Nios® V Processor Reference Manual

Download PDF
ID 683632
Date 1/27/2025
Public
Document Table of Contents
Document Table of Contents x
1. Overview 2. Nios® V/c Processor 3. Nios® V/m Processor 4. Nios® V/g Processor 5. Nios V Processor Reference Manual Archives 6. Document Revision History for the Nios® V Processor Reference Manual
1. Overview x
1.1. Quartus® Prime Software Support
2. Nios® V/c Processor x
2.1. Processor Performance Benchmarks 2.2. Processor Pipeline 2.3. Processor Architecture 2.4. Programming Model 2.5. Core Implementation
2.3. Processor Architecture x
2.3.1. General-Purpose Register File 2.3.2. Arithmetic Logic Unit 2.3.3. Reset and Debug Signals 2.3.4. Instruction Cycles 2.3.5. Memory and I/O Organization 2.3.6. Error Correction Code (ECC)
2.3.5. Memory and I/O Organization x
2.3.5.1. Instruction and Data Buses 2.3.5.2. Address Map
2.3.5.1. Instruction and Data Buses x
2.3.5.1.1. Instruction Manager Port 2.3.5.1.2. Data Manager Port 2.3.5.1.3. Choosing a Suitable Interface
2.4. Programming Model x
2.4.1. Privilege Levels
2.5. Core Implementation x
2.5.1. Instruction Set Reference
3. Nios® V/m Processor x
3.1. Processor Performance Benchmarks 3.2. Processor Pipeline 3.3. Processor Architecture 3.4. Programming Model 3.5. Core Implementation
3.1. Processor Performance Benchmarks x
3.1.1. Pipelined 3.1.2. Non-pipelined
3.2. Processor Pipeline x
3.2.1. Pipelined Architecture 3.2.2. Non-pipelined Architecture
3.3. Processor Architecture x
3.3.1. General-Purpose Register File 3.3.2. Arithmetic Logic Unit 3.3.3. Reset and Debug Signals 3.3.4. Control and Status Registers 3.3.5. Instruction Cycles 3.3.6. Trap Controller (CLINT) 3.3.7. Memory and I/O Organization 3.3.8. RISC-V based Debug Module 3.3.9. Error Correction Code (ECC)
3.3.6. Trap Controller (CLINT) x
3.3.6.1. Related Control and Status Register 3.3.6.2. Direct Mode 3.3.6.3. Vectored Mode
3.3.6.1. Related Control and Status Register x
3.3.6.1.1. Machine Status Register (mstatus) 3.3.6.1.2. Machine Trap-Vector Base-Address Register (mtvec) 3.3.6.1.3. Machine Interrupt Register (mip and mie) 3.3.6.1.4. Machine Exception Program Counter Register (mepc) 3.3.6.1.5. Machine Cause Register (mcause) 3.3.6.1.6. Machine Trap Value Register (mtval)
3.3.6.2. Direct Mode x
3.3.6.2.1. Hardware Implementation 3.3.6.2.2. Software Implementation
3.3.6.3. Vectored Mode x
3.3.6.3.1. Hardware Implementation 3.3.6.3.2. Software Implementation
3.3.7. Memory and I/O Organization x
3.3.7.1. Instruction and Data Buses 3.3.7.2. Address Map
3.3.7.1. Instruction and Data Buses x
3.3.7.1.1. Instruction Manager Port 3.3.7.1.2. Data Manager Port 3.3.7.1.3. Choosing a Suitable Interface
3.3.8. RISC-V based Debug Module x
3.3.8.1. Debug Module 3.3.8.2. Abstract Commands 3.3.8.3. Program Buffer 3.3.8.4. Debug Mode 3.3.8.5. Hardware Trigger Module 3.3.8.6. JTAG Debug Transport Module 3.3.8.7. Nios® V Processor Implementation
3.3.8.1. Debug Module x
3.3.8.1.1. Reset/Halt/Run Control 3.3.8.1.2. Hart States 3.3.8.1.3. Debug Module Register
3.3.8.2. Abstract Commands x
3.3.8.2.1. Access Register Command 3.3.8.2.2. Using Access Register Command
3.3.8.3. Program Buffer x
3.3.8.3.1. Using Program Buffer
3.3.8.4. Debug Mode x
3.3.8.4.1. Enter Debug Mode 3.3.8.4.2. Exit Debug Mode 3.3.8.4.3. Core Debug Registers
3.3.8.5. Hardware Trigger Module x
3.3.8.5.1. Trigger Registers 3.3.8.5.2. Type of Address/Data Match Triggers 3.3.8.5.3. Using Triggers
3.3.8.6. JTAG Debug Transport Module x
3.3.8.6.1. JTAG DTM Registers 3.3.8.6.2. Accessing Debug Module Interface
3.4. Programming Model x
3.4.1. Privilege Levels 3.4.2. Control and Status Registers (CSR) Mapping
3.4.1. Privilege Levels x
3.4.1.1. Machine Mode (M-mode) 3.4.1.2. Debug Mode (D-mode)
3.4.2. Control and Status Registers (CSR) Mapping x
3.4.2.1. Control and Status Register Field 3.4.2.2. Memory Mapped Registers Fields
