RP2350

My guess is it's essentially free to knock up a RISC-V core (there have been a couple in tiny tapeout) while ARM are getting aggressive and raising the price of their cores - by building a part than can run with 2 ARM cores or 2 RISCV cores RPi are thinking about their future, not only do they have a way to keep costs down if ARM tries to raise them simply by abandoning ARM but they also have a bargaining chip on the table when that negotiation happens

There's going to be a pretty large number of code bases that when they get recompiled for RISC-V just work, that's going to be scary for ARM

The effort of interconnect to the rest of the chip isn't worth the tradeoff. https://news.ycombinator.com/item?id=41196841

Not that I am too familiar with the matter, but maybe the cores are not "real" independent cores, but share so much of the surrounding die that you have to make the choice. So it is not a quadcore where you can use only two, but a dualcore with an alternative architecture-part.

In a different thread someone said that the part implementing the opcodes are really small, so doing both isn't that much of a tradeoff.

If the cores share resources like ALU, routing, multiplier, etc. you can't run two cores that share same circuitry at same time.

Edit: Correction: Eben Upton said the cores share no logic. https://news.ycombinator.com/item?id=41196880.