ARM Community: Software Enablement - ARM Community

Thumb-2 can make use of the same conditional execution features that the ARM instruction set provides. For conditionally executing one or two instructions, this mechanism can provide code-size and performance benefits over the (more conventional) conditional branching mechanism.

I noted at the end of the last post in this series that this mechanism is not directly available to Thumb. Instead, Thumb-2 has an instruction — it — which can provide the same functionality as ARM conditional execution. In this article, I will describe the it instruction, and I will also explain a few caveats of condition-setting instructions in Thumb-2. Note that the it instruction is only available to Thumb-2, and so most of this article will not be relevant to the old Thumb instruction set 1.

This article introduces the shifting operations provided by NEON, and shows how they can be used to convert image data between commonly used color depths. Previous articles in this series: Part 1: Loads and Stores, Part 2: Dealing with Leftovers and Part 3: Matrix Multiplication.

A shift on NEON is very similar to shifts you may have used in scalar ARM code. The shift moves the bits in each element of a vector left or right. Bits that fall of the left or right of each element are discarded; they are not shifted to adjacent elements.

The amount to shift can be specified with a literal encoded in the instruction, or with an additional shift vector. When using a shift vector, the shift applied to each element of the input vector depends on the value of the corresponding element in the shift vector. The elements in the shift vector are treated as signed values, so left, right and zero shifts are possible, on a per-element basis.

A right shift operating on a vector of signed elements, indicated by the type attached to the instruct...

We have seen how to load and store data with NEON, and how to handle the leftovers resulting from vector processing. Let us move on to doing some useful data processing – multiplying matrices.

In this post, we will look at how to efficiently multiply four-by-four matrices together, an operation frequently used in the world of 3D graphics. We will assume that the matrices are stored in memory in column-major order – this is the format used by OpenGL-ES.

We start by examining the matrix mutiply operation in detail, by expanding the calculation, and identifying sub-operations that can be implemented using NEON instructions.

Notice that in the diagram, we multiply each column of the first matrix (in red) by a corresponding single value in the second matrix (blue) then add together the results for each element to give a column of results. This operation is repeated for each of the four columns in the result matrix.

...

In the first post on NEON about loads and stores we looked at transferring data between the NEON processing unit and memory. In this post, we deal with an often encountered problem: input data that is not a multiple of the length of the vectors you want to process. You need to handle the leftover elements at the start or end of the array - what is the best way to do this on NEON?

Using NEON typically involves operating on vectors of data from four to sixteen elements in length. Frequently, you will find that your array is not a multiple of that length, and you have to process those leftover elements separately.

For example, you want to load, process and store eight elements per iteration using NEON, but your array is 21 elements long. The first two iterations go well, but for the third, there are only five elements remaining to be processed. What do you do?

There are three ways to handle these leftovers. The methods vary in requirements, performance, and code size. They are listed below in order, with the fastest approach first.

If you can change the size of the arrays that you are processing, increase the length of the array to the next multiple of the vector size using padding elements. This allows you to read and write beyond the end of your data without corrupting ad...