3.5. Core Implementation x
3.5.1. Instruction Set Reference
4. Nios® V/g Processor x
4.1. Processor Performance Benchmarks 4.2. Processor Pipeline 4.3. Processor Architecture 4.4. Programming Model 4.5. Core Implementation
4.3. Processor Architecture x
4.3.1. General-Purpose Register File 4.3.2. Arithmetic Logic Unit 4.3.3. Multipy and Divide Units 4.3.4. Floating-Point Unit 4.3.5. Custom Instruction 4.3.6. Instruction Cycles 4.3.7. Reset and Debug Signals 4.3.8. Control and Status Registers 4.3.9. Trap Controller (CLINT) 4.3.10. Memory and I/O Organization 4.3.11. RISC-V based Debug Module 4.3.12. Error Correction Code (ECC) 4.3.13. Branch Prediction 4.3.14. Lockstep Module
4.3.4. Floating-Point Unit x
4.3.4.1. IEEE 754 Exception Conditions 4.3.4.2. Floating Point Operations
4.3.4.2. Floating Point Operations x
4.3.4.2.1. Floating Point Classification
4.3.9. Trap Controller (CLINT) x
4.3.9.1. Related Control and Status Register 4.3.9.2. Direct Mode 4.3.9.3. Vectored Mode
4.3.9.1. Related Control and Status Register x
4.3.9.1.1. Machine Status Register (mstatus) 4.3.9.1.2. Machine Trap-Vector Base-Address Register (mtvec) 4.3.9.1.3. Machine Interrupt Register (mip and mie) 4.3.9.1.4. Machine Exception Program Counter Register (mepc) 4.3.9.1.5. Machine Cause Register (mcause) 4.3.9.1.6. Machine Trap Value Register (mtval) 4.3.9.1.7. Machine Second Trap Value Register (mtval2)
4.3.9.2. Direct Mode x
4.3.9.2.1. Hardware Implementation 4.3.9.2.2. Software Implementation
4.3.9.3. Vectored Mode x
4.3.9.3.1. Hardware Implementation 4.3.9.3.2. Software Implementation
4.3.10. Memory and I/O Organization x
4.3.10.1. Instruction and Data Buses 4.3.10.2. Address Map 4.3.10.3. Cache Memory 4.3.10.4. Tightly Coupled Memory
4.3.10.1. Instruction and Data Buses x
4.3.10.1.1. Instruction Manager Port 4.3.10.1.2. Data Manager Port
4.3.10.3. Cache Memory x
4.3.10.3.1. Instruction Cache 4.3.10.3.2. Data Cache 4.3.10.3.3. Configurable Cache Memory 4.3.10.3.4. Bypassing Cache (Peripheral Region) 4.3.10.3.5. Using Cache Memory Effectively
4.3.10.4. Tightly Coupled Memory x
4.3.10.4.1. Instruction and Data Tightly Coupled Memory 4.3.10.4.2. Accessing Tightly Coupled Memory 4.3.10.4.3. Using Tightly Coupled Memory Effectively
4.3.11. RISC-V based Debug Module x
4.3.11.1. Debug Module 4.3.11.2. Abstract Commands 4.3.11.3. Program Buffer 4.3.11.4. Debug Mode 4.3.11.5. Hardware Trigger Module 4.3.11.6. JTAG Debug Transport Module 4.3.11.7. Nios® V Processor Implementation
4.3.11.1. Debug Module x
4.3.11.1.1. Reset/Halt/Run Control 4.3.11.1.2. Hart States 4.3.11.1.3. Debug Module Register
4.3.11.2. Abstract Commands x
4.3.11.2.1. Access Register Command 4.3.11.2.2. Using Access Register Command
4.3.11.3. Program Buffer x
4.3.11.3.1. Using Program Buffer
4.3.11.4. Debug Mode x
4.3.11.4.1. Enter Debug Mode 4.3.11.4.2. Exit Debug Mode 4.3.11.4.3. Core Debug Registers
4.3.11.5. Hardware Trigger Module x
4.3.11.5.1. Trigger Registers 4.3.11.5.2. Type of Address/Data Match Triggers 4.3.11.5.3. Using Triggers
4.3.11.6. JTAG Debug Transport Module x
4.3.11.6.1. JTAG DTM Registers 4.3.11.6.2. Accessing Debug Module Interface
4.3.12. Error Correction Code (ECC) x
4.3.12.1. ECC Interface 4.3.12.2. ECC Error Injection 4.3.12.3. Affected CSR during ECC Event
4.4. Programming Model x
4.4.1. Privilege Levels 4.4.2. Control and Status Registers (CSR) Mapping (CLINT)
4.4.1. Privilege Levels x
4.4.1.1. Machine Mode (M-mode) 4.4.1.2. Debug Mode (D-mode)
4.4.2. Control and Status Registers (CSR) Mapping (CLINT) x
4.4.2.1. Control and Status Register Field (CLINT) 4.4.2.2. Memory Mapped Registers Fields
4.5. Core Implementation x
4.5.1. Instruction Set Reference
1. Overview
2. Nios® V/c Processor
3. Nios® V/m Processor
4. Nios® V/g Processor
5. Nios V Processor Reference Manual Archives
6. Document Revision History for the Nios® V Processor Reference Manual

