Competitors will always find ways of attacking each other.
In reality in any kind of engineering in the real world there's always trade offs of different approaches. These choices are often made with cost being one of the factors.
AMD's trade of is a one size fits all die that can be used in a lot of products quite effectively. And on a relative basis, IPC wise it's really not that far. As a result it can offer this at an attractive price.
Intel on the other hand is making a lot of different dies which comes with lower cross core communication latencies. Also, their more expensive / mature process allows for higher Mhz clocks. This comes with a higher premium MSPR for products in the same class.
People can figure out what is more important to their use cases and we can all argue about the benefits/downsides of each approach.
I would say the market segmentation with AVX512 (different chips supporting different features) is somewhat maddening. Plus a limited use case of HPC software make effective use of that. Most does not.
In reality in any kind of engineering in the real world there's always trade offs of different approaches. These choices are often made with cost being one of the factors.
AMD's trade of is a one size fits all die that can be used in a lot of products quite effectively. And on a relative basis, IPC wise it's really not that far. As a result it can offer this at an attractive price.
Intel on the other hand is making a lot of different dies which comes with lower cross core communication latencies. Also, their more expensive / mature process allows for higher Mhz clocks. This comes with a higher premium MSPR for products in the same class.
People can figure out what is more important to their use cases and we can all argue about the benefits/downsides of each approach.
I would say the market segmentation with AVX512 (different chips supporting different features) is somewhat maddening. Plus a limited use case of HPC software make effective use of that. Most does not.