China getting x86 license?

[parvadomus]

Or is just VIA joining another company? Getting help from China goverment looks like they might want the license to replace their Loongson processors.
Link

 

[ultimatebob]

Interesting. I guess that we now know what will be powering all of those sub $250 PC's that Microsoft just offered $15 Windows 8.1 licenses for:

http://news.cnet.com/8301-10805_3-5...-slash-windows-8.1-price-on-low-cost-devices/

For a sub $250 PC, even Intel's sub $50 Celeron and Atom chips are a stretch. Now, a mass produced $15 to $20 Chinese Via chip on the other hand... now you got parts for a $200 laptop with a little profit margin to spare.

 

[Homeles]

Interesting. I guess that we now know what will be powering all of those sub $250 PC's that Microsoft just offered $15 Windows 8.1 licenses for:

http://news.cnet.com/8301-10805_3-5...-slash-windows-8.1-price-on-low-cost-devices/

For a sub $250 PC, even Intel's sub $50 Celeron and Atom chips are a stretch. Now, a mass produced $15 to $20 Chinese Via chip on the other hand... now you got parts for a $200 laptop with a little profit margin to spare.

And a helluva lot crappier product, to boot. Last I checked, VIA didn't have anything competitive, even with AMD.

 

[Ertaz]

And a helluva lot crappier product, to boot. Last I checked, VIA didn't have anything competitive, even with AMD.

Compete against what? They will be the only game in town once they roll out.

 

[Homeles]

Compete against what? They will be the only game in town once they roll out.

Even if this were true, the moment a competitor caught on, they'd be squashed.

I'm not really sure what you're getting at, but China isn't some sort of big secret that only VIA knows about. The industry as a whole is well aware of China and emerging markets.

 

[SPBHM]

Via Nano should be a solid base for them to work and develop "good enough" cheap CPUs, if this means VIA is going to have a lot of resources incoming,

 

[Fox5]

Current VIA x86 offerings aren't cheaper than Atom or Brazos, at least in the US.

 

[Phynaz]

Technology != license.

 

[NTMBK]

Isn't VIA's x86 license tied to a specific FSB interface, which seriously limits what they can do? Or did I imagine that?

 

[Fox5]

I'd imagine AMD would be an easier purchase at this point anyway if China really wants an x86 license.

 

[norseamd]

there should be wasting their time on making their own design better to ramp up production and flood the world market. a national processor from china will make the resources of intel look like some amateur project. they can dabble in x86 if they want but they need to use royalty free if they want to be able to make anything and not have any meddling from western companies.

how is x86 under patent anyways anymore? patents are for 20 years right?

 

[Maximilian]

how is x86 under patent anyways anymore? patents are for 20 years right?

Probably all the extra stuff thats been added since x86 was created.

 

[ehume]

there should be wasting their time on making their own design better to ramp up production and flood the world market. a national processor from china will make the resources of intel look like some amateur project. they can dabble in x86 if they want but they need to use royalty free if they want to be able to make anything and not have any meddling from western companies.

how is x86 under patent anyways anymore? patents are for 20 years right?

Does the software baked into the design come under patent or copyright? And if instead it is a formal Trade Secret, it would not have a time limit.

 

[norseamd]

Does the software baked into the design come under patent or copyright? And if instead it is a formal Trade Secret, it would not have a time limit.

throwing loop holes into patent law are they now

 

[ehume]

throwing loop holes into patent law are they now

Loopholes are what lawyers do.

 

[jhu]

how is x86 under patent anyways anymore? patents are for 20 years right?

The ISA isn't patented (and can't be). The implementation of instructions are. All patents related to the original Pentium have expired. So if you want to make a 32-bit x86 processor exactly the way the Pentium is implemented, there's nothing Intel can do about it. However if you want to do x86 -〉RISC translation the way Pentium Pro does it, then you'll have to wait until 2015 when all patents related to the Pentium Pro have expired.

 

[norseamd]

The ISA isn't patented (and can't be). The implementation of instructions are. All patents related to the original Pentium have expired. So if you want to make a 32-bit x86 processor exactly the way the Pentium is implemented, there's nothing Intel can do about it. However if you want to do x86 -〉RISC translation the way Pentium Pro does it, then you'll have to wait until 2015 when all patents related to the Pentium Pro have expired.

so everything for the x86 has a expiration date?

 

[Headfoot]

The ISA isn't patented (and can't be). The implementation of instructions are. All patents related to the original Pentium have expired. So if you want to make a 32-bit x86 processor exactly the way the Pentium is implemented, there's nothing Intel can do about it. However if you want to do x86 -〉RISC translation the way Pentium Pro does it, then you'll have to wait until 2015 when all patents related to the Pentium Pro have expired.

The ISA could easily have been patented. Whether it was or not, I don't know. I would imagine it is or that some critical (related) portion of it is, otherwise people would be making x86 CPUs left and right. You can legally reverse engineer a trade secret so if it were only protected in that manner it would have been reverse engineered by now. It's copyright that does not allow the protection of ideas, only expression. I believe that's what you're getting at there. There is a special statutory copyright-like protection for lithographic masks which disallows copying on the layout level, at least in the US.

*none of this is legal advice

 

[jhu]

so everything for the x86 has a expiration date?

yes

 

[jhu]

The ISA could easily have been patented. Whether it was or not, I don't know.

Let's assume the x86 ISA was patented. That would have been back in 1978 for the 16-bit ISA and 1985 for the 32-bit ISA. Those patents would have expired already.

I would imagine it is or that some critical (related) portion of it is, otherwise people would be making x86 CPUs left and right.

Do you remember the mid-1990s? The players were Intel, AMD, IBM, Cyrix, Rise, NexGen, and IDT as I recall (perhaps there were others I'm missing). We're now down to Intel, AMD, and Via (Cyrix). The problem isn't just making a fast enough x86 chip, it's also getting people to buy the chip.

 

[NTMBK]

Let's assume the x86 ISA was patented. That would have been back in 1978 for the 16-bit ISA and 1985 for the 32-bit ISA. Those patents would have expired already.

But would anyone in their right mind want an x86 chip with no 64-bit support? With no SSE? Those are pretty fundamental these days- SSE was such a big improvement over x87 that many compilers automatically emit SSE instructions for scalar floating point, and anything compiled with those settings flat out wouldn't run on a "Pentium ISA" chip.

 

[norseamd]

what about opensparc

 

[Soulkeeper]

It's the "extensions" to the x86 ISA where the licensing/royalties come into play
ie: sse, avx, x86_64, etc.

 

[jhu]

It's the "extensions" to the x86 ISA where the licensing/royalties come into play
ie: sse, avx, x86_64, etc.

It's not just those particular extension implementations, it's also things like instruction decoding, branch prediction, power management, etc.

 

[Headfoot]

Yep, so easily encompassed by the statement "critical (related) portion." Like the article says though, the most powerful reason is likely the FTC deal and anti trust concerns after the AMD lawsuit

 

[Soulkeeper]

It's not just those particular extension implementations, it's also things like instruction decoding, branch prediction, power management, etc.

to be fair I did use "etc."