Beastian (noemail.delete@this.aol.com) on April 19, 2019 9:05 am wrote:
> Does anyone think that the upcoming expiry of patents on the core x86-64 instruction set
> which includes SSE2, maybe sooner than 2023 (as the Athlon 64 was introduced in 2003) would
> open the flood gates for third party implementations of compatible processors?

The "x86-64" patents are only one part of the patent puzzle.

There is a huge mass of patents, not all expiring soon, on CPU implementation - much of which isn't specific to x86, which can also be used to keep competitors out.

Just do a patent search for Intel as the assignee and ask yourself if you can easily implement a high performance CPU without using any of those techniques.

Now, it is my opinion that almost everyone in the CPU space regularly and mutually violate each others implementation patents, but it is rarely litigated because (a) the implementation is often invisible and thus hard to prosecute and (b) mutually assured destruction.

Exceptions who actually do litigate frequently include WARF who, by virtue of not making CPUs themselves don't need to worry about (b), although it is not exactly clear to be how they deal with (a), i.e., how they know a particular patented technique is being used. Perhaps observation of behavior is enough, for patents written "behaviorally" rather than wrt a specific hardware implementation.