• 1 Post
  • 80 Comments
Joined 2 years ago
cake
Cake day: July 2nd, 2023

help-circle




  • I’d never looked at them before, but yeah that super flower super modular supply looks pretty sweet. It looks like it has a ton of ports that I assume can be wired up as whatever you need.

    For me, the splitters were just generic: they plug to an existing molex out connector and give you 5 SATAs on a ribbon.

    https://a.co/d/gXtQ3Qp is what I’d bought, just for reference. The power supply I used them with wasn’t modular (ancient) and so whatever it had was what there was.

    Maybe I misread, but if you are planning on having two different PSUs in play for the same system, it’s my understanding that it’s important to make sure the DC outputs share a common ground, which might be a little extra wiring.


  • Depending on how power hungry the drives are, and if your PSU has enough spare power, you can get cable splitters. I had some spare molex ports which I plugged a cable from Amazon that split it into 5 SATA power connectors.

    You don’t want to infinitely split cables though, as tempting as that can be, because there are real electrical limits to doing that. Also just because a power supply is rated at X watts, that’s the total. Hard drives will use the 5V and 12V rails and usually there are individual limits on each rail.

    Upgrading the PSU is another option. Probably the cleanest easiest best solution IMO. But even then, you probably can’t find a PSU that’ll give you 12 SATA connectors out of the box so you’ll probably need some splitters in there anyways.

    In my case specifically, I’ve actually got a second power supply (because i already had it and it was otherwise just gathering dust) powering the extra drives. It’s a bit more complicated to get set up but, it’s an option as well.

    Edit: also if you’re asking yourself where can you physically PUT the drives, I 3D printed these and slapped some fans on them:

    https://www.thingiverse.com/thing:4875498




  • Critical is that HOW you learn this is trial and error.

    Most people can imagine the result of combining two images, say a frog riding a turtle. We can imagine what a handful of wet spaghetti might sound like being dropped onto the hood of a car. We can imagine what a fluffy bunny that’s been rolling in sand might feel like.

    But that isn’t just because those senses are somehow intrinsically better for synthesis and prediction. We just got a ton more practice with them. As kids we got to draw, we got to play with toys, we touched everything, we bashed all kinds of stuff together.

    But most of us, we just got the food prepared for us with no awareness of the properties of the constituent ingredients.

    You gotta act like a toddler in the kitchen to grow that part of your brain.


  • As others have said, running out of motherboard SATA slots doesn’t mean you need a new machine to support expansion.

    You can get m2 adapter slots for more SATA drives.

    If you think you’ll be building a NAS in the future, and are cheap like I am, you might consider getting a pci-e expansion card for SAS rather than SATA drives. They’re backwards compatibile with SATA drives, but open you up to being able to use SAS drives which are common in enterprise data centers. You can get used lots of those drives on eBay WAY cheaper per TB when the data centers hour them out.

    I’ve got a machine with 16 SAS drives running the unRaid OS, and I’m very happy with it for data hoarding and media serving. The drives (with shipping) cost $5/TB.


  • I am not implying, I am explicitly saying the process of memory recall is error-prone.

    And further to the original commenters point, we already have enough understanding of the underlying physical mechanics of memory to be able to say that pass-by-value is a more appropriate analogue to how memory works than pass by reference.

    If you fuzz the value of a value by 10%, your value is still within %10 of the original value. The same can not be said for pointers.

    That isn’t an explanation of how we arrive at an understanding of how memory works. It’s just an easily understandable statement for a computer scientist to help “prime the pump” that there may be some low-hanging reasons why thinking of human memory in terms of pointers might not be a great analogue.



  • As someone with a kid… I agree 100% w/ this. It’s like when old people say “youth is wasted on the young”

    Don’t you dare waste your freedom! Don’t answer your (relative) lack of responsibility with arbitrary and self imposed limits. Call in sick and fuck off to the mountains on a Wednesday and spark a J. Just because I can’t doesn’t mean I don’t desperately want that for you.




  • I don’t see the two environments as necessarily being at odds in any way.

    If implementing feature X is going to take a developer 10 days… It’s going to take a developer 10 days. I can say the deadline is 1 day all I want, it’s going to take 10 days.

    If I want to get my Volkswagen golf down a 1/4 mile, it doesn’t matter how hard I push the gas pedal, it’s going to take as long as it takes.

    In a corporate environment, if deadlines are what you’re optimizing for, you have options. You can cut scope. You can add resources. You can decrease quality. You can forgo time intensive processes designed to reduce risk. These are still all agile activities. Making deliberate decisions, and continually evaluating those decisions is agile.

    Agile doesn’t mean there are no timelines or goals. It’s just that the design and implementation are routinely examined for suitability to your ultimate goals.

    So I actually think agile is better suited to corporate environments because of how volatile the definition of delivered value is. Open source projects usually have a less volatile vision