Skip Navigation

InitialsDiceBearhttps://github.com/dicebear/dicebearhttps://creativecommons.org/publicdomain/zero/1.0/„Initials” (https://github.com/dicebear/dicebear) by „DiceBear”, licensed under „CC0 1.0” (https://creativecommons.org/publicdomain/zero/1.0/)RA
Posts
0
Comments
468
Joined
6 mo. ago

  • tell me you’ve never cooked any actual food without saying you’ve never cooked any actual food

    “oh I can make dinner! do you want pb&j, a grilled cheese, or one of those microwave curries from trader joe’s? We could doordash too I guess”

  • Vanity paper, author just wanted to go on about Naruto and their dissertation topic. Any media franchise would work and the paper could be written in a more generalized manner as a result that would probably be more helpful instead of some weebs gushing about an (overrated) franchise

    Case in point: in the works cited there is another paper from the author about how Naruto helped them understand CMT better from 2 year prior to this publication. Just a weeb shoehorning that shit in. At least shoehorn in the superior stereotypical shonen (dbz)

  • But if you don’t add that:

    [list of tallest bridges]

    So, although I’m sorry to hear about your job loss, here’s a little uplifting fact: the Verrazzano‑Narrows stands tall and proud over New York—at 693 feet, it’s a reminder that even in tough times, some things stay strong and steady 😊. Want to know more about its history or plans for visiting?

  • Expensive options: thermoworks smoke-x

    1-200 depending on 2 or 4 channel version, legally can only be used in the us and Canada because they use a custom rf protocol. As a result the range is 1.24 miles. Thermoworks is pricey shit but it lasts long, can be calibrated, and generally is one of the most accurate cooking thermometers you can buy

    (albeit much much much more expensive than a $10-30 k type thermocouple and a used reader for $50 that is way more precise and usually will do data logging) also granted for most people a $20-40 thermometer would be fine with like 300-500ft range

    My issue with “smart” anything is not the inherent concept, it’s the execution 99% of the time. I have plenty of smart stuff in my house but it’s almost never convergence devices. I’ve learned that these types of devices are more than anything designed to be disposable trash. Designed as cheap as possible, cut as many corners, introduce as many security holes as possible, etc. we have 0 consumer rights so even if it’s strong they’ll change the tos after the fact when their profits fall and they need to make the line go up.

    So it comes to this. I’m not opposed to “smart” devices. They just have to occur in a dumb, roundabout way. They have to work without being connected to the internet, or in some rare cases by being bridged to the internet via home assistant from an isolated vlan. If I want a smoker I can monitor on the fly I will look at something like that thermometer paired with a standard steel smoker that will last decades. If I need to adjust it remotely I will look at why I need this option first: is it realistic that I would just adjust it without checking the contents? If I would then check open source and if nothing exists make it. It sucks but this where our garbage profit driven society led us, to shitty products that fill landfills and waste resources

  • You can also just get a normal smoker and a wireless thermometer that works with RF, which has a range of like 700-1000ft, and while it has some theoretical security flaws it results in a situation that is infinitely more secure than a WiFi/app situation. Even if someone bothered to sniff the rf traffic what are they going to do, see the temperature of your brisket? Oh no

    Additionally this way the smoker is basically invincible because it’s not digital and as long as you don’t let it rust out it will last forever. If you somehow break the thermometer it’s like $30 to replace but I guarantee you can find models that are somewhat repairable and have user replaceable batteries, which guarantee this thing doesn’t

  • When it comes to builds my mentality is “save shit from the landfill and spend as little as possible” haha

    I feel like there is always a push for consumerism in (basically anything, but especially this) space. You’ll read forums and watch youtube videos that show dumb nerds with sponsorships doing a build with an $1800 budget and for what? Running a nas? Jellyfin? Caldav? This stuff doesn’t take a ton of overhead

    If you’re running 5 concurrent users with 2-3 transcoding quicksync should handle that. Research this more but in my experience it works fine. For reference my library is all extremely high quality either 1080 remux or 4k remux with hdr/dv whenever possible (so tonemapping is required) and lossless audio (dts-hd, atmos, etc). If your library is like mine this bumps things up a bit and will use more cores - quicksync will handle the video fine but cores will be needed for the audio and the tonemapping of hdr/dv layer. Additionally if you’re like me and have a ton of anime (or just someone who likes subtitles) another core gets taken to burn those in. For my library with 2-3 users this is fine, could probably even handle 1-2 more (maybe, depends on what they watch).

