• 0 Posts
  • 193 Comments
Joined 3 years ago
cake
Cake day: June 18th, 2023

help-circle

  • For those who don’t want to read several pages of unnecessary text telling you what you probably already know:

    The math, while pretty involved, may tell a straightforward story (if you’re interested in the details of our analysis, see the Appendix). OpenAI has contracted 900K memory wafers per month from Samsung and SK Hynix. Partner commentary seems to indicate that’s a monthly number, so that represents 10.8 million wafers over 12 months. In terms of demand, a fully built-out 10GW Stargate cluster would require ~3 million GB200 Bianca Boards. Each board requires ~50% of a memory wafer in total; split between the HBM3e stacks embedded into its two B200 GPU (~30%) and its 480 GB of LPDDR5X system memory (~20%). That puts total wafer demand for the entire cluster at ~3 million wafers.

    Therefore, according to our best estimates, OpenAI likely needs less than 30% of the 10.8 million wafers it’s planning to buy

    So this is just putting some numbers to what a lot of people already guessed. The AI companies are not just buying a ton of RAM to build out their data centers. They aren’t buying enough other components to even use that RAM. They’re buying it so that no one else can.




  • For a the past few years, I had wondered why videogames, movies, and TV shows nowadays feel so… Bland. Meaningless. Soulless. Corporate. Like, I know they ARE corporate, but these industries have all been dominated by gigantic corporations for my entire life. What changed recently? Am I just getting old and curmudgeonly and preferring content that was made back when I was younger?

    Then I was watching DoorMonster talk about some show (I could be wrong, but I think it was the video about how Arcane had a great Season 1 that was largely ruined by Season 2) where they kept joking about not accusing them of using AI to write things.

    Then it clicked. The Writer’s Strike from May-September 2023. On paper, the Writer’s Guild secured restrictions on the use of AI. And I can’t point to anything specific and say “that was clearly written by AI”. But I can say that for the past few years everything put out by pretty much every company has felt very… “Meh”. Nothing new has grabbed me and said “wow I need to watch/play that”. Could be a coincidence, but I also have to wonder whether AI tools involved in writing and visuals have cost us something intangible that I can still feel.


  • This may be unpopular, but I think this is great news.

    Skyrim became one of the best-sellign games of all time in part BECAUSE of how great it is to see your character get ragdolled into the lithosphere by a giant, or to watch the chaos of spawning thousands of wheels of cheese on top of the throat of the world and watching them roll down.

    An Elder Scrolls game that was built around having realistic physics, or being restricted to more cinematic movement and knteractions, would lose a key essence of what made the earlier games great.

    I don’t want engaging combat in Elder Scrolls. If I want combat that I have to pay attention to, I’ll go play a Souls game or a fighting game or one of the thousands of games that have tried to be “Skyrim with better combat” that have languished in obscurity because they miss the point.





  • They also have guidelines for “user generated content” which includes reviews, and you can report people for violating those guidelines.

    Sure Valve does not pay for moderators to check things proactively. I quite like that they don’t have AI or some other half-assed attempt at “moderation” like other platforms have. I hate the way that the whole Internet has moved to censor “fuck” and made up the word “unalive” because the automated systems of platforms I don’t even use have decided they are the arbitora of what language is allowed.

    I think the responsibility to monitor reviews should lie with whoever controls the Steam page: I would assume the publisher most of the time? The publisher and developer should be looking at reviews anyways. Add in the ability for users to vote reviews as helpful or unhelpful and I think it’s one of the better systems left on the internet.



  • BioShock 1 and Infinite both have the same problem.

    On your first time through, the story pulls you through the game. There setting and characters are so mysterious and interesting you’re compelled to figure out what the hell happened and get to the bottom of it. You might notice, on your first run, that the games are really easy and the gun play isn’t particularly good. The actual gameplay gets repetitive, basically moving from big room to big room shooting things.

    The special powers are fun the first couple times you use them but are mostly situational and the kind of thing other games just use items for (land mines, grenades, etc), just re-skinned.

    Then at the end there’s a big reveal. Some plot twist that re-contextualizes the whole game and leaves you thinking about the game for an entire week.

    Then you replay them and realize… The big twist at the end? There’s almost zero foreshadowing and it would be impossible to have predicted either of them on your first playthrough.

    There are plenty of factions that have different political ideologies, but they are nothing more than a setting. The most obvious is how they spent the first half of Infinite pretty clearly establishing that Comstock and his associates were violently oppressing the working class in Colombia and that Daisy Fitzroy’s rebellion was both personally and ideologically justified. Then all of a sudden Booker is there enemy because… He thinks they were too violent in their pursuit to overthrow that oppression or something? It really felt like the devs just needed to throw more enemies at you in the back half of the game so they made a flimsy excuse to do that.

    The BioShock games give the illusion of talking about politics and ideology, but really the only message is just “extremism bad”.






  • They make some good points about how we view “classic” games too.

    A lot of 16-bit games are remembered fondly because of things like “look at how many colors are on the screen at once! Look at how big the sprites are- they’re almost as big as the arcade version! Hear how there are 4 separate audio tracks that kind of almost sound like real instruments sometimes!”.

    Mario 64 is a great example for me. I hear other people was nostalgic about how incredible it was to be able to move in 3D space at the time, and how they spent hours just wandering around levels and marveling at the technology. For me, I did that with Crash Bandicoot (which came out a few months earlier in the US). And shortly after Spyro blew them both out of the water with its incredibly smooth controls and, imo, better graphics and sound. When I’ve tried to go back and play Mario 64 I find it a clunky mess of a game, more of a tech demo than anything else.

    On the one hand I can respect the pioneers. The original thinkers who push the frontiers of what art can be. On the other hand, those games that rely so heavily on being “revolutionary for their time” often don’t hold up well decades later when tons of games have done what they did better. I think it’s possible to appreciate those games for what they did without enjoying going back and playing them.

    When I look back at what I’ve played the past couple years, games like Control and Horizon: Zero Dawn stick out. I don’t think either one of them had anything particularly innovative or new. I see any games coming out today where I say “wow that’s a Control-like” game. But what they did do was execute on a high level, with a lot of polish and very few flaws. I think that’s the biggest strength of AAA games: execution, not innovation.