Which of the following statements correctly describe differences between tensor cores and CUDA cores when working with GPUs used for machine learning?

A tensor core is a special processing area added to GPUs for matrix operations, while a CUDA core is a special processing area added to GPUs for floating point tasks.

A tensor core specifies one type of programming interface, while a CUDA core specifies another kind of programming interface accessible from the same GPU.

A tensor core uses high$-$speed memory for matrix manipulation, while a CUDA core relies on general memory for the same task.

A tensor core defines the number of $44$ matrix manipulations for a single clock cycle, while a CUDA core defines the number of single$-$precision multiply$-$accumulate operations in a single clock cycle.

A tensor core defines the number of low$-$cost operations in a single clock cycle, while a CUDA core defines the number of highcost operations in a single clock cycle.