I don't understand the hype because I'm out of the loop. Is the only advantage the lower hardware requirements, thus cost? Is there something I'm missing?
OpenAI o1 and Deepseek r1 have similar performance (o1 is a bit better at reasoning though you can see r1’s though process which you could argue trumps the competition).
OpenAI o1 api cost: $60/million output tokens. Deepseek r1 api cost: $2.19/million output tokens.
> Is the only advantage the lower hardware requirements, thus cost?
Yes, but the keep thing is it performs nearly as well as models that are 100x as expensive.
The lower price drastically changes possible utility. For example, I've been rocking RooCode since R1 came out. R1 can do about 95% of the tasks Claude can, but at 1% of the cost. I might burn $10 to $20 per hour on Claude tokens. While spending less than $1 on Deepseek when doing the same task.
Yeah it's a lot more efficient, it's also a very advanced model that answers questions in a multi-step way, like OpenAI-O1, it performs extremely well.
My understanding is Surface Linux is mostly the project built around developing kernel patches to be upstreamed, so the specific Kernel is more focused around external patches. Here’s their comment on this from their GitHub [0]:
We aim to send all the changes we make here upstream, but this may take time. This kernel allows us to ship new features faster, as we do not have to adhere to the upstream release schedule (and, for better or worse, code standards). We also rely on it to test and prototype patches before sending them upstream, which is crucial because we maintainers cannot test on all Surface devices (which also means we may break things along the way).
So should you install this custom kernel and the associated packages? It depends: We generally recommend you try your standard distribution kernel first. If that works well for you, great! But if you're missing any features or experiencing issues, take a look at our feature matrix and give our kernel and packages a try. If your device is not listed as supported yet, feel free to open an issue.
This is interesting. Just the other day I did a search on DDG and had the same thing happen, a full page of .it TLDs that vaguely matched what I was searching for. This is the only time I have seen that and unfortunately I don't remember what search terms I used, probably something related to my recent video card issues.
I used namecheap to register my domain, and they have an option for dynamic DNS without ongoing fees (besides domain fees, I guess). I would be surprised if other registrars didn't have something similar.
Be warned these plans require you to drill a couple holes in the conduit. I managed with just a handheld drill though the holes weren't quite aligned. Definitely use a file or hacksaw to create a flat spot on the conduit where you want the hole so the drill bit doesn't wander as easily.
One favorite trick of mine is to print out 1:1 drill pattern drawings and center-punch through the paper onto my metal workpiece for all the drill locations. Fast and accurate.
Basically, 2D printers (you know, those $150 things) are exceptionally high precision and accuracy tools for making 2D drawings. I've been using printers for years and it never occurred to me you could use it to print a (for example) 10cm square.
Fusion360 seems to try to prevent you from exporting PDFs with the free hobbyist version. One tip that I discovered was to create a CUPS printer on a Linux VM that saves PDF files.
(I found it printed slightly off sized if I sent a 2D drawing straight to the printer.)
Yeah, you have to correctly configure the output to get dimensional accuracy. but it shouldn't involve finding some magic scaling factor for X and Y that makes things accurate.
I've used Inkscape to make basic shapes, and pay for Fusion 360. TBH I've never actually thought to take one of my 3D Fusion models and use it to make a 2D template for drilling. That makes sense...
My favored approach to offset drilling / drilling on curved surfaces is to use an endmill. Doesn't wander, goes straight in. Of course, you need an endmill for that approach, but id you're building a CNC those should be in ready supply.
interestingly you can also mill a flat or indexes to mate tubes together. I’ve also contemplated making one-piece saddles to mate extrusions, filling the gaps with zero-expansion epoxy.
Tubes and extrusions you buy cheap rarely have dimensioning and tolerance you’d want to accept out of the box. To get what you need, best just to use geometry of hole centers, and adjustable fine parts.
I set a limit of 1/128 inch on any garage woodworking projects. This is 8 mil (thousandths of an inch) or 0.2 mm. Wood and plastics (and even aluminum) fluctuate from moisture and temperature enough to make this a lower limit of reasonable value, though I’m getting closer to 5 mil in router precision. It’s not a fine carpentry shop and I’m not making anything that really needs better than eyeball precision (hand marking) which would be about 1/32 inch.
Applying geometric dimensioning and tolerance to design has been a liberating experience. I’m not a mechanical engineer or even otherwise anywhere close to the industry so I really had no idea how to assess or compare designs.
Bitwarden offers this as well, but I don't really understand why you would want it. If someone compromises your password manager, 2FA is now worthless. Or am I misunderstanding how it works?
Your understanding is correct. 1Password requires a key in addition to the master password. And finally, 1Password can have 2FA for itself, which is stored on my Authy. These are reasons why I am comfortable storing my 2FA codes on it.
Bitwarden has 2FA support too, but does not have the unique key feature that 1Password has.
