SPARC Architecture

Based on "The SPARC Architecture Manual: Version 8", SPARC International

Introduction to SPARC Architecture

SPARC is a RISC computer architecture. RISC stands for "Reduced Instruction Set Computer".
Such computers use few instructions with a fixed size, which is usually the size of the word in the architecture.

A word is the most common number of bytes used to represent a piece of information.

The few uniform size instructions used in that type of architecture allow for building simpler and faster machines.
The simplicity comes from the need of fewer hardware units to implement the instructions.
The speed gain comes from the use of pipelining - executing several instructions at their different stages of execution
simultaneously.

Register Files

Registers are memory cells located on the CPU (Central Processing Unit) chip.
In this architecture there are 32 general purpose integer registers and 32 floating point registers.
The small number of registers makes access to these is very fast, which is important for speeding up program execution.

Integer General Purpose Register File

The 32 general purpose integer registers are named from %r0 to %r31. They also have aliases according to the different role they play
in a program's execution:

%g0 .. %g7 global registers; same as %r0 .. %r7
%o0 .. %o7 out      registers; same as %r8 .. %r15
%l0 .. %l7    local   registers; same as %r16 .. %r23
%i0 .. %i7      in        registers; same as %r24 .. %r31

%r0 is always 0

Floating-Point Register File

Floating-Point registers have the names: %f0 .. %f31. They are used to hold real numbers.
When a bigger number is needed to be stored, they can be used in pairs and groups of four.

Special Symbol Names

%psr Processor State Register
The processor state register has the following structure:

Condition Code Bits

Condition code bits are part of %psr. They are set during execution of some assembly instructions. Their contents can be checked afterwards and acted upon,
N is set if result of arithmetic operation (cc version of an instruction) has a most significant bit 1.
Z is set if result of arithmetic operation (cc version of an instruction) is zero.

%wim Window Invalid Mask Register
%tbr Trap Base Register
%y Y register

The %y register is used for multiplication and division. It holds the 32 most significant bits of the dividend in division and the 32 most significant bits of the result in multiplication.

%fsr Floating-Point State Register

Floating Point Condition Codes

E 0 if freg_rs1=freg_rs2
L 1 if freg_rs1<freg_rs2
G 2 if freg_rs1>freg_rs2
U 3 if freg_rs1 ? freg_{rs2 (unordered -
one or both of the numbers were NaN )}

%csr    Coprocessor-State Register
%fq      Floating-Point Queue
%cq     Coprocessor Queue

%hi

Unary operator which extracts high 22 bits of its operand.

%lo

Unary operator which extracts low 10 bits of its operand.