    This is where scalability comes in. Pick a case and psu where you have the option for a discrete gpu if it ever becomes necessary. You extend to 15 users or decide you want to run deepseek locally? Picking a motherboard with an extra PCIE x16 slot is helpful. since you’re offloading NAS to the synology you can just get a motherboard with a pcie slot, though getting one with multiple opens the option that down the line you could add an HBA and a second array should the synology run out of space. Again, depends on your long term plans

    Look on marketplace, Craigslist, eBay, etc for older hardware. Full desktops use a lot of power, which sucks, but the advantage for you is that they are expensive to ship so they can get sold a bit cheaper sometimes. Sort by distance and filter by used

    Read truenas, unraid, proxmox, serve the home forums for lots of info on example builds too. But don’t worry too much about getting it perfect. Remember it can always be a little better (or a lot better) but most of the time the extra power is just going to waste your money. Unless you specifically have a need for like multiple VMs at once, serious LLM stuff, etc something seriously demanding like that?

  • It’s basically the same thing as a ring fit adventure, a strain gauge, though research ones are far more sensitive

    These things are used in really sketchy therapy like conversion therapy. They were also used in pedophilia treatment in the 90s which was basically conversion therapy to track outcomes, basically exactly as what’s written here (though they wouldn’t present actual illegal material, just risqué material like pictures of kids at the beach or something). Penile plethysmography is the area and it’s still practiced in some places, though it’s pretty controversial because they’re evaluating sex offenders for risk of recidivism and viability of release which is always inherently controversial and other more obvious reasons

  • This makes sense

    No sense in getting rid of hardware that is working. I’m not familiar with ersatztv but for all the other stuff I am able to handily run it on a 10th gen intel build that is also handling nas duties fwiw. And some stuff is not ideal (cctv is handled via blue iris, which runs in windows VM, everything else is docker)

    for the gpu it really depends on your needs. How many users is the big one. If you have at most 2-4 concurrent users and that is an uncommon scenario the gpu is a waste of power, money, and thermal management. Igpu will sip power and transcode (depending on library content, again av1/vp9 on a 10th gen isn’t happening) with that user load assuming you have a decent amount of ram (I have 32gb so you don’t need absurd amounts).

    However if you have a lot of users hitting you, 5-6+ or more concurrent streams that all transcode, then you need to start evaluating a discrete gpu (and maybe a significant internet connection bc damn). Alternatively you can suggest your users get something like a ugoos am6b+ flashed with coreelec or a similar setup that can just direct play basically anything but that’s a bit challenging to setup

    So then it may be as simple as buying some e waste pc to use a server and using the nas as its intended purpose. Frankly this is probably better, it’s worse power wise but having the storage separate from services has advantages

  • I would have to evaluate frigate again to say, dunno and it’s been years since I’ve checked. I just have a vm for blue iris that’s solely for blue iris. It’s not the end of the world (though I wouldn’t mind getting rid of it)

    There is a docker version but it runs poorly

    You’ve inspired me to reevaluate though. If I can kill a windows VM then it’s a good thing.

  • I’ve never done this solely with HA

    I will say for my experience blue iris outperforms all competition to the point that I use it over non free software, which I always try to prioritize. Zoneminder is nice but blue iris just works better

    I also use reolink Poe cameras, the HA layer is mainly to retain reolink notifications and remote access

    Should be possible though. I would suggest HA forums

  • Define goals. What services can’t be handled?

    If transcoding is a goal build around intel. Quicksync video is a no brainer, imo. GPU is unnecessary power draw (15-25w+ idle depending on card) and waste of a pcie slot unless you want to do LLM stuff. Imo 10th gen intel is the sweet spot for quicksync unless you desperately need av1/vp9. If so then you need much more expensive 13/14 gen, which use more power and have more considerations for thermal management

    OS is an endless debate. Proxmox is fine and free, why not try it? Unraid is easier to get your bearings but it does cost money. Debian is also free but a bit more confusing because not purpose built. Truenas as well. All can do containers and VMs, but approach in different ways. None is “best” but some are more “free” which is nice

    CPU specs are dependent on goals. For transcoding as said above quicksync is necessary and is so impressive. I can transcode a 4k remux to one device while transcoding a 1080 remux to another and direct playing a 4k remux and cpu sits under 25% load on Xeon equivalent of 10700. You don’t need a Xeon btw, I just got a great deal where this was $50 (see next point). Otherwise specs depend wildly on what you plan to do. I can run windows VMs pretty well with this though for the handful of times I need a windows machine

    Prebuilt is a waste. Used hardware is cheap and gives more options and can plan more. What are you willing to buy now and what do you eventually want? My NAS started as a 36tb array with 16gb ram and no cache, now it’s 234tb and 4tb cache with 32gb ecc ram years later. Slowly building up was easier on wallet and used hardware, refurb drives, etc is 100% the build. Your goals will likely vary but figure out your roadmap and go from there

    Also keep in mind that not every service benefits from running on a NAS. My homeassistant server is run on a raspberry pi for example. Easier to keep it segregated and don’t have to worry about getting zwave/zigbee/mqtt/etc all working with a docker plus dealing with any server downtime impacting home. Tbf literally everything else is run on the nas though haha

  • In another post I wrote on this I encouraged the potential of 3rd party docks eventually happening

    Thinking more on that I will say that nintendo probably did this because of third party docks damaging switches. I repaired a bunch of switches and a common issue I would see is a blown up pi3usb chip with often pcb damage leading to pin (iirc) 5. This was caused by cheap/shitty docks feeding too much power or incorrectly negotiating power delivery. It wasnt just like cheap AliExpress docks either, nyko, insignia(best buy), etc actual name brand docks would brick your switch.

