Some behaviors of the solver can be set via environment variable that will override settings on the UI. This can be useful in a studio environment to make sure all users use the desired settings for a project.

Solver

PARADIGM_SOLVER_SOURCE_USE_NANOVDB

Value: Boolean - Default: ON

Sets if NanoVDB should be used when sourcing VDBs. Setting this t 0 means sourcing will occur on CPU then the result copied to the compute device.

PARADIGM_COLLISION_MASK

Value: Boolean - Default: ON

Masks collision fields such as sourceVelocity by the collision source volume.

PARADIGM_KERNEL_JACOBI_ORIGINAL

Value: Boolean - Default: OFF

Will use an un-optimized version of the Jacobi kernel.

Compute

PARADIGM_COMPUTE_FORCE_FINISH

Value: Boolean - Default: OFF

Forces the solver to finish each compute task immediately after it is queued. This is for debugging and should not be enable for normal use as it negatively impacts performance.

API

PARADIGM_COMPUTE_API_USE_OPENCL

Value: Boolean - Default: OFF

Has the solver use OpenCL. This overrides the user API parameter option.

PARADIGM_COMPUTE_API_USE_METAL

Value: Boolean - Default: OFF

Has the solver use Metal. This overrides the user API parameter option.

PARADIGM_COMPUTE_API_USE_CUDA

Value: Boolean - Default: OFF

Has the solver use Cuda. This overrides the user API parameter option.

OpenCL

PARADIGM_COMPUTE_OCL_EXTRA_COMPILER_OPTIONS

Value: String - Default: “”

Add extra compiler options for OpenCL. For example “-cl-fp32-correctly-rounded-divide-sqrt”, “-cl-fast-relaxed-math”, “-cl-unsafe-math-optimizations”. You can multiple separated by a space character.

PARADIGM_COMPUTE_OCL_USE_CPU_DEVICES

Value: Boolean - Default: ON

Enables or disables the use of CPU devices.

PARADIGM_COMPUTE_OCL_USE_GPU_DEVICES

Value: Boolean - Default: ON

Enables or disables the use of GPU devices.

PARADIGM_COMPUTE_OCL_USE_UNIFIED_MEMORY_DEVICES

Value: Boolean - Default: ON

Enables or disables the use of unified memory devices such as CPUs or integrated GPUs.

PARADIGM_COMPUTE_OCL_USE_INTEL_INTERGRATED_GRAPHICS_DEVICES

Value: Boolean - Default: OFF

Enables or disables the use of Intel integrated GPUs.

PARADIGM_OCL_FLOAT_CONSTANTS_USE_CONST

Value: Boolean - Default: OFF

Have constants use the “const” qualifier instead of “__constant” when declared in shader code.

PARADIGM_COMPUTE_OCL_AMD_GPU_OPTIONS

Value: Boolean - Default: ON

Adds additional flags for AMD GPUs.