If it's in the CPU I bought, how should it never have had that feature that's clearly in the CPU I bought?

This is the CPU equivalent of those car makers wanting a subscription to enable the heated seats. Maybe AMD will enable it for $5 a month or something.

It's basically buying a car and having heated seats installed even if you didn't pay for them. They did it because it simplifies production. If you choose to enable it yourself, it's unsupported - so if you activate the heated seats yourself that sets your car on fire, they may not warrant the vehicle against the damage and insurance might deny coverage. And yes, usually the heated seats are just left unconnected, so people have hooked their own power connections and switches to manually turn them on and off.

Likewise, producing a a die is very expensive - it's like $100K per mask, and you need 20-30 masks per chip (so about $2-3M to produce a mask set which needs ot be done before you can make one chip). Those chips are then fused so they can be customized per requirements. So one die design can fulfill several lines of processors from low to mid to high end chips and create product differentiation.

Of course, the documentation also will usually not describe features you're not supposed to have,usually those registers are marked as "must be set to zero" and configuration registers are not documented. It's why you often find missing registers in register listings.

Enterprising people who have access often can discover hidden functionality if they try misconfiguring the register and seeing what happens. But such things are unofficial.

Of course, it's entirely possible that because to fix some bugs, they may need to disconnect some blocks so they could re-use the transistors - because often you can get away with just re-wiring the transistors rather than having to remake the entire mask set. It's what makes the difference between say, B1 to B2 steppings from B5 to C0 steppings - the B1 to B2 usually just means a metal layer rework so it's much cheaper as you only need to redo a subset of masks. When they go from B5 to C0, it usually indicates that a whole new mask set was created.

But it could easily mean that they fused out the MEU so you couldn't unofficially enable it, or maybe they borrowed the transistors to fix some other flaw.

Well, school has been sucking the life out of reading for over a century nowadays. It's always been a problem - honestly I love reading and read a ton of stuff. But I also remember reading being a chore in school because the stuff you're forced to read generally isn't very interesting.

My school didn't force me to read 1984. So one day out of boredom I read it because it was referenced in a lot of places, and I found it a very interesting read. Maybe not if I was forced to read it, long before anyone really pointed out why I should be reading it.

Maybe school should start emphasizing why the reading materials chosen are actually relevant instead of just going through Jane Eyre and analyzing all the subplots.

Of course, this is in early elementary school though. By college you already should have a basic amount of reading level because you gotta read your textbooks. But also by then most reading I woiuld do would be recreational, and some from genres I really dislike in my days growing up like history. But give me a book on the history of computing? Yes please!

School does a bad job at encouraging students to read in the early years. But by college one should already be able to be literate. Then again, with your phone pinging every 5 seconds, chances are the real reason is shortened attention spans - you can't sit down and read a page without your mind wandering or waiting for that ping.

I remember back in 1997 seeing countless job postings that needed 20+ years of Cisco experience, and have seen the same sort of insanity repeated with every new tech fad that comes along. Why would the biggest tech fad of all be any different?

At least though Cisco was around since 1984 so you could get the better part of 20 years of Cisco experience.

The worst was 1996 or so and needing 5+ years of Java. Which has only come out the year before.

I always wonder though what recruiters and HR folks think when all their applicants all seem to fall well short of their requirements/

That said, what Trump did was crazy, and Iran might be more likely to get a nuclear weapon now than they were a year ago. If you want something done right, don't trust Trump to do it because he's like an MBA: his primary skill is knowing how to take power.

That happened during Trump 1.0 when he ripped up the agreement Obama negotiated with Iran. Something that took the better part of a decade to accomplish - 5 years of informal back and forth negotiations followed by an intense 18 month diplomatic formal negotiation.

And by the end of it, Iran had nuclear inspectors crawling everywhere, including ones from Israel per the agreement. That was the status until Trump ripped up the agreement, and Iran kicked everyone out as retaliation.