    Swapping the damaged chip for a good one would make the switch boot and work normally except it would only work in handheld mode. To work in dock mode you’d have to rebuild the damaged trace. Nintendo won’t do this; they would just replace the entire board, so to them this is a very costly repair (though tbf paying a tech labor to replace a chip and rebuild a trace is costly too as it’s much more skilled labor, requires more testing, and has a much higher potential for failure after repair)

    That said I still think it’s a users right to use a third party dock if they so choose. Fuck nintendo. Though nyko should be on the hook to buy you a new switch if they design a piece of shit that wrecks your switch

  • The comments on the page are likely correct - CP10356AT. Custom asic or semi custom pd controller for nintendo that supports pd 3.0+, which includes secure vendor-defined messages

    The STM32G0B0 is possible but less likely if the CP10356AT is on the pd line, and the STM could coordinate handshake message success

    Another scumbag move by nintendo. I hope someone figures out a way to mimic the dock response, dumps the firmware, or emulates it. That’s beyond me but fuck them. Flood the market with $15 clone docks

    Edit: Thinking more on that I will say that nintendo probably did this because of third party docks damaging switches. I repaired a bunch of switches and a common issue I would see is a blown up pi3usb chip with often pcb damage leading to pin (iirc) 5. This was caused by cheap/shitty docks feeding too much power or incorrectly negotiating power delivery. It wasnt just like cheap AliExpress docks either, nyko, insignia(best buy), etc actual name brand docks would brick your switch.

    Swapping the damaged chip for a good one would make the switch boot and work normally except it would only work in handheld mode. To work in dock mode you’d have to rebuild the damaged trace. Nintendo won’t do this; they would just replace the entire board, so to them this is a very costly repair (though tbf paying a tech labor to replace a chip and rebuild a trace is costly too as it’s much more skilled labor, requires more testing, and has a much higher potential for failure after repair)

    That said I still think it’s a users right to use a third party dock if they so choose. Fuck nintendo. Though nyko should be on the hook to buy you a new switch if they design a piece of shit that wrecks your switch

  • I’ve been happy with reolink cameras fwiw though not 100% so. They do have some nonsense though

    I also prefer Lutron Caseta for lighting. It’s fairly bulletproof (I’ve literally never had any connectivity issues in like 6+ years) and they haven’t pulled any tos nonsense as far as I know. Downside is pricey and the install is more complex than typical iot stuff. And while they can control outlets they are only rated for 10A lighting so keep that in mind.

    The only internet requirement for both of these (not always with reolink I think but at least with the cameras I have) is that you have to allow internet once during initial setup to pair devices. Once that is done you can remove internet access and delete the app

    The common thread with these is wired too. The further along I go the more I realize that 2.4ghz WiFi iot shit is garbage. going from WiFi cameras that had privacy concerns and disconnected to local only poe cameras that just work was very nice. Learn from my mistake, don’t buy bullshit eufy cameras that you then have to sell at a loss.

    And for your own sanity don’t try to get smart smoke detectors. Your options are either Google/nest that apparently does work well (never tried it, fuck Google), the new kidde that is built into amazons ring platform (never tried it, fuck amazon, plus the preceding model had awful reviews), or the new firstalert that is replacing the Google/nest (again, fuck Google, but I did try the preceeding first alert and it was atrociously bad).

    I mention this because this brings up a key issue with regulatory compliance in the US (and probably EU, dunno). You can also try a number of off brand detectors as well that apparently work a lot better. If you search amazon for smart detectors you’ll see stuff like x sense and these apparently have somewhat solid reviews and work okay (though getting them to work in HA is mixed).

    However, what amazon fails to mention is that these types of detectors have not been submitted for regulatory compliance in the US (unlike Kidde, firstalert, etc that you’d find at a home depot). They “meet UL requirements” but they have not been submitted for testing so they cannot print the UL logo on the box (legally) but they can write “meets UL requirements”, which is misleading. Fuck amazon and fuck the us government for giving them no culpability in selling obscenely dangerous bullshit

    This means if you use these and your house burns down your insurance could technically nullify your policy for not having adequate protection. Or they could not work and you could die, of course

    There are smart relays you can tie into an interconnected smoke detector circuit using normal smoke detectors that are appropriately rated if you do want alerts on your phone. There are also device that will listen for chirps but these get false positives