# MIPS Specification ## Programmer-visible state | Resource | Width | Count | |---|---:|---:| | General-purpose registers | 32-bit | 32 | | PC | 32-bit | 1 | | HI | 32-bit | 1 | | LO | 32-bit | 1 | | CP0 registers | implementation-defined control state | implementation-dependent | GPR 0 is architecturally constant zero. Writes to it do not produce a persistent nonzero value. ## Instruction formats Classic MIPS uses fixed 32-bit instructions with R, I, and J encoding families. SLeeLa keeps instruction decode separate from execution so fields can feed register selection, immediate generation, ALU control, memory control, and branch control. ## Execution The model includes: - register-register ALU operations; - immediate operations; - loads and stores; - conditional branches; - jumps and jump-and-link; - shifts; - multiply/divide with HI/LO; - system-control operations through CP0; - exception and interrupt entry. ## Pipeline The base model represents the classic five conceptual stages: 1. IF 2. ID 3. EX 4. MEM 5. WB This is a functional timing abstraction, not a claim that every MIPS implementation physically used exactly five stages. ## Branch delay Classic MIPS executes the instruction in the branch delay slot. citeturn0search20 ## Privilege and CP0 CP0 contains privileged system state including status and exception-related state. MIPS documentation shows the CP0 Status register controlling interrupt, exception, privilege, and related operating state. citeturn0search19 ## Cache/MMU/DMA boundary Cache and TLB organization are implementation-dependent. The base ISA description does not invent a particular cache size or TLB entry count. DMA is modeled as an external bus master unless a specific MIPS SoC/core documents otherwise.