ARM.SoC.Architecture

The coprocessor takes cpa low but leaves cpb high the

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: purpose. Since the data transfer is not interruptible once it has started, coprocessors should limit the maximum transfer length to 16 words (the same as a maximum length load or store multiple instruction) so as not to compromise the ARM's interrupt response. Examples and exercises 103 Pre-emptive execution A coprocessor may begin executing an instruction as soon as it enters its pipeline so long as it can recover its state if the handshake does not ultimately complete. All activity must be id em potent (repeatable with identical results) up to the point of commitment. 4.6 Examples and exercises Why does r15 give pc + 8 in the first cycle of an instruction and pc + 12 in subsequent cycles on an ARM7? This is the ARM pipeline being exposed to the programmer. Referring back to Figure 4.2 on page 77, we can see that the pc value was incremented once when the current instruction (' 1' in the figure below) was fetched and once when its successor ('2') was fetched, giving pc + 8 at the start of the first execute cycle. During the first execute cycle a third instruction ('3') is fetched, giving pc + 12 in all subsequent execute cycles. While multi-cycle instructions interrupt the pipeline flow they do not affect this aspect of the behaviour. An instruction always fetches the next-instruction-but-one during its first execute cycle, so r15 always progresses from pc + 8 at the start of the first execute cycle to pc + 12 at the start of the second (and subsequent) execute cycle(s). (Note that other ARM processors do not share this behaviour, so it should never be relied upon when writing ARM programs.) Example 4.1 Exercise 4.1.1 Draw a pipeline flow diagram along the lines of the one above to illustrate the timing of an ARM branch instruction. (The branch target is computed in the first execute cycle of the instruction and issued to memory in the following cycle.) How many execute cycles are there after the branch target calculation and before the instruction at the branch target is ready to execute? What does the processor use these execute cycles for? Exercise...
View Full Document

This document was uploaded on 10/30/2011 for the course CSE 378 380 at SUNY Buffalo.

Ask a homework question - tutors are online