Environment Variables

MLX uses environment variables to configure compilation, numerical precision, backend behavior, and distributed execution. Set them before starting the process. Many variables are read when the corresponding subsystem is first initialized and changing them later may have no effect.

Boolean variables use 0 to disable and a nonzero integer to enable unless otherwise noted.

General

MLX_DISABLE_COMPILE

Disable compilation globally. This variable is enabled by its presence, so setting it to 0 also disables compilation. Calling mlx.core.enable_compile() overrides it.

MLX_ENABLE_TF32

Allow reduced-precision float32 matrix-multiplication family operations on supported hardware. The default is 1. Set it to 0 to keep these operations in full float32 precision. See Numerical Precision.

Distributed

The variables required to initialize a distributed process depend on the backend. Distributed Communication describes their formats and how mlx.launch sets them.

MLX_RANK

The zero-based rank of the current process. This is used by the Ring, JACCL, and NCCL backends. JACCL_RANK is accepted as a higher-priority alias by JACCL.

MLX_HOSTFILE

The path to the JSON host file used by the Ring backend.

MLX_RING_VERBOSE

Enable verbose logging for the Ring backend. This variable is enabled by its presence.

MLX_IBV_DEVICES

The path to the JSON device-connectivity file used by JACCL. JACCL_IBV_DEVICES is accepted as a higher-priority alias.

MLX_JACCL_COORDINATOR

The coordinator address in IP:port form used to establish JACCL connections. JACCL_COORDINATOR is accepted as a higher-priority alias.

MLX_JACCL_RING

Prefer a ring topology for JACCL. This variable is enabled by its presence. JACCL_RING is accepted as a higher-priority alias.

MLX_WORLD_SIZE

The total number of processes in an NCCL group.

MLX_NCCL_TIMEOUT

The timeout in milliseconds for establishing NCCL bootstrap connections. The default is 300000.

MLX_MPI_LIBNAME

Override the MPI dynamic-library name. The default is libmpi.dylib on macOS and libmpi.so on other platforms.

The NCCL backend also requires NCCL_HOST_IP and NCCL_PORT. Setting NCCL_DEBUG=INFO enables additional logging while MLX establishes the bootstrap connection. NCCL itself recognizes additional NCCL environment variables. CUDA_VISIBLE_DEVICES selects the local CUDA device for each process and is handled by the CUDA runtime.

Metal

MLX_METAL_FAST_SYNCH

Enable the faster Metal CPU/GPU synchronization path. The default is 0. This requires Metal 3.2 or later (macOS 15 or later, or iOS 18 or later).

Advanced tuning

These variables tune MLX implementation details. They are primarily intended for development, diagnostics, and performance experiments. The defaults are selected automatically for the current hardware and are appropriate for most users. Their behavior may change as the implementation evolves.

MLX_BFS_MAX_WIDTH

Set the breadth-first-search width limit used when constructing an evaluation tape. The default is 20.

MLX_MAX_OPS_PER_BUFFER

Override the maximum number of operations encoded in one Metal command buffer or CUDA graph. The default depends on the device.

MLX_MAX_MB_PER_BUFFER

Override the approximate memory limit, in megabytes, for one Metal command buffer or CUDA graph. The default depends on the device.

MLX_METAL_GPU_ARCH

Override the Metal GPU architecture string reported to MLX. This affects architecture-specific kernel and scheduling choices, but does not change the capabilities of the physical GPU. Forcing an architecture that does not match the GPU can select incompatible kernels and produce incorrect results.

MLX_SDPA_BLOCKS

Override the number of reduction blocks used by the Metal scaled dot-product attention kernel. Positive values are rounded up to a multiple of 32.

CUDA

The MLX-prefixed variables in this section are advanced CUDA backend controls.

MLX_USE_CUDA_GRAPHS

Enable CUDA graph capture and replay. The default is 1.

MLX_SAVE_CUDA_GRAPHS_DOT_FILE

Use the specified value as the filename prefix when writing captured CUDA graphs to numbered DOT files. An unset or empty value disables the output.

MLX_PTX_CACHE_DIR

Override the directory used to cache runtime-compiled PTX. By default MLX uses an mlx/<version>/ptx directory under the system temporary directory.

MLX_CUDA_USE_CUDNN_SDPA

Allow the CUDA backend to use cuDNN scaled dot-product attention when the inputs and device are supported. The default is 1.

MLX_CUDA_CONV_CACHE_SIZE

Set the CUDA convolution cache capacity. The default is 128.

MLX_CUDA_FFT_CACHE_SIZE

Set the CUDA FFT plan cache capacity. The default is 128.

MLX_CUDA_GRAPH_CACHE_SIZE

Set the CUDA graph cache capacity. The default is 400.

MLX_CUDA_SDPA_CACHE_SIZE

Set the CUDA forward scaled dot-product attention cache capacity. The default is 256.

MLX_CUDA_SDPA_BACKWARD_CACHE_SIZE

Set the CUDA backward scaled dot-product attention cache capacity. The default is 64.

MLX_ENABLE_CACHE_THRASHING_CHECK

Detect repeated CUDA cache misses and raise an error suggesting a larger cache capacity. The default is 1.

MLX also uses CUDA_HOME or CUDA_PATH to locate CUDA headers for runtime kernel compilation when they cannot be found in the Python environment.