The RP2350 is in fact even more exciting/interesting:

The burned-in bootloader is only setting the initial state of each CPU core, they can in fact be individually switched between Arm and RISC-V mode, and yes you can in fact run one of each CPU cores in parallel!

So a fast way to experiment with RISC-V code is leaving a 'monitoring' process on core 0 in ARM mode that watches for new code to be uploaded over USB with small standard libraries, stop core 1, unpack the fresh code to SRAM, restart core 1 again in RISC-V mode.

RP2350s with 8MB of PSRAM can in fact run RISC-V Linux, though it would be one of those 'sans bootloader' installs with just a raw kernel in that case, and serial console over a RPi debug cable.

https://github.com/Mr-Bossman/...

So yes, it could in theory have an RP2350-unique spin and as more upstream features happen there like SD card and/or HSTX DVI output it could happen.

Basically all major hosting providers even describe this technique as a cheap easy way to add load balancing without a dedicated load balancer, this 'fast flux' method is just a way lots of people did it if they didn't control the DNS server either by constantly pushing new DNS records instead to cycle between.

It's been used for decades, plural. Just suddenly it's a big security threat because it makes tracking more complicated somehow?

It's not an issue of silicon. They still support 32-bit data types just fine.

It's 100% a driver code issue.

Especially with all the Recall and AI crapware on top of the "TPM 2.0 required" bullshit to lock down devices away from your own control, and anyone's surprised folks are going "BLEEP THAT!" to that faustian choice?

And 90% of the difference in cameras is not the sensor, there's only a handful of those on the market in the end.

Install the Google PXL version of the camera app and you'll regain most if not all of the camera quality basically instantly because you'll get the properly massaged sensor output instead of the raw bits that the OEM vendor camera app spits out.

This site for the challenge is absolute dumpster-fire trash tier for interface.

It tries to enforce 'committing' code like git, but never actually stores anything so you lose all progress even running a test then going back to your code... because it resets back to the 'challenge' point in it's internal commit tree.

And it also has several LLM Generative "prompt engineering" challenges later in the list it looks like, so have another nope.

And the forced loud autoplay videos are just the third ribbon-on-top nope of the whole thing, what a waste.

The 'reset' button is recessed on the left, there's nothing stopping you from plugging that into a GPIO or other header on your system instead to do exactly that. :D

As someone that's built several PCs with the GD09 this retro case is based on, it's one of the best airflow cases on the market and has gobs and gobs of room to route the cables all the way up against the front of the case with good places to cable-tie them in place across the front framework, and 100% of the airflow is to the sides by default. It supports either 'positive pressure' case with flow out the top/rear, or cross-flow towards either side which I usually prefer.

It's not a case with glass panels to see the insides, it's a case to setup once in a clean way internally then just use the computer for years after in comfort, with only having to occasionally take the fan filter panels off to dust them off and clip them back in place.

...it's /bundle/10143 and /bundle/14082, the "SEGA Mega Drive and Genesis Classics" and "Dreamcast Collection" bundles, to see if you're missing anything you want to pick up.

And there's damnably few (basically none) studies even attempting to find causation versus correlation, and many of those aren't finding that being on social media harms kids, but that harmed kids flock to social media as their only available avenue to find others when those in the guardian roles fail them.

And yes, I'm being specific in choosing my words to be as general as I meant to be, please don't try to waffle these rice-cooker pancakes.

Their annual recurring (I.E. subscription) revenue in 2023 was almost 3.5 billion, with a gross profit margin of 75-80% on that income.

They could (maybe with indigestion) stomach a 5.5 billion loss after several years similar to 2023 in recent history.

That's literally all certificate pinning was.

It was a header that announced valid hashes of the public key of the certificate. Literally all it was.

Pinning had two problems which is why every browser removed support less than 5 years after adding it. (2015-2019)

1) It didn't scale. At all. Even internally it had issues trying to scale to allow for updating the certificate, and absolutely no way to recover from some mistakes as you'd functionally lock yourself out of fixing them because all browsers that obeyed the standard would refuse updates unless they matched. And the size of the pinning data grew exponentially in very short order as more sites tried to do it, if someone could inject headers into your site they could lock everyone out with a malicious header, etc.

2) Many SSL vendors didn't implement the standards properly and often required entirely new private keys for every renewal, which was directly fighting the pinning standards. This is still an issue even today when even NIST has for years and years now recommended re-use of private keys for years at a time. At least Let's Encrypt does THAT part correctly!

Quake actually did this even, so rapid tapping smooths from the inertia (which is generation-dependent, Generations on Q3A is great to explore this).

The real issue is ADAD spam messes up the movement prediction because it counteracts the core idea of most of the anti-lag tech.

Inertia + a tiny bit of enforced minimum latency on things (like 50ms) so the inertia can be smoothed in would indeed fix like 99% of these complaints because it puts a 'A-10 BRRT speed limit' on how much tapping matters but still lets folks use it for precise positioning as much as they want.

Honestly I wish more games (especially racing games) did they so that rapid left-neutral tapping could control the steering better than it does instead of expecting/demanding everyone use an analog input for steering/throttle/brake like they do a lot of the time now.