Intel® High Level Synthesis Compiler Pro Edition: Reference Manual

Download PDF
ID 683349
Date 10/04/2021
Public

A newer version of this document is available. Customers should click here to go to the newest version.

Document Table of Contents
Document Table of Contents x
1. Intel® HLS Compiler Pro Edition Reference Manual 2. Compiler 3. C Language and Library Support 4. Component Interfaces 5. Component Memories (Memory Attributes) 6. Loops in Components 7. Component Concurrency 8. Arbitrary Precision Math Support 9. Component Target Frequency 10. Systems of Tasks 11. Libraries 12. Advanced Hardware Synthesis Controls 13. Intel® High Level Synthesis Compiler Pro Edition Reference Summary A. Advanced Math Source Code Libraries B. Supported Math Functions C. Cyclone® V Restrictions D. Intel® HLS Compiler Pro Edition Reference Manual Archives E. Document Revision History of the Intel® HLS Compiler Pro Edition Reference Manual
2. Compiler x
2.1. Intel® HLS Compiler Pro Edition Command Options 2.2. Using Libraries in Your Component 2.3. Compiler Interoperability 2.4. Intel® HLS Compiler Hardware Model
2.3. Compiler Interoperability x
2.3.1. Compiling Your Testbench and Component Separately
3. C Language and Library Support x
3.1. Supported C and C++ Subset for Component Synthesis 3.2. C and C++ Libraries 3.3. Templated and Overloaded Functions 3.4. Intel® HLS Compiler Pro Edition Compiler-Defined Preprocessor Macros
3.3. Templated and Overloaded Functions x
3.3.1. Templated Functions 3.3.2. Overloaded Functions 3.3.3. Function Name Mapping
4. Component Interfaces x
4.1. Component Invocation Interface 4.2. Avalon® Streaming Interfaces 4.3. Pipes 4.4. Avalon® Memory-Mapped Host Interfaces 4.5. Agent Interfaces 4.6. Stable Component Parameters 4.7. Global Variables 4.8. Structs in Component Interfaces 4.9. Reset Behavior
4.1. Component Invocation Interface x
4.1.1. Scalar Parameters 4.1.2. Pointer and Reference Parameters 4.1.3. Interface Definition Example: Component Invocation Interface Control Attributes 4.1.4. Interface Definition Example: Component with Both Scalar and Pointer Arguments
4.3. Pipes x
4.3.1. The pipe Class and Its Use
4.4. Avalon® Memory-Mapped Host Interfaces x
4.4.1. Memory-Mapped Host Testbench Constructor 4.4.2. Implicit and Explicit Examples of Describing a Memory Interface 4.4.3. Avalon® Memory-Mapped Host Interfaces and Load-Store Units
4.4.3. Avalon® Memory-Mapped Host Interfaces and Load-Store Units x
4.4.3.1. Load-Store Unit Types 4.4.3.2. Memory-Access Coalescing and Load-Store Units
4.5. Agent Interfaces x
4.5.1. Control and Status Register (CSR) Agent 4.5.2. Agent Memories
5. Component Memories (Memory Attributes) x
5.1. Static Variables
6. Loops in Components x
6.1. Loop Initiation Interval (ii Pragma) 6.2. Loop-Carried Dependencies (ivdep Pragma) 6.3. Loop Coalescing (loop_coalesce Pragma) 6.4. Loop Unrolling (unroll Pragma) 6.5. Loop Concurrency (max_concurrency Pragma) 6.6. Loop Iteration Speculation (speculated_iterations Pragma) 6.7. Loop Pipelining Control (disable_loop_pipelining Pragma) 6.8. Loop Interleaving Control (max_interleaving Pragma) 6.9. Loop Fusion
6.9. Loop Fusion x
6.9.1. Loop Fusion Control (loop_fuse Pragma) 6.9.2. Loop Fusion Exemption (nofusion pragma)
7. Component Concurrency x
7.1. Serial Equivalence within a Memory Space or I/O 7.2. Concurrency Control (hls_max_concurrency Attribute) 7.3. Component Pipelining Control (hls_disable_component_pipelining Attribute)
8. Arbitrary Precision Math Support x
8.1. Declaring ac_int Data Types 8.2. Declaring ac_fixed Data Types 8.3. Declaring ac_complex Data Types 8.4. AC Data Types and Native Compilers 8.5. Declaring hls_float Data Types
8.1. Declaring ac_int Data Types x
8.1.1. Important Usage Information on the ac_int Data Type 8.1.2. Integer Promotion and ac_int Data Types 8.1.3. Debugging Your Use of the ac_int Data Type
8.2. Declaring ac_fixed Data Types x
8.2.1. Arbitrary Precision Fixed-Point Literals in Operations
8.5. Declaring hls_float Data Types x
8.5.1. Operators and Return Types Supported by the hls_float Data Type 8.5.2. Additional Data Types Provided By hls_float.h
9. Component Target Frequency x
9.1. Effects of Specifying Target II and Target fMAX
10. Systems of Tasks x
10.1. Task Functions 10.2. Internal Streams 10.3. System of Tasks Simulation
11. Libraries x
