uMagNUS (v1.3.0)

Added new features:

• Exchange interaction between pairs of regions
• Spin torque between pairs of regions

uMagNUS (v1.2.2)

Patch release.

• The initialization of the XORWOW PRNG is fixed. Each thread will initialize with the same seed and will discard a number of subsequences based on their thread id. The length of each discarded subsequence is 2^67.
• Kernel codes are fixed to ensure successful compile on Intel GPUs. Tested in Windows 11 (x64) running 8th Gen i7 with UHD 620 GPU and Nvidia MX150.

uMagNUS (v1.2.1)

Patch release:

• The latest VkFFT library has issues with accuracy. Reverted to an earlier version that passes key tests.

uMagNUS (v1.2)

Feature update release. The following new features have been added:

• Support for THREEFRY random number generator for stochastic LLG simulations (faster than previous MTGP RNG)
• Support for multiple spin-torque and VCMA interfaces for simulating complex magnetic tunnel junction structures (e.g., penta-layer MTJs with dual MgO)

uMagNUS (v1.1)

uMagNUS v1.1 Release

Feature set update to uMagNUS:

• Support using MTGP random number generator for stochastic LLG simulations
• Support for Newell's method to calculate demagnetizing field

uMagNUS (v1.0)

Version 1.0 Release of uMagNUS

uMagNUS is a rewrite of MuMax3 (https://github.com/mumax/3) to run using OpenCL 1.2 instead of CUDA.
OpenCL 1.2 is a widely supported GPGPU standard, allowing GPUs from AMD, Intel and Nvidia to run uMagNUS.

All original MuMax3 problem file definitions and frontend interface are preserved in uMagNUS.