slli shifts bits to the left by a fixed amount. It is the same operation as sll — slide the bits left, fill the right with 0s, drop what falls off the left, doubling the value for each place moved (the sll page explains the mechanics) — but the number of places to shift is written into the instruction as a constant rather than read from a register. The extra i means immediate, the term for a constant baked into the instruction.

The form is slli rd, rs1, amount, where amount is a fixed number from 0 to 31. So slli t1, t0, 3 always shifts t0 left by 3 places, multiplying it by 8.

This is one of the most common instructions in real programs, because of how arrays work in memory. Each item in an array of 32-bit values takes 4 bytes, so to find item number i you multiply i by 4 — and multiplying by 4 is shifting left by 2. The recurring pattern is: shift the index to turn it into a byte offset, add the array's starting address, then load. You will write it constantly, and spotting the shift-add-load trio makes other people's assembly far easier to read.

Use slli whenever the shift distance is known as you write the code, which is most of the time; use the register form sll only when the amount is computed while running. The example shifts 5 left by 3, giving 5 times 8, or 40.