4.3.4.2. Floating Point Operations

The table below provides a detailed summary of the FPU operations.
Table 93.  Floating Point Operation Summary
Category Operation Cycles2 Result Subnormal Rounding3 GCC Inference
Arithmetic FDIV.S 14 a ÷ b Flush-to-0 RNE a / b
FSUB.S 1 a - b Flush-to-0 RNE a - b
FADD.S 1 a + b Flush-to-0 RNE a + b
FMUL.S 2 a x b Flush-to-0 RNE a * b
FSQRT.S 12 √a Flush-to-0 Faithful 4 sqrt(a)
FMIN.S 2 (a < b) ? a : b Supported RNE fminf()
FMAX.S 2 (a < b) ? b : a Supported RNE fmaxf()
Fused Arithmetic5 FMADD.S 3 (a x b) + c Flush-to-0 RNE (a * b) + c
FMSUB.S 3 (a x b) – c Flush-to-0 RNE (a * b) - c
FNMSUB.S 3 -(a x b) + c Flush-to-0 RNE -(a * b) + c
FNMADD.S 3 -(a x b) - c Flush-to-0 RNE -(a * b) - c
Conversion FCVT.S.W / FCVT.S.WU 3 int_to_float(a) Supported None Casting
FCVT.W.S / FCVT.WU.S 3 float_to_int(a) Supported Round towards Zero Casting
Round to Nearest, ties to Max Magnitude roundf(a)
Compare FLT.S 1 (a < b) ? 1 : 0 Flush-to-0 RNE a < b
FLE.S 1 (a ≤ b) ? 1 : 0 Flush-to-0 RNE a <= b
FEQ.S 1 (a = b) ? 1 : 0 Flush-to-0 RNE a == b
Sign Injection FSGNJN.S (FNEG.S) 1 -a Supported RNE -a
FSGNJX.S (FABS.S) 1 |a| Supported RNE fabsf(a)
Classification FCLASS.S 2 Refer to topic Floating Point Classification. Supported None fpclassify(a)
Note: Assume a, b, and c as single-precision floating point values. Nios® V Processor Fused Arithmetic has a rounding stage between the multiplier and addition.
The following list describes the header in the table above:
  • Operation —Provides the name of the floating-point operation. The names match the names of the corresponding RISC-V floating-point instructions.
  • Cycle —Specifies the number of cycles it takes to execute the instruction.
  • Result—Describes the computation performed by the operation.
  • Subnormal—Describes how the operation treats subnormal inputs and subnormal outputs. Subnormals are numbers with a magnitude less than approximately 1.17549435082e-38.
  • Rounding —Describes how the FPU rounds the result.
  • GCC Inference—Shows the C code from which GCC infers the instruction operation.

When optimizing a Floating Point Unit (FPU) within a processor, the FSQRT (floating-point square root) and FDIV (floating-point division) operations are critical components. Because of their complexity and longer execution time, they can influence the maximum frequency (Fmax) of the FPU. Refer to the following guidelines to help you decide whether to enable or disable the FSQRT and FDIV.

Table 94.  Recommendations to Enable or Disable the FSQRT and FDIV
Action Guidelines
Enable Enable the FQSRT and FDIV:
  • If your application frequently performs floating-point square root and floating-point division calculations
  • To prioritize the number of instructions executed per cycle
Disable Disable the FQSRT and FDIV:
  • If your application rarely or never uses floating-point square root and floating-point division calculations
  • To prioritize logic utilization and achieve a higher Fmax
  • If you're using the Nios® V processor in a MAX® 10 FPGA

In summary, deciding whether to enable or disable FSQRT and FDIV in a Floating Point Unit (FPU) is a choice between two main goals. You can aim for a higher maximum frequency (Fmax) and lower logic usage, or you can focus on improving performance for floating-point operations. This decision should depend on the specific needs and limits of your application.


Section Content
Floating Point Classification

Related Information
2 Preliminary results.
3 Round-to-Nearest, ties to Even (RNE).
4 Faithful rounding has a maximum error of 1 Unit of Least Precision (ULP) as compared to the 0.5 ULP in RNE. Faithful rounding is employed to save area and reduce the latency of FSQRT.S.
5 GCC toolchain infers Fused Arithmetic when the optimization level is -O3 or higher.
Level Two Title