nki.isa.rand2#

nki.isa.rand2(dst, min, max, name=None)[source]#

Generate pseudo random numbers with uniform distribution using Vector Engine.

Note

Available only on NeuronCore-v4 and newer.

This instruction generates pseudo random numbers and stores them into SBUF/PSUM. The generated values follow a uniform distribution within the specified [min, max] range.

Key features:

  • Uses XORWOW PRNG algorithm for high-quality random number generation

  • Generates FP32 random values with uniform distribution

  • Supports output conversion to various data types

Memory types.

The output dst tile can be in SBUF or PSUM.

Data types.

The output dst tile can be any of: float8_e4m3, float8_e5m2, float16, bfloat16, float32, tfloat32, int8, int16, int32, uint8, uint16, or uint32.

Tile size.

The partition dimension size of dst must not exceed 128. The number of elements per partition of dst must not exceed the physical size of each SBUF/PSUM partition.

Constraints.

  • Supported arch versions: NeuronCore-v4+.

  • Supported engines: Vector.

  • min < max for valid range.

Parameters:

  • dst – the destination tensor to write random values to

  • min – minimum value for uniform distribution range (FP32), can be a scalar or vector value

  • max – maximum value for uniform distribution range (FP32), can be a scalar or vector value