The best I got from this is "status quo" - the US side saying that. Iran's side just added reparations. In other words, nothing happened other than hundreds of billions of dollars wasted, lives wasted, and we're back to where things were on Feb 27.

Except now Iran knows it has a powerful economic weapon it can wield effectively.

And prices will not drop - everyone's been using up the strategic reserve to make up for the loss of transport. Those reserves need to be replenished, which is why gas prices will remain elevated.

At best, because the strategic reserves were likely reaching levels never seen since they were established in the 80s, is likely why the US had to agree to basically a nothing plan - because in a month or two, gas prices will shoot up again and it'll be like the 70s.

Add to that "The US doesn't need anything Canada has". Well, if you kill off importing oil from Canada, that trade deficit turns into a surplus with Canada. Oil imports are the only reason there is a trade deficit.

Funny how we worry about tiny errors in sqrt on a computer and yet we rave about hallucinating AI. I think fundamentally the problem with AI is that the common non-techie is believing the answer from AI like they do from a sqrt. I still recall I think it was Intel that had the major flub in some IEEE math result that many wanted to have Intel do a complete recall. I can't remember the details it has been so long and I didn't own the product so it did not affect me. If it was an AI error, no one would even notice it was such a tiny error like microsoft's calculator.

Because people do expect their calculators to give exact results. And small errors in calculations bring questions.

AI hallucinations are giving people huge problems - you see it in forums where people ask for help, show the AI guidance they used and then everyone is pointing out the AI slop they followed was incorrect. But then the poster gets indignant and refuses to take on any suggestions to fix their problems because they believe AI is better than humans.

As for the Intel bug, it's because early Pentium 60 and 66MHz chips had an error in the FDIV instruction that would give an incorrect result. Intel tried to play it down as an error so minor no one would encounter it, but that just made people angrier. Because it's not that no one would encounter it, but that if one did, they wouldn't know and might be relying on it. That and the fact that the Pentium was an expensive chip, so people spending several hundred dollars on a CPU would want one that was correct.

The error was in the division tables that are on-chip. Like many floating point operations, the instruction relies on internal tables to speed up calculations. The problem was the code to generate the tables was faulty and produced a zero instead of a value. That faulty table was burned into the table ROM of the chip and produced the erroneous values. CPUs with a correct table didn't have an issue. (I believe the table was a log table as it's easiest way to speed up a division by turning it into a subtraction).

Details: https://en.wikipedia.org/wiki/...

I fixed that for you. Can't wait for the mental gymnastics fiat supporters have to go through to explain how inflation is good for the economy

The economy relies on money circulating around. You need to spend money on stuff which will keep the economic engine going because that money then gets spent on other things and so on.

Now, there are actually 4 events - deflation, zero growth, inflation, and hyper inflaction. All except inflation are bad. Deflation is bad because it encourages savings - why buy a loaf of bread today when tomorrow it's cheaper? Sure you need to eat, but maybe you need to eat a little less so we can stretch what we have a little longer so we can avoid buying what we need the longest.

Zero growth is similar - though at least it means you eat as necessary since it's not going up or down. But there's no incentive to do anything.

Hyperinflation, or really, inflation beyond mild inflation is bad - and that's generally considered more than 2%. Prices are going up so fast that wages can't keep up and you really need to buy now which contributes to more inflation.

Mild inflation is where you generally want to be - usually between half to 2% at most. Here your savings will devalue over time, but not so fast you need to spend it all immediately with none for the future, but also means you want to generate value added products from the economy - gathering raw materials, refining it, processing it, producing products, etc. Each economic step adds value and the economy grows.

Bitcoin's fault is it's deflationary. Spending it is stupid - it's why BTC commerce is a fringe activity. No one's using bitcoin for everyday transactions because it costs more and it's stupid. You're better off collecting bitcoin because it's a fixed quantity and once you run out, that's it. That's why the big whales in BTC aren't doing anything - lots of coins are stored up in cold wallets. Winklevoss twins reportedly own over 20% of available coins. Of course, they have enough assets elsewhere that they're not using BTC at all for day to day living.

