The second gen ones are still incredibly cheap for what you get and will most likely have a long support window (they’re still pushing updates for the old hd models from nine years ago).

As far as being open source, I have done diy linux streaming or time shifting boxes for twenty years before switching to the appletv and the fact that it can connect up to a smb, ntfs or jellyfin is enough for me.

Just trying to save our illustrious op some money, time and heartache.

I use appletv. It’s the best available option I’m aware of and certainly the best one in the ~$100 price range.

Yeah!

So work on dumping and encoding your raws on the internal nvme SSD and then save the finished stuff and project folders you’re not using on the magnetic drives.

That’ll give you the fastest paging. That’s a ddr4 board right? The old wisdom used to be that you’d get faster performance with only two sticks of ram but that’s from back when two 32gb sticks was the same price as four 16gb sticks. Don’t shy away from using all four slots.

Oh I agree, chromecasts and any android streaming box are a hard no. It’s just really important to recognize that replacing one with a mini pc is going to be much more expensive than just using a better streaming box.

Oh lol. I forgot that amd CPUs can be put in that situation too.

I usually use Debian and have had to use legacy (bios) booting with csm loaded on those boards before. so if you can tolerate an old man distribution, try legacy booting the Debian text mode net installer.

Unless you are just trying to use up hardware you already have laying around, do not do this. The experience is extremely suboptimal and much more expensive than just buying a better streaming box.

If you have not spent money yet there is still time to avoid this mistake.

You do have an asus prime case. The smallest is the 201.

Here is the manual showing where to mount three 3.5” drives:

https://www.manualslib.com/manual/2851351/Asus-Prime-Ap201.html?page=19#manual

You bolt them to the bottom.

Why are you even using a dedicated video card anyway? If this is the same b550 board in your other post you already have onboard video…

That card probably also has an insanely old firmware too, but you ought to be able to set a 550 motherboard to bios mode instead of uefi and get past it.

I think your bios can be updated to revision 4304

What case? I can’t tell myself because you blurred out the info…

You may end up needing to use attached storage either in the form of a nas or some kind of usb bullshit.

E: asusprime ap201?

No.

The person who said it’ll be too slow and you’ll ruin your usb is right but also:

Your system supports a nvme and sata. Dumping raw to nvme then working with it there so it doesn’t take so long to page into ram will be significantly faster.

Use spinning drives for storage though.

And to bring it aaaaaalllll the way back to what your post is about: the industry signaling it would accept a limitation of some sort means that it predicts (using privileged information available to it via deep connections at all levels from design to manufacturing marketing and sales!) it will be unable to meet current demand with new hardware over the time that new hardware will be viable.

They’d have to raise prices.

And the only reason they wouldn’t want to raise prices is because they wouldn’t be profitable if they did.

Openai cancelled ipo, trumpbucks 2.0, etc. are foreshocks.

I think it’s a little disingenuous to imply that deepseek is a small model. Even flash requires 160gb of vram and it’s important to note that’s a size aimed at a constraint, the vram density of hbm equipment from two generations ago. (And the density allowed by using last generations consumer equipment plugged into ten amp circuits with doubled vram see all those 48gb 4090s floating around!)

They didn’t arbitrarily decide that they’d use a smaller model, they were targeting a constraint. Which of course was really juicy because of all the demand for text inference and the limited new hardware to fulfill it.

The paper you cite is true, I haven’t read it but it based on the title it can’t really be wrong unless they just get wildly over their skis with their claims…

You’re right that everything’s gotta walk through the layers and more layers means a longer walk. That’s limited by memory and interconnect bandwidth though which is still doubling or close to it every year or close to it.

Which means it would have to be twice as fast to use a smaller model on cutting edge hardware to be a real “wall”.

We are not near the end of the memory bandwidth road yet, quantum tunneling isn’t rearing its head again as the workaday wafers that ferry serialized streams from place to place get faster and more numerous. Because they don’t need to get smaller really. The port of New York can grow and sprawl and sprout heretofore unseen support structures like coolers and voltage regulators and coolers for its voltage regulators.

I’ve seen in action what you describe though. On a, sensible chuckle, tiny card like a 3050 slotted into a system with pcie3 very small models run faster than those who can barely fit in the vram with their little bitty context, but that difference shrinks significantly when the card is a 5060 or something that still has limited vram but can load faster due to a faster pcie interconnect.

Now that may seem like apples and oranges because it’s two completely different things but the point of the comparison is to show that the difference between performance measures in time to first, tps or whatever other measurement might be in vogue at the moment when two models are compared shrinks when the interconnect and memory bandwidth gets faster.

To butcher a car metaphor, you can turbocharge an ls and get more power but you’re not escaping engine wear=crankshaft rotations. That might not be as butchered as it first seemed even though the domains are all shifted.

You said there’s no reason to think you can just keep making the model bigger and keep getting improved capability.

there’s the structure of the neural network itself. Fundamentally, adding nodes and layers increases the ability of the model to handle more complex input.

Then there’s the actual models we see in use. They are literally as large as the hardware allows. The only reason to use smaller models are to fit some constraint.

So both by the book and in practice bigger is always better.

Now we can’t always go big. I can’t afford to purchase a dgx or even upgrade my wiring to power it, let alone pay the power bill it would rack up or all the other utilities alone when my wife leaves me because of the sound.

My computer can only fit so many expansion cards and pcie is so slow compared to hbm that I’m better off running a small model quickly that fits on one card as opposed to a larger one slowly across several cards.

But those are all constraints. When I replace my motherboard with supermicro gpu host fabric I no longer am limited by the pcie bandwidth and can quickly use models that fit across several cards.

I do agree with you that the future is smaller models, not because of the fundamental nature of the concepts involved but because of the complex constraints that are coming into play.

You could create a way to measure the idea of capability that would bear that out but from a pure discrete mathematics perspective, no, you only get better with a larger memory footprint.

There’s a lot of ways to make that faster or make that behave like a process running on a bigger memory footprint, but ultimately that’s the constraint.

And companies competing in the field of ai can’t justify the expense of cutting down their gigantic model to only know how to identify wood because that has a known and limited impact. They already said they’re shooting for unlimited immeasurable impact on the scale of replacing all human labor and got massive funding for it.

It doesn’t matter if it’s easier to do one backflip, you asked me to triple dog dare you to do a million backflips. Well… we’re waiting!

You can’t do that with tor browser.

Above doors or above other windows. They’re the ones up high that pivot open and closed.

I have never seen any implementation of a microphone that has “critical” components in series with it.

I’d advise against that from an engineering perspective because microphone failure is often a break in the conductive path and standard troubleshooting of non audio in failures generally doesn’t place a high priority on checking mics for continuity.

Even on a throwaway component like a remote, requiring microphone continuity would introduce unnecessary complexity and more chance for problems.

Ligmoid