goose performance as IBM does with its Power architecture, which can dynamically switch from 2, 4, or 8 threads per core. (Sometimes, the threading in the Power9 and Power10 chips is set in firmware and the cores are fat or skinny, depending.)

If you look on the right in the chart above at the Zen 3 block diagram, you can see there are two ways into the machine. The 32 KB instruction cache is still driven by a decoder that can drive four instructions per clock cycle into the op queue. And the way into the chip is through the branch predictor on the far right that can put instructions into the op cache and deliver eight macro ops per cycle. The dispatcher decouples the two sides of the Zen 3 pipeline integer and floating point and can do six macro ops per cycle to either unit.

That front end to the integer and floating point units in the Zen 3 core has a lot of tweaks, starting with a n L1 cache branch target buffer that is twice the size of the one in the Zen 2 core, at 1,024 entries.

Clark says that the branch predictor on the front end has more bandwidth, which means it can pull more branches out per clock cycle. The Zen 3 core also features what Clark calls a no bubble branch prediction mechanism, which he explains thus:

When you pull out a target address from the branch predictor, you then need to obviously put that back into the branch predictor to get the next address. Typically, that turnaround time creates a bubble. We have a unique mechanism where we can eliminate that bubble cycle and therefore be able to continuously pull out branch targets every cycle. We do still get some branches wrong in the execution units, but getting those addresses back and getting the target instructions of the machine we improve the latency of that from Zen 2.

There are also some efficiency improvements in the op caches faster sequencing of fetches and finer-grained switching of op cache pipelines that help that Zen 3 front end drive that 19 percent IPC improvement (which is an average across a bunch of different workloads that have been used to gauge IPC on cores in the Opteron days that is now used on Zen cores in the Epyc era).

So thats the front end the air intake manifold and fuel lines in a car engine analogy, we supposed. What about the integer and floating point cylinders? Here is a zoom into the execution engine in the Zen 3 core:

With the Zen 3 core, there is a much wider integer unit now, with four ALUs and dedicated branch and storage units, as you can see from the chart above comparing and contrasting with the Zen 2 block diagram in the chart further up in this story.

Here is a drill down into the integer execution unit, which Clark says has a design goal of having larger structures to extract more instruction level parallelism (ILP) from applications to feed this part of the execution engine; its units, in general, have lower latency, too. The combined effect is more integer IPC.

As you can see, everything increases by a little bit or a lot, bringing a different set of throughout and balance to the combined set of units.

(We wonder if the engineers play a kind of video game, tweaking this or that in a simulator so see the effects, or if the EDA tools do this work, as well. We suspect the former and that like much design there is a knack for it and as an architecture hardens a bit, you throw it out and start over. This Zen 3 core does not look that different from a Zen 2 core to our eyes certainly not like the jump from Sledgehammer to Bulldozer to Piledriver to Steamroller cores.)

With Zen 3, there are four integer scheduler units instead of seven with Zen 2 (why seven, which so not base 2 and therefore violates our sensibilities?), and the same eight ports come out of the integer register file as with the Zen 2 integer unit. Rather than the schedulers being paired to an arithmetic logic unit (ALU) or an address generation unit (AGU), they are shared, allowing for balanced use across workloads. There are still four ALUs on the integer block, as you can see, but one of them has its own branch unit embedded in it and another one has a store unit embedded in it. Similarly, there are still three AGUs, but one has a store unit embedded in it. And, there is a branch unit pulled out separately.

Its still the same number of ALUs, but they are much more available and have much higher utilization, says Clark. With queue combinations, with shared ALU/AGU schedulers, which we can pick from independently, the pickers can get a better view of more operations to therefore find more instruction level parallelism in the workloads. And by offloading those extra store data and branches, those things dont really return things back to the register file so you dont really have to take the cost of having more write ports into the register file just more read ports.

With the Zen 3 core, the floating point unit is also wider, with six pipelines to be able to accept the input from that six-wide dispatch unit.

The floating point multiply/accumulate and add units have store units pulled out separately now, too. The reorder buffer has been increased in size so the Zen 3 core has a larger window to get more floating point instructions in flight. And, as with the integer units, the floating point to integer conversion units and store units are separated out from the add and multiple/accumulate units so they dont collide or cause backups while still preserving the number of add and multiple/accumulate units compared to Zen 2. The floating point register file is 256 bits wide (same as with the Zen 2, which had a pair of 128-bit registers), and importantly for AI inference workloads, the INT8 bandwidth is twice that of the Zen 2 core, with two IMACs and two ALU pipes. The Zen 3 core can do two 256-but multiply accumulate operations per cycle.

If you are going to chew on more data and instructions, you have to be able to load and store more data and instructions, so the load/store units in the Zen 3 core have also been beefed up:

The Zen 3 core can do three loads per cycle or two stores per cycle, compared to two loads and one store per cycle (tied together, that is an and statement, not an or statement for the Zen 2 core). The load/store units have higher bandwidth per clock and, like the integer and floating point units, have greater flexibility in what they can do at any given time. Which drives up the ILP to get to that higher IPC.

Importantly, the Zen 3 core has six translation lookaside buffer (TLB) walkers, which walk that memory cache, which stores virtual memory addresses for physical memory in the DDR4 DRAM attached to each processor. This increased TLB capability, says Clarke, helps deal with server workloads that have a lot of random accesses to main memory or that have applications that have large memory footprints that span multiple pages of main memory.

And finally, the Zen 3 core has a bunch of instructions that are added, as follows:

Next up, we will be taking a look at the competitive landscape as AMD sees it for the Milan Epyc 7003 processors.

Here is the original post:

Deep Dive Into AMDs Milan Epyc 7003 Architecture - The Next Platform