VLIW and ILP

So how to meet these simultaneous requirements is a big question. The requirements can be simultaneously met by traditional processors, which are energy hogs. In fact, an egg can be fried on a server class chip, and they're not cheap.

What is instruction level parallelism (ILP) and why is useful?

The first requirement of high performance is met by instruction level parallelism that allows executing multiple instructions simultaneously to increase performance.

What is the Superscalar/VLIW processor design?

There are many forms of instruction level parallelism. The different forms of instruction level parallelism manifest as

1. Superscalars

2. In order processors

3. VLIW, and

4. In and out of order execution.

In order to understand the requirements for embedded processor design, let's do a quick overview of VLIW, superscalar, in and out of order processors and you should be able to figure out which architecture would be better for embedded systems in particular.

In VLIW, a single inspection is loaded but it consists of multiple server instruction that execute in parallel. Each server instruction is a simple RISK instruction and multiple server instructions are loaded at once.

Unlike VLIW, only one instruction is assured in a cycle in superscalar.

• Instructions that execute in the same order they are shared are known as in order processors.

• Instructions that execute out of order with regard to their issue are called as out of order processes.

So multiple possibilities exist. So the best possible processor design for embedded processor is it

• Superscalars, out of order execution style? Or

• Superscalars, in order execution style? Or

• VLIW, in order execution style? Or

• VLIW, out of order execution style?

You can see that, there have different choices, and different combinations of these choices.