47
edits
Changes
Jump to navigation
Jump to search
Basics and psq_l
Paired singles are a unique part of the Gekko/[[Broadway]] processors used in the Gamecube and Wii. They provide fast vector math by keeping two single-precision floating point numbers in a single floating pointer register, and multiplying across between registers. This page will demonstrate how these instructions are to be used.
== Quantization and Dequantization ==
For conversion from non-floats, in order to allow for greater flexibility, there is a form of scaling implemented. All quantization is controlled by the GQRs (Graphics Quantization Registers). The GQRs are 32bit registers containing the conversion types and scaling factors for storing and loading. (During loading, it dequantizes. During storing, it quantizes.)
== Loading and Storing ==
To load and store Paired-singles, one must use the psq_l and psq_st instructions respectively, or one of their variants.
=== psq_l ===
psq_l frD, d(rA), W, I
This instruction dequantizes values from the memory address in d+rA and puts them into PS0 and PS1 in frD. If W is 1, however, it only dequantizes one number, and places that into PS0. PS1 is loaded with 1.0 always when W is 1. I specifies the GQR to use for dequantization parameters. The two numbers read from the memory are directly after each other, regardless of size (for example, if the GQR specified to load as a u16, you would have d+rA point to a two-element array of u16s)
== Quantization and Dequantization ==
For conversion from non-floats, in order to allow for greater flexibility, there is a form of scaling implemented. All quantization is controlled by the GQRs (Graphics Quantization Registers). The GQRs are 32bit registers containing the conversion types and scaling factors for storing and loading. (During loading, it dequantizes. During storing, it quantizes.)
== Loading and Storing ==
To load and store Paired-singles, one must use the psq_l and psq_st instructions respectively, or one of their variants.
=== psq_l ===
psq_l frD, d(rA), W, I
This instruction dequantizes values from the memory address in d+rA and puts them into PS0 and PS1 in frD. If W is 1, however, it only dequantizes one number, and places that into PS0. PS1 is loaded with 1.0 always when W is 1. I specifies the GQR to use for dequantization parameters. The two numbers read from the memory are directly after each other, regardless of size (for example, if the GQR specified to load as a u16, you would have d+rA point to a two-element array of u16s)