Amusingly, what you insist upon calling a flaw is functioning exactly like the documentation that Intel provides for the instruction. Some software vendors wrote some pretty bad code that makes some assumptions that the documentation is pretty explicit in showing that those assumptions cannot be made.
I suspect you aren't really interested in the actual facts though.
It takes a pretty odd leap of logic to take a product that functions exactly as the documentation for that product says it does, have someone treat the product as if it functions differently than the documentation that has been provided, and then claim the product is at fault when it is performing exactly as specified.
Please, explain to us how you can call this a flaw when the instruction performs exactly as Intel claims it performs? An implementation that exactly fits the wording of the specs AMD published for the instruction when it made it. Sadly, the documentation of the spec allowed for multiple differing implementations that have material differences. That looks to me like a flaw in the original documentation.
If you would like to discuss specifics on how Intel's implementation of the instruction doesn't fit either their documentation of their implementation, or the wording of AMDs spec for the instruction, feel free. Otherwise, we can only assume that your claims are just an attempt at FUD.
I really don't expect that you'll actually respond.
Facebook
Twitter