Using registers in Miips assembly language

Question

I'm trying to understand how miips instructions work, and looking at the bne instruction. I think I understand it at a high level, I just have a few questions.

For example, let's say I use the command bne $R0, $R10, X. this is the same as saying if $R0 does not equal $R10, jump to address X (where address X is calculated as PC+4+4*X, and X is some integer).

• When you look at a standard register table (I found several on Google that were identical, I assumed they are standard), there are 31 registers. So when you specify a value for X, does that mean X is limited to values such that 4*X < 31? (or more restricted since several registers in the range 0-31 are reserved for system use?), or is that going to be an address in memory, and not a register?

• I know that if I use $t0-$t7 I am using registers 8-15, and $s0-$s7 uses registers 17-23, but where is $R0-$Rk (for some integer k<32, I assume) located? Is that the same 32 registers I would use for $t0, $s5, etc? For example, are $R10, register 10, and $t2 the same thing?

Answer 1

When you look at a standard register table (I found several on Google that were identical, I assumed they are standard), there are 31 registers.

This is not exactly correct. There are 32 general-purpose registers on MIPS, counting r0, as well as a number of special-purpose registers such as pc, lo and hi, and floating-point registers.

So when you specify a value for X, does that mean X is limited to values such that 4*X < 31? (or more restricted since several registers in the range 0-31 are reserved for system use?), or is that going to be an address in memory, and not a register?

Neither. The value for X is specified as a constant within the instruction. The encoding looks like:

BNE: 0001 01ss ssst tttt  iiii iiii iiii iiii

The encoding for this instruction begins with a 6-bit header (000101), uses 5 bits for each of the source and destination registers (s and t), and uses the remaining 16 bits of the instruction for a signed 16-bit offset (i). This offset is treated as a count of instructions relative to the next PC - as a result, you can jump up to 64 KB in each direction.

I know that if I use $t0-$t7 I am using registers 8-15, and $s0-$s7 uses registers 17-23, but where is $R0-$Rk (for some integer k<32, I assume) located? Is that the same 32 registers I would use for $t0, $s5, etc? For example, are $R10, register 10, and $t2 the same thing?

Yes — some MIPS general-purpose registers have multiple names. r10 and "register 10" both refer to the register by its location, whereas t2 refers to it by its conventional use as a Temporary register.