I imagine this could be useful for iterated crypto use cases, where you need to brute force hash(hash(hash(…))) as quickly as possible, you can’t parallelize by

I imagine this could be useful for iterated crypto use cases, where you need to brute force hash(hash(hash(…))) as quickly as possible, you can’t parallelize by design, and some hardware failure is acceptable for extra speed. Perhaps this is a way for Intel to either avoid their inventory being eaten up by miners solving these problems, or to upsell to other CPU lines?