Some applications use those unused bits to add tags to pointers but it’s important to mask those out before attempting to dereference the address. I’m not sure about ARM but x86-64 requires bits 49-63 to be copies of bit 48 (kinda like sign-extension), ironically to ensure that no one is using those bits to store extra data.

deleted by creator

Removed by mod

I actually added detail that wasn’t already discussed in the article?

We don’t even have true 64-bit addressing yet. x86-64 uses only 48 bits of a 64 bit address and 64-bit ARM can use anything between 40 and 52 depending on the specific configuration.

This would be a lot more readable with some paragraph breaks.

The person who correctly guesses when the AI bubble is gonna pop and shorts Nvidia stock is gonna make a lot of money. Call it The Big Short 2: Electric Boogaloo.

America doesn’t do anything big unless it’s to beat either China or Russia. Maybe this collider will be the impetus we need to build a bigger one.

It is being used to develop a quantum compass – an instrument that will exploit the behaviour of subatomic matter in order to develop devices that can accurately pinpoint their locations no matter where they are placed,

[…]

The aim of the Imperial College project […] is to create a device that is not only accurate in fixing its position, but also does not rely on receiving external signals.

These statements imply the device can know exactly where it is in space just by measuring some purely internal quantum effect, which conflicts with the principles of Lorentz invariance and relativity.

Both are constructed around the same idea that there’s nothing special in the laws of physics that changes with where you are or how fast you’re going. That observation is what led the conclusion that the speed of light is the same in every reference frame, and to Einstein developing the theory of relativity.

In reality, the device needs an external signal to learn its initial position. And it’s unlikely to be perfectly accurate so it may still need periodic updates, just hopefully a lot less frequently.

The London Underground is actually kind of a dumb use-case because it’s fixed infrastructure. You can just have something like RFID tags around the track that the train reads as it goes by. And there’s going to be sensors in the track that report trains’ presence to a central control room. It’s just a good setting to test the device.

What it’s really potentially quite useful for is nuclear submarines since they can stay underwater pretty much as long as their food supplies last, and knowing their position without using sonar or being able to receive GPS signals is quite important for navigation and obstacle avoidance. But the author was probably told to downplay potential military applications.

The article describes the device working in ways that violate relativity, but the actual technical description is a lot cooler.

It’s not a quantum compass, really. It’s a quantum accelerometer and gyroscope. The hope is that its accuracy will lend itself to long-term inertial guidance, which normally needs regular GPS updates to correct errors which accumulate over time.

Given how much modern games stream data in and out of VRAM, I think it would actually be quite a significant issue. Although, for modern games the 520M would probably be below minimum requirements anyway. It was just to illustrate my point.

The processor it’s using is linked in the article: https://www.cnx-software.com/2022/08/29/starfive-jh7110-risc-v-processor-specifications/

It’s a system-on-chip (SoC) design with an embedded GPU, the Imagination BXE-4-32, which appears to be designed mainly for smart TVs and set-top boxes.

The SoC itself only has two PCIe 2.0 lanes on separate interfaces so you can’t use both for the same device, and one is shared with the USB 3.0 interface.

That’s not even enough bandwidth to drive an entry-level notebook GPU from over a decade ago. Seriously: the GeForce GT 520M, launched January 2011, wants a full PCIe 2.0 x16 interface. Same with the Raedeon HD 6330M. You could probably get away with just 8 lanes if you had to, but not only one.

The other commenter wasn’t kidding by saying you could get more power out of a Raspberry Pi 4. It’s even mentioned in the article.

Any highlights from those in the know?

I dunno about other apps, but OKCupid does have this option.

Not once have I encountered a trans person on a dating app who wasn’t 100% transparent about it. Some even asked me after matching, “you’re aware that I’m trans, right?” just to be sure.

There’s no logical reason to falsely pretend to be cis on a dating app to get matches. If someone’s cool about it then it’s better to know up front, right? And if they’re not, then you probably don’t want to waste your time on them.

The “justification” for this app is just bigotry, plain and simple. Fuck TERFs.

I think a new kind of human being is taking shape, under the presumption that what most counts about us can be quantified and codified, but that presumption, while it eliminates many forms of friction and oppression, is also progressively devastating the deepest forms of human communion,

That’s not the Internet’s fault, that’s Capitalism’s. The Internet just accelerated it.

Go back 20 years, and the Internet didn’t give a flying fuck about who you were and what you could be manipulated into buying or thinking. It showed you the same ads no matter what you were into. It was just a place where people posted cool shit they’d created. Venture capitalism made it what it is today.

It executes on a native thread in the background. That way it doesn’t stall the Javascript execution loop, even if you give it a gigabyte of data to hash.

Chess engines have outplayed humans for thirty years, and they didn’t need teraflops of computing power to do it.

Generative AI is actively harmful to the environment, slowing the phase-out of coal in the US and guzzling billions of gallons of water. It’s likely going to kill jobs and it’s already filling the internet and the academic world with garbage. It’s also likely a bubble that will burst before long, potentially bringing the economy down with it.

I’ll give you noise cancellation and text-to-speech, that’s pretty cool.

But personally, I’d rather have more CUDA cores.

Meteor Lake was taped out in May 2021 and launched in December 2023. Still much slower than the pace of LLM development, to be fair. It seems more like an “if you build it, they will come” approach. But that’s also how we got stuck with (for most consumer purposes) useless tensor cores on our GPUs. Does anyone even give a shit about raytracing/DLSS anymore?

It actually sounds like Microsoft is betraying Intel for Qualcomm, since their upcoming processor in the new Surface tablet is the only one that actually meets the requirements. So it looks like Microsoft doesn’t give two shits about supporting existing hardware either way.

I sure as hell don’t, but it seems extra pointless when it can’t even run the workloads it was designed for.