Intel® C++ Compiler 16.0 User and Reference Guide
This topic only applies to Intel® 64 architecture targeting the Intel® Xeon Phi™ coprocessor x100 product family (formerly code name Knights Corner).
The use of C++ classes for SIMD operations on Intel® 64 architecture targeting the Intel® Xeon Phi™ coprocessor x100 product family (formerly code name Knights Corner) is based on the concept of operating on arrays, or vectors of data, in parallel.
Consider the addition of two vectors, A and B, where each vector contains sixteen elements. Using integer vector class, the elements A[i] and B[i] from each array are summed as shown in the following example.
int a[16], b[16], c[16];
for (i=0; i<16; i++) /* needs sixteen iterations */
c[i] = a[i] + b[i]; /* returns c[0], c[1], c[2], c[3], …, c[15] */
The following example shows the same results using one operation with integer vector Classes.
Is32vec16 ivecA, ivecB, ivec C; /*needs one iteration */
ivecC = ivecA + ivecB; /*returns 16 elements inside ivecC */
These C++ classes provide parallelism, which is not easily implemented using typical mechanisms of C++. The following table provides details of these class libraries.
| Instruction Set | Class | Signedness | Data Type | Size | Elements | Header File |
|---|---|---|---|---|---|---|
|
Intel® Initial Many Core Instructions (Intel® IMCI) |
F64vec8 |
unspecified |
double | 64 | 8 | micvec.h |
| F32vec16 |
unspecified |
float | 32 | 16 | micvec.h | |
| M512 |
unspecified |
__m512 | 512 | 1 | micvec.h | |
| I64vec8 |
unspecified |
long int | 64 | 8 | micvec.h | |
| I32vec16 |
unspecified |
int | 32 | 16 | micvec.h | |
| Is32vec16 |
signed |
int | 32 | 16 | micvec.h | |
| Iu32vec16 |
unsigned |
int | 32 | 16 | micvec.h |
Most classes contain similar functionality for all data types and are represented by all available intrinsics. However, some capabilities do not translate from one data type to another without suffering from poor performance, and are therefore excluded from individual classes. The I64vec8 class currently supports logical and data manipulation operations but not arithmetic operations.
Intrinsics that take immediate values and cannot be expressed easily in classes are not implemented.
The required class header files are installed in the include directory with the compiler. To enable the classes, use the #include directive in your program file as shown below:
#include <micvec.h>You can enable these classes within native code or within offloaded code that is written to contain target-specific code typically protected by the __MIC__ macro.