Intel Announces 10nm Is Shipping

[InvalidError]

Intel has yearly gross income comparable to several small countries, so a few million or even a billion in R&D might sting, but it definitely wouldn't hurt them in any impactful way.

If Intel halted all production and sales of CPUs until all plausibly exploitable flaws were fixed, they'd also lose tens of billions in sales in the meantime and billions more in repeat sales due to the R&D cycle getting extended by several months just to do the extra research to identify any possible new exploits that may have crept up in the new design.

Also, don't forget that people are ultimately going to have to pay for all this extra overhead and those lost opportunities. I'm hopefully being sarcastic here when I say: "hello $200 Pentiums."

 

[Darkbreeze]

Yeah, that's the other, somewhat less obvious side of problems like this. Companies rarely EVER "take it in the xxx", they just pass the cost on to consumers. Same with utility companies. Same with cable services. Seems like the only kind of manufacturers that can't get away with doing this are automobile manufacturers. If we can just figure out what it is we do that forces them to eat the costs when they screwe up and apply that to these others, we as consumers would be a lot better off.

I think it's MOSTLY just the simple fact that when it comes to vehicles we have many choices and when it comes to CPUs, cable tv, internet access or power and gas, we have few choices, so they know they can get away with it. There definitely needs to be a third x86 CPU manufacturer come on the scene. Seems like when there is at least three of anything, competition tends to seriously benefit the consumer.

TBH, I don't think you are being unrealistic when you say "hello $200.00 Pentiums". I mean, if you look at the price of a 16GB memory kit this time last year, compared to now, it makes it look a whole lot more PROBABLE than just possible.

 

[InvalidError]

There definitely needs to be a third x86 CPU manufacturer come on the scene.

Personally, I'd say x86 simply needs to die so we can start fresh with an ISA that doesn't have 40 years of backwards-compatibility kludges and bugs-come-essential-features adding to the regression testing burden. The more convoluted an instruction set and hardware architecture are, the greater the likelihood of unforeseeable bugs.

Also, now that all major platforms are pushing for platform-agnostic applications with local native re-compiling (Java on nearly everything, ART/Dalvik-flavored Java on Android, .NET/MSIL on Windows and Linux, Swift on IOS, etc.) instead of straight native binaries, the underlying CPU architecture is becoming irrelevant to a growing segment of personal and office computing. Now would be a good time for the industry to come up with a solid and clean open standard ISA for modern-day computing.

 

[Darkbreeze]

Whatever you just said, and I think I understood half of it (ok, maybe a bit more than that), sounds appropriate to me. A clean slate is rarely a bad thing. I don't know that it's REALISTIC given that even with the current atmosphere these companies don't seem to feel compelled to look for new solutions and are merely happy to open new boxes of band aids, but I certainly agree that something better would be welcome.