juliangoldsmith

It probably doesn't help that Syncthing is a direct competitor to Google Drive.

The OnePlus 6 can run mainline Linux, though apparently audio in calls drops sometimes. [0]

0: https://wiki.postmarketos.org/wiki/OnePlus_6_(oneplus-enchil...

Oura could provide their app, but also make it possible to talk to the ring directly via Bluetooth. Not only would this solve the privacy issues, but it would make third-party apps more reliable (due to no more round-trip).

The amount of work Oura would need to do would be trivial: they'd need to provide a basic description of the format of the packets the ring sends over Bluetooth (or just release protocol headers/internal client libraries). They'd also need to not put in work to deliberately prevent other apps from connecting (which some manufacturers do; not sure about Oura).

That still requires going through Oura's API, which is where the second/third-party data collection happens.

I think the concern is privacy, not the user's access to the data.

The BFL scam was where they were mining on equipment before sending it out, right? It's too bad they went that way. I bought two BFL Jalapenos with BTC I'd mined, and the hardware was pretty nice.

As I understand it, Spotify pools together everyone's payments and distributes it by number of listens across all users. You or I would expect we'd be supporting our favorite artists by listening to them, but most of our subscription fees actually go to whoever's at the top of the charts. That restaurant playing "Top 40" on repeat effectively outweighs us both.

To be fair, that machine is pretty high-spec for NT 3.51, given that the minimum requirements are 12 MB RAM and a 386 (which tops out at 40 MHz). It'd certainly be nice to see that kind of performance in newer Windows versions, though.

A quick Googling shows that Vulkan had an experimental, Android-only Vulkan backend around version 1.7. It doesn't appear to still exist in current versions.

Autograd hides the backward pass in that implementation, but this code spells it out explicitly.

R&R is still one of the largest. The elitism was definitely something I saw when I wrote the RCI integration at Dealer Wizard around 2014. I've heard they've been trying to change the culture internally. I'm not sure how that's going, but it'd certainly be welcome.

Does Fortellis even accept other DMSes? I've dealt with it a bit to integrate with eLeads (now CDK CRM), but there didn't seem to be much on there apart from CDK's APIs.

I believe the litigation was about data integration, due to the mentions of Authenticom. Authenticom was a vendor my company used until we needed to start getting data directly from R&R. (IIRC we were a third party on a lawsuit related to that.)

Switching DMSes is likely disruptive, but is still pretty common. We've had dealerships do it every now and then.

ADP is CDK as of a few years ago. I've seen a fair number of Auto/Mate stores in addition to DealerTrack, but not as many as CDK and R&R.

That 90% is for both of the two biggest DMSes, CDK and Reynolds & Reynolds. Both DMSes have been around since the '70s or so, so they're probably considered a safe option by a lot of dealers.

The rating was the minimum number of reads the motors in the CD drive of the Playstation would work for. Most likely, the rating was really low compared to what the drives could actually do.

One alternative to PCIe risers is OCuLink/SFF-8611. It lets you plug one end of a thin cable into a PCIe (or M.2) card and the other into a board with a PCIe slot.

The main downside is that it cuts bandwidth to PCIe x4, but for most usage (e.g. LLM inference), that would be fine.

Came here to ask this. Depending on the SoC, bringing up Linux via postmarketOS could be pretty easy.

Is any of this related to newer standards like Open Rack?

Strattera, an SNRI, is one.

For Gaudi2, it looks like 21/24 ports are internal to the server. I highly doubt those have actual individual cables. Most likely they're just carried on PCBs like any other signal.

100GBe is only supported on twinax anyway, so Cat8 is irrelevant here. The other 3 ports are probably QSFP or something.

This might work for used cars, but probably won't for new cars. Dealers frequently lose money on the sale price of new vehicles, but make up for it on finance/insurance and profit from selling any trade-ins.

Having an existing offer is a good idea, but I wouldn't totally write off dealer financing.

When you go through the dealer for financing, they get a portion of the interest that the bank takes out (finance commission). That lets them take the price of the car below what they paid for it, because they can still make a profit on the financing and trade-in.

Apart from that, there are sometimes pretty big rebates and rate incentives that OEMs offer when you take their financing.

I built a system that estimates what monthly payments a dealer can offer a customer based on what incentives they might qualify for. It's pretty common to see a difference of several hundred dollars a month between the best combinations and the worst ones (even with the same term).

Unless you're wearing it tight enough to leave a mark around your mouth and nose, it's not doing anything for you.

My wife, an ICU nurse, spent a year reusing N95s because people who didn't need them bought them all up, only to wear them in a way that didn't seal properly and provided no protection.

That type of "churn and burn" doesn't work long-term. Acquiring new customers has a cost, and that cost will increase the more customers you've burned. The more customers you burn, the more likely it is that potential customers will have heard something negative about your service.

Providing good customer service improves the bottom line in the long run. Management that is unwilling to do that is either focused only on short-term gains or is totally incompetent.

Resolving issues quickly also reduces the amount of money the company needs to spend on support.

With photolithography, what most chips are made with, the resist is developed by exposing it to UV light. A mask is used so that the desired parts of the resist are left behind, and the rest is washed away. Contact photolithography places the mask directly against the silicon wafer because it helps the optics during exposure, but the mask doesn't directly remove the resist.

In nanoimprint lithograhy, a stamp is pressed into the resist on wafer and leaves the desired pattern behind.

The difference is that the resist is removed mechanically instead of chemically.

Not sure why you're getting downvoted. It performs decent enough on my Ryzen 3600X with 64GB of RAM. It definitely wouldn't be usable for production or fine-tuning, but it's fine for experimenting.

Does it still require the hack for exception handling? I remember there being a missing symbol from Windows NT, for SEH or something.

GPUs are great for things that are parallelizable, but a lot of things aren't.

Technologies like GPUDirect are interesting in that vein. Theoretically you can DMA data directly from an NVMe drive on another machine, across an Infiniband link, and into VRAM without the data ever touching the CPU.

You should only need the weights for the experts you want to run. The experts clock in at around 400 MB each (based on the 800 GB figure given elsewhere). A 24 GB GPU could fit around 60 experts, so it might be usable with a couple of old M40s.

They weren't the ones who went before Congress, asking for regulations that would prevent new startups taking root. (Something like a minimum of 25 employees dedicated to moderation against bias, as I recall. That would be trivial for OpenAI, but impossible when on a budget.)