![]() |
AI Engine
(AIE) r2p15.2
|
Casting intrinsics allow casting between vector types of the same size. More...
Casting intrinsics allow casting between vector types of the same size.
For the single/double accumlator conversion intrinsics it may seem at first that the vectors are of different types (v8acc80 is 640-bit and v8cacc48 is 768-bit). However, the following two points provide more clarity on casting:
Single accumulator conversions | |
| v8acc48 | as_v8acc48 (v8acc48) |
| v4cacc48 | as_v4cacc48 (v8acc48) |
| v4acc80 | as_v4acc80 (v8acc48) |
| v2cacc80 | as_v2cacc80 (v8acc48) |
| v8acc48 | as_v8acc48 (v4cacc48) |
| v4cacc48 | as_v4cacc48 (v4cacc48) |
| v4acc80 | as_v4acc80 (v4cacc48) |
| v2cacc80 | as_v2cacc80 (v4cacc48) |
| v8acc48 | as_v8acc48 (v4acc80) |
| v4cacc48 | as_v4cacc48 (v4acc80) |
| v4acc80 | as_v4acc80 (v4acc80) |
| v2cacc80 | as_v2cacc80 (v4acc80) |
| v8acc48 | as_v8acc48 (v2cacc80) |
| v4cacc48 | as_v4cacc48 (v2cacc80) |
| v4acc80 | as_v4acc80 (v2cacc80) |
| v2cacc80 | as_v2cacc80 (v2cacc80) |
Double accumulator conversions | |
| v16acc48 | as_v16acc48 (v16acc48) |
| v8cacc48 | as_v8cacc48 (v16acc48) |
| v8acc80 | as_v8acc80 (v16acc48) |
| v4cacc80 | as_v4cacc80 (v16acc48) |
| v16acc48 | as_v16acc48 (v8cacc48) |
| v8cacc48 | as_v8cacc48 (v8cacc48) |
| v8acc80 | as_v8acc80 (v8cacc48) |
| v4cacc80 | as_v4cacc80 (v8cacc48) |
| v16acc48 | as_v16acc48 (v8acc80) |
| v8cacc48 | as_v8cacc48 (v8acc80) |
| v8acc80 | as_v8acc80 (v8acc80) |
| v4cacc80 | as_v4cacc80 (v8acc80) |
| v16acc48 | as_v16acc48 (v4cacc80) |
| v8cacc48 | as_v8cacc48 (v4cacc80) |
| v8acc80 | as_v8acc80 (v4cacc80) |
| v4cacc80 | as_v4cacc80 (v4cacc80) |
Scalar conversions | |
| int | as_int32 (float) |
| float | as_float (int) |
Scalar complex conversions | |
| int | as_int (cint16) |
| cint16 | as_cint16 (unsigned) |
| unsigned short | as_short (cint8) |
| cint8 | as_cint8 (unsigned short) |
| cint16 as_cint16 | ( | unsigned | ) |
| cint8 as_cint8 | ( | unsigned | short | ) |
| float as_float | ( | int | ) |
| int as_int | ( | cint16 | ) |
| int as_int32 | ( | float | ) |
| unsigned short as_short | ( | cint8 | ) |