Replies to post #112478 on Intel Corp (INTC)

[ibc]

I understand that you lose efficiency/performance in this scenario.
The question is, how much do you give up vs. how much do you gain by attracting more customers to fill the fabs?
From your hyperbole, it sounds like any kind of compromise is fatal.

Does this also mean that it's unlikely that Apple would use Intel fabs for AX processors since the different design methodologies would be too difficult to overcome?

the focus is being expanded to include low power and low cost computing, graphics, and wireless. Intel is only weak in areas of rapid change, high fragmentation

Hold on now... Fragmentation is happening now inside the SOC. Intel's superior process and design methodologies have been tuned for MPUs over 20 years. The more heterogeneous nature of computers (GPU, DSP, wireless etc...) have conflicting requirements.

One simple example is that GPUs can tolerate more latency vs. CPUs
If I might overly simplify it - wide and slow is preferrable for GPU vs. narrow and fast for CPU.
This has implications for your optimal interconnect and transistors - for example, the FP-SIMD density in a GPU is higher than the FP-SIMD in a CPU.

You can't be all things to all people, even on the same die.
Somewhere, there's a trade-off being made.