In RaspberryPI architecture (Broadcom BM2328 SoC), we can see that ARM core physical memory is mapped to VideoCore memory through a second MMU (the first maps ARM virtual to physical memory). What is the use of this configuration? Doesn't the use of 2 MMU's increase the latency?


One MMU maps ARM virtual addresses (used by Linux userland and the kernel) into ARM physical addresses.

The other MMU maps ARM physical addresses into bus addresses (to access the peripherals such as the gpio, DMA, SPI, UARTs etc. etc.).

I doubt that either MMU adds any latency.


| improve this answer | |
  • Thanks, but does it mean that VideoCore GPU is what really manages every memory accesses? – Pierre Jan 10 '15 at 10:19
  • @Pierre I do not know. I don't think so. I believe it's only the peripheral address space which needs to be accessed via the VC GPU MMU. You'd need to find someone familiar with ARM architecture - and that is not me. – joan Jan 10 '15 at 10:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.