11.1. Static-Object Libraries 11.2. Creating a Static-Object Library 11.3. Creating Objects From HLS Code 11.4. Creating Objects From RTL Code 11.5. Packaging Object Files Into a Library
11.3. Creating Objects From HLS Code x
11.3.1. Creating an Object File From HLS Code 11.3.2. Supported OpenCL* Language Constructs
11.4. Creating Objects From RTL Code x
11.4.1. RTL Modules and the HLS Pipeline 11.4.2. Creating a Static-Object File from an RTL Module
11.4.1. RTL Modules and the HLS Pipeline x
11.4.1.1. Integration of an RTL Module into the HLS Pipeline 11.4.1.2. RTL Module Interfaces 11.4.1.3. RTL Reset and Clock Signals 11.4.1.4. Object Manifest File Syntax 11.4.1.5. Mapping HLS Data Types to RTL Signals 11.4.1.6. HLS Emulation Models for RTL-Based Functions 11.4.1.7. Potential Incompatibility between RTL Modules and Partial Reconfiguration 11.4.1.8. Stall-Free RTL 11.4.1.9. RTL Module Restrictions and Limitations for HLS Libraries
11.4.1.2. RTL Module Interfaces x
11.4.1.2.1. RTL Module Interface Signals
11.4.1.3. RTL Reset and Clock Signals x
11.4.1.3.1. Intel® Agilex™ and Intel® Stratix® 10 Design-Specific Reset Requirements for Stall-Free and Stallable RTL Modules
11.4.1.4. Object Manifest File Syntax x
11.4.1.4.1. XML Elements for ATTRIBUTES 11.4.1.4.2. XML Elements for INTERFACE 11.4.1.4.3. XML Elements for RESOURCES
11.4.1.4.1. XML Elements for ATTRIBUTES x
11.4.1.4.1.1. Interaction Between ALLOW_MERGINGand HAS_SIDE_EFFECTS Elements
12. Advanced Hardware Synthesis Controls x
12.1. Hardware Implementation Control for Math Functions 12.2. The hls_fpga_reg() Function
12.1. Hardware Implementation Control for Math Functions x
12.1.1. Math Function Hardware Implementation Summary 12.1.2. The --dsp-mode Command Option 12.1.3. The ihc::math_dsp_control Function
13. Intel® High Level Synthesis Compiler Pro Edition Reference Summary x
13.1. Intel® HLS Compiler Pro Edition i++ Command-Line Arguments 13.2. Intel® HLS Compiler Pro Edition Header Files 13.3. Intel® HLS Compiler Pro Edition Compiler-Defined Preprocessor Macros 13.4. Intel® HLS Compiler Pro Edition Keywords 13.5. Intel® HLS Compiler Pro Edition Simulation API (Testbench Only) 13.6. Intel® HLS Compiler Pro Edition Component Memory Attributes 13.7. Intel® HLS Compiler Pro Edition Loop Pragmas 13.8. Intel® HLS Compiler Pro Edition Scope Pragmas 13.9. Intel® HLS Compiler Pro Edition Component Attributes 13.10. Intel® HLS Compiler Pro Edition Component Default Interfaces 13.11. Intel® HLS Compiler Pro Edition Component Invocation Interface Control Attributes 13.12. Intel® HLS Compiler Pro Edition Component Macros 13.13. Intel® HLS Compiler Pro Edition Systems of Tasks API 13.14. Intel® HLS Compiler Pro Edition Pipes API 13.15. Intel® HLS Compiler Pro Edition Streaming Input Interfaces 13.16. Intel® HLS Compiler Pro Edition Streaming Output Interfaces 13.17. Intel® HLS Compiler Pro Edition Memory-Mapped Interfaces 13.18. Intel® HLS Compiler Pro Edition Load-Store Unit Control 13.19. Intel® HLS Compiler Pro Edition Arbitrary Precision Data Types
13.13. Intel® HLS Compiler Pro Edition Systems of Tasks API x
13.13.1. ihc::stream Class
A. Advanced Math Source Code Libraries x
A.1. Random Number Generator Library A.2. Matrix Multiplication Library A.3. Cholesky Decomposition Library
B. Supported Math Functions x
B.1. Math Functions Provided by the math.h Header File B.2. Math Functions Provided by the extendedmath.h Header File B.3. Math Functions Provided by the ac_fixed_math.h Header File B.4. Math Functions Provided by the hls_float.h Header File B.5. Math Functions Provided by the hls_float_math.h Header File B.6. Default Rounding Schemes and Subnormal Number Support
1. Intel® HLS Compiler Pro Edition Reference Manual
2. Compiler
3. C Language and Library Support
4. Component Interfaces
5. Component Memories (Memory Attributes)
6. Loops in Components
7. Component Concurrency
8. Arbitrary Precision Math Support
9. Component Target Frequency
10. Systems of Tasks
11. Libraries
12. Advanced Hardware Synthesis Controls
13. Intel® High Level Synthesis Compiler Pro Edition Reference Summary
A. Advanced Math Source Code Libraries
B. Supported Math Functions
C. Cyclone® V Restrictions
D. Intel® HLS Compiler Pro Edition Reference Manual Archives
E. Document Revision History of the Intel® HLS Compiler Pro Edition Reference Manual