The real scam to Bitcoin is simple - it's trivial to start another coin. It's why any exchange has hundreds to thousands of coins they support. You know the big ones, but the rest are all small bit players. You also know Trump coin is in there.

Meanwhile, starting your own fiat currency is a lot harder. It does happen now and again, but very few actually last, almost none span beyond a local town and get very little coverage. But there is the odd alternative currency that does - any Canadian knows about Canadian Tire Money, which in the early days were paper reward coupons but before they stopped issuing physical notes started having real money-like protections because many people began accepting them informally.

What is practically useless are all the airplanes in the air at any one time, most of which burn more fuel in one flight than it takes to heat a 200 homes for an entire winter. We need electric aircraft - badly.

But electric aircraft are impractical from many aspects. You see, unlike a car whose engine is basically idling most of the time an aircraft engine is working hard. A car engine may generate 200+hp on 1.5L displacement, but an aircraft engine can get 160hp... from 5L. But the aircraft engine will be developing 80-100hp just flying through the air, while the car engine is likely only doing 10hp to maintain speed on the highway.

Car engines have been repurposed - but are severely de-rated.

The end result is the car engine spends a lot of time doing nothing and running inefficiently, which leaves open possibilities like hybrid drivetrains where you can run the engine more efficiently and use the excess to charge a battery for later use.

Hybrid engines have been tried for airplanes, and they aren't very successful because there aren't many opportunities to generate electricity.

Electric aircraft are a reality for short (20 min) flights or training, but it's not likely to be beyond it.

What part of people don't give a shit as long as their water and power are not affected is unclear to you? That's it. Data centers in small towns will always affect the local infrastructure. In decades past, datacenters had to build all the infrastructure they needed. And the locals didn't give a shit as long as they did that. What is happening now is the rush to build new datacenters wants to skip over the infrastructure problems.

Until recently, datacenters were located near urban areas - the power, water and more importantly, data connectivity required it.

The ONLY reason they're going after small towns is cheap land. But small towns rarely have big city infrastructure - the power and water are sized for the expected growth - a data center can easily consume double or more than what the town does and overwhelm the local utilities. And other times, it's simply beyond what the resources have available - aquifers only replenish so fast.

And in other places, it's water rights being bought up for data centers - but it also means excess use costs a lot more.

I find it interesting that GPS, Galileo and BeiDou share 2/3 of their base frequencies, but GLONASS doesn't - its overlap is additional frequencies. I'm not a comms guy, but I do wonder if that means Russia can interfere with GPS, Galileo and BeiDou simultaneously without affecting their own gear significantly.

Most "GPS" receiver modules these days support GPS and GLONASS. GLONASS support has been available for well over a decade on most modules used on phones these days. You can buy a very modern GNSS receiver module that support GPS, GLONASS, BeiDou, Galileo and a couple of others for a "6 system" receiver on AliExpress for a few bucks.

Multi-system GNSS receivers are common, easily available and can be dropped in and used. Nowadays they're just a single chip solution where you feed in RF and it comes out as standard NMEA strings.

Computer Science is about the science of computing - algorithms and other things. It's the same as studying physics, chemistry, and other similar sciences - providing the future theoretical foundations. Here a Turing machine can exist properly.

Computer Engineering is the about application of science - and like other engineering like civil, mechanical, electrical, etc. You have to deal with real-world compromises - just like not all cows are perfect spheres, you have to consider actual aerodynamics of your bovine. You can't create a Turing machine here because you're dealing with real-world limitations. But you can come up with limited versions to do useful work.

Neither are vocational training - they are after all not trade courses. If you want programmers, you can go after the trade schools teaching programmers. These will be folks you can hand a requirement spec to and get code out.

