Dad, architectural designer, former SMB sysadmin and still-current home-labber, sometimes sim-racing modder, enthusiastic everything-hobbyist. he/him.

  • 1 Post
  • 54 Comments
Joined 1 year ago
cake
Cake day: June 9th, 2023

help-circle
  • M1 gets most of its performance-per-watt efficiency by running much farther down the voltage curve than Intel or AMD usually tune their silicon for, and having a really wide core design to take advantage of the extra instruction-level parallelism that can be extracted from the ARM instruction set relative to x86. It’s a great design, but the relatively minor gains from M1 to M2 suggest that there’s not that much more in terms of optimization available in the architecture, and the x86 manufacturers have been able to close a big chunk of the gap in their own subsequent products by increasing their own IPC with things like extra cache and better branch prediction, while also ramping down power targets to put their competing thin-and-light laptop parts in better parts of the power curve, where they’re not hitting diminishing performance returns.

    The really dismal truth of the matter is that semiconductor fabrication is reaching a point of maturity in its development, and there aren’t any more huge gains to be made in transistor density in silicon. ASML is pouring in Herculean effort to reduce feature sizes at a much lower rate than in years past, and each step forward increases cost and complexity by eyewatering amounts. We’re reaching the physical limits of silicon now, and if there’s going to be another big, sustained leap forward in performance, efficient, or density, it’s probably going to have to come in the form of a new semiconductor material with more advantageous quantum behavior.






  • The thing about EVE is that caveat emptor is the social contract, and as long as you don’t get your dander up about being pirates or scammed, the “bad guys” are more than willing to help you learn to avoid the next trap. I was once part of an “anti-piracy” roaming fleet that got bored with the quiet night, and ended up jumping a newbie who was hanging out where they shouldn’t have been, tackling his ship, and proceeding to ransom it back to him for the princely sum of 1 ISK while we laughed our heads off in Teamspeak. Then we sent him on his way with a few hundred thousand ISK extra, some pointers on highsec versus lowsec, and the valuable lesson that there were always sharks on the prowl for easy prey.


  • Speculations indicate that Navi 3.5 might enable integrated graphics with performance comparable to an Nvidia RTX 3070.

    Uh huh. Given that the Radeon 780M that represents the current state of the art in Zen4 iGPUs is still trailing a discrete 3050 (by no means a strong performer itself) by about 30% on average, this seems wildly optimistic. Don’t get me wrong, I would love a beastly iGPU, but this seems less like informed speculation and more like fanboy hype.



  • I don’t know if there’s many major outlets that are primarily investigative in the era of the 24/7 news cycle and the accompanying need to always have something fresh on the front page, but at least in the English-speaking world the various newspapers of record (think places like the New York Times or the The Guardian) still have a decent newsroom and publish original investigative pieces. In audio formats, NPR and the various constellations of associated organizations like the Center for Investigative Reporting do excellent work as well. There’s also organizations like Bellingcat that specialize in deep-dive investigations using open-source intelligence, presented in a “just-the-facts” format without editorialization.


  • Because a significant chunk of what gets passed off as journalism on such sites is just writing copy – for example, regurgitating press releases, or repackaging the work of another outlet that actually did do the legwork of investigative journalism. I don’t think there’s anything inherently wrong with using AI tools to speed up the task of summarizing some other text for republishing, but I do question the value of such work in the first place.

    It’s going to be a long, long time until artificial intelligence can do the work of a true investigative journalist.


  • SpaceX’s Starship launch was much-memed-upon but it honestly went as well or better than could be expected given the development approach the company takes. That said, it’s clear that the test cadence is being rushed at Elon’s behest (launching without a proper pad deluge system, for instance) and they’ve reached a size of rocket that having something go wrong in flight could cause serious damage, and isn’t just an opportunity to deploy funny acronyms and giggle.

    That said, SpaceX is one of the few things he’s doing that isn’t a total clusterfuck, and that’s got a lot to do with the much more competent people he has running the company under his nominal leadership. Gwynne Shotwell has been very well-regarded and tends to do a good job of insulating the rest of the company from Elmo’s dumbest whims.


  • Sure, I could, but for me the whole appeal of a federated service is its interconnected nature. If I have to create an account on the “right” instance to interact with all the communities I’m interested in – potentially repeatedly, given that operators may choose to defederate from each other at any time and for any reason – that has already defeated the purpose of the exercise.

    I see this as a major driver of the “default instance” issue that both Mastodon and Lemmy have experienced. If you both have a hard time finding content off your home instance and have no guarantee of continued ability to interact with that content down the road, the safest choice of home instance becomes whichever is biggest, and from there the network effects just make the problem progressively worse over time. Federated services need some way to reduce the friction of traversing the Fediverse if they’re going to avoid that.


  • I’ll be honest, “dig through instances and code your own interface” is kinda where I’m at right now with Lemmy. I love Beehaw as an instance, but because of the demo it’s attracted there are relatively niche topics I am interested in that it can’t support an active community for. Those communities are already humming along on lemmy.world – but I can’t interact with them because of the (justifiable) decision to defederate from them due to moderation concerns.

    So where does that leave me? I am trying to stand up my own instance right now… but the build directions available don’t work for my homelab setup (for some reason the backend only responds to requests from localhost, which means I can’t set Lemmy up to work with my existing reverse proxy?). I guess I could go rooting around in the code to change that, but at that point I’m committed to maintaining a personal fork of Lemmy just to be able to use it in a way that is analogous to how I used Reddit… which just worked out of the box.

    I really wish that in addition to federation, Lemmy offered some sort of OAuth-style portable identity that would allow users to interact directly with instances off their “home” one. As it stands, the Fediverse has ended up a bit more balkanized than I think was intended or anticipated