A.3. Cholesky Decomposition Library

The Cholesky Decomposition library provided with the Intel HLS Compiler provides an FPGA-optimized templated library to factor a Hermitian (symmetric) positive-definite matrix into the product of a lower triangular matrix and its conjugate transpose .

Header File

To include the Cholesky decomposition library in your component, add the following line to your component: 
#include ”HLS/cholesky_decompose.h”
The header file is self-documented. You can review the header file to learn how to use the Cholesky decomposition library. For high performance, use the –ffp-reassociate command flag when compiling your component.

Variants

You can call four different variants. All variants are in the ihc::cholesky namespace.

Cholesky decomposition iterates on the matrix heavily. For the best component performance, the component need low latency access to the matrix iterated on, and the matrix must be vectorized. However, this requirement is not always easy to meet, so two of the variants allow you to use separate pieces of memory for the iterative decomposition and recording the final output.

The variants provided are as follows:

  • cholesky_decompose_real(A_input, L_iter, n);

    Used for real-valued matrices.

  • cholesky_decompose_real(A_input, L_output, L_iter, n);

    Used for real-valued matrices with separate memory for iterative decomposition and final output.

  • cholesky_decompose_complex(A_input, L_iter, n);

    Used for complex-valued matrices.

  • cholesky_decompose_complex(A_input, L_output, L_iter, n);

    Used for complex-valued matrices with separate memory for iterative decomposition and final output.

Arguments

A_input

A pointer to an array of size MATRIX_SIZE * MATRIX_SIZE, for providing input A matrix.

Its data width and alignment should match that of a single element in the matrix.

L_iter

A pointer to an array of size MATRIX_SIZE * MATRIX_SIZE, for holding the L matrix that the program can iterate on.

In the 3-argument variants, this is the same place for recording the result. Its data width and alignment should match that of a vector of elements in the matrix, with specified vectorization width.

If a pointer to component memory is given, the compiler should be able to optimize that automatically.

L_output

A pointer to an array of size MATRIX_SIZE * MATRIX_SIZE, for recording the final matrix result in the 4-argument variants.

Its data width and alignment should match that of a single element in the matrix.

n

The actual matrix size.

It should not be greater than MATRIX_SIZE

Template Arguments

FP_T
The floating-point data type. Can be float or double.
VEC_SIZE_PWR
The logarithm (base 2) of the vectorization width. Vectorization with a width of is used for the access of L_iter matrix and the dot product computation.
This header uses the Intel® HLS Compiler ivdep pragma with its safelen() clause, Dummy iterations matching the safelen() values are inserted to manage memory access dependency more efficiently.
The ivdep pragma with its safelen() clause are used for the following template arguments in the Cholesky decomposition algorithm:
  • INNER_SAFELEN value applies to the partial_dot array,
  • OUTER_SAFELEN_OVERWRITE value applies to the L_iter matrix
Although a reasonable estimate of the safelen() value is provided, their value might require some tuning for different devices, clock targets, precisions, and memory arrangements.
Use an iterative approach for finding their optimal values. First, try large conservative estimates until the compilation result does not demonstrate any II issues and gives you a a satisfactory fMAX. Then, use the Function Viewer (part of the High-Level Design Reports) to examine the schedule of the load and store nodes for the memory the safelen() value applies to. The difference of start cycle of the load and store should approximately match the optimal value to use for that safelen() value.
INNER_SAFELEN
The INNER_SAFELEN value applies to the partial_dot array. Its value roughly matches the latency of index calculation and a floating point addition in the algorithm. This safelen() value does not grow with matrix size. A default value of 16 is given, which should be ideal for single precision. For double precision, increase it accordingly.
OUTER_SAFELEN_OVERWRITE
The OUTER_SAFELEN_OVERWRITE value applies to the L_iter matrix, and is related to intercolumn dependencies. An estimate of this safelen() value is provided and along with an estimation of how this value grows along with matrix size. However, the optimal value for this template argument depends on clock target values, target device families, and the precision used.

Use this parameter to overwrite the set safelen() value.

If OUTER_SAFELEN_OVERWRITE=-1, the safelen() value is tuned for single-precision float data types on an Intel® Arria® 10 device with a default clock target. The tuning also assumes that the L_iter matrix is in component memory with proper vectorization provided.

Level Two Title