quadrants.algorithms # previous quadrants.ad next quadrants.experimental Show Source

tensor_types scalar_tensors matrix_vector tensor compound_types buffer_view static sub_functions parallelization ```

image submodule for image io. video submodule for exporting results to video files. diagnose submodule for printing system environment information.

Quadrants offers two underlying tensor implementations, [`qd.field` and `qd.ndarray`](tensor_types.md). They have different runtime/compile-time trade-offs, and different physical memory layouts can suit …

Matrix and Vector # Quadrants provides qd.Matrix and qd.Vector types for small, fixed-size linear algebra inside kernels. These are stored in GPU registers and are unrolled at compile time, so they …

Tensor types # There are two core tensor types: ndarray (qd.ndarray) field (qd.field) See also tensor for a unified qd.tensor(...) factory that picks between the two via a backend= keyword. In addition, when …

Both have identical APIs (creation, slice-syntax load/store, qd.outer rank-1 updates, cholesky_, solve_triangular_, SharedArray interop) and use subgroup shuffles for cross-thread communication …

Troubleshooting # In case of crash/seg fault # run without cache - or clear cache - to see if this resolves the issue if running without cache solves the seg fault, then clear the cache To run without cache:

Block-level primitives operate on the threads of a single CUDA thread block (CTA) / AMDGPU workgroup / Vulkan or Metal workgroup. They include thread barriers, memory fences, shared …

Module Contents # quadrants.lang.simt.grid.arch_uses_spv(arch)[source] # quadrants.lang.simt.grid.mem_fence()[source] # quadrants.lang.simt.grid.memfence()[source ...

quadrants.lang.kernel_impl.pyfunc(fn:Callable)→quadrants.lang._quadrants_callable.QuadrantsCallable[source] # Marks a function as callable in both Quadrants and Python scopes. When called inside the …