Computer engineers generally tend to be people who can design systems - you hand them a product idea and they can break it down to stuff that programmers can code for, Computer scientists are the ones who you stuff into the R&D departments to produce the technology that the engineers will create and apply.

It's not just cheap cars. After all, in the 80s Hyundai released the Pony - a stupidly cheap car that sold big at the time. But anyone who's tried to start one on a rainy day knows, they weren't great.

Set back the Korean car industry for decades because everyone knows, knew, or had personal experience with the Pony.

BYD's cheap EVs, though, are really good. Quality is fantastic - something Tesla needs to take note of because there's no excuses for gap and paint issues on cars costing 5-10x more.

Even better - no spyware - BYD is too cheap to install a cellular modem in the car. So if you want infotainment, you have Apple CarPlay or Android Auto.

It should be telling that the Ford CEO refuses to return his Chinese made car because he's still learning things about it.

It's also not inherently safer. The problem with current Li-Ion isn't the chemistry - it's the charge density. That's why a discharged spicy pillow isn't. You can stab both a charged and a discharged battery, and only the charged one will catch fire. The discharged one does absolutely nothing.

The breakpoint is around 50% - below that and they're rather tame. Above that and they get spicy.

If the battery technology offers an equivalent charge density, well, it'll have the same problems. Storing that much energy results in fires. Doesn't matter if it's electrical, chemical or other - it's why batteries catch fire, gas-powered stuff catches fire, diesel, etc.

Yes, even LiFePO4 and sodium-ion batteries have been made to catch file as well. Same reason - it's not the chemistry, it's the stored energy.

Well, I'm sure governments get better rates. But yes, it's likely a nationalism thing. Stripe, being American and Adyen being European. People are dropping American tech when they can switch, and I'm guessing the UK contract was up.

And while they may be expensive, it's probably cheaper since they can handle card payments online without having to do all the PCI security stuff.

Not just that, but subscriptions have probably hurt as well because it's a lot harder to audit when everyone is subscribing and now you're paying per user, not per computer.

After all, you can buy Microsoft Office outright for $400, and it's good on one PC. Or you can subscribe to M365 and now you can use it on 5 PCs - 1 of which may be a user's personal home computer, and 5 mobile devices or tablets. And 4 PCs is basically an IT department's allocation of computers for a person - they can have a desktop and a laptop. Their next desktop and laptop on replacement will consume 4 licenses temporarily and they have their home PC license. Then when the user returns their old desktop and laptop, those licenses are recycled and that user has 2 free computer licenses again.

And yes, Trump basically screwed over the US tech companies again because until then, thoughts to switch to altenratives were just murmurs and no one really did anything for decades. But now in the space of a year, everyone is switching to sovereign systems, or even worse, open-source. Everyone threatened it to get better licensing, but now it's actually happening.

No, they are not too depleted. Cars are very power hungry - the average 4 banger with 150 horsepower has over 100kW on tap - which would drain the vast majority of EV batteries in somehwere between half an hour and an hour. Meanwhile, the average house consumes about 1-1.5kW on average per day.

Even the largest grid scale storage only really capable of doing 1MW. And those installations are much larger than 10 EV batteries.

Just because a used EV battery can't supply its 100kW peak doesn't mean it can't be remade as a grid scale pack where it can peak at 10kW. That's why used EV batteries are being reused as grid scale or home scale batteries - 10kW is plenty for peak household in many instances, and even home scale batteries are usually only 20kWh or so (which is less than 1/5th of a large EV battery pack).

Used EV batteries simply can't supply the peak power anymore - their internal impedance increases as they grow old. But they can supply lower power levels for a number of years after. This could mean an EV battery pack gets 30+ years of use - the first half as an EV, the second half as grid scale or home scale battery.

It's actually why battery recycling is in its infancy - there's simply very little in the way of getting sufficient batteries for recycling.