(jax-201-gpu-memory-allocation)= # GPU memory allocation **JAX will preallocate 75% of the total GPU memory when the first JAX operation is run.** Preallocating minimizes allocation overhead and memory fragmentation, but can sometimes cause out-of-memory (OOM) errors. If your JAX process fails with OOM, the following environment variables can be used to override the default behavior: **`XLA_PYTHON_CLIENT_PREALLOCATE=false`** This disables the preallocation behavior. JAX will instead allocate GPU memory as needed, potentially decreasing the overall memory usage. However, this behavior is more prone to GPU memory fragmentation, meaning a JAX program that uses most of the available GPU memory may OOM with preallocation disabled. **`XLA_CLIENT_MEM_FRACTION=.XX`** If preallocation is enabled, this makes JAX preallocate XX% of the total GPU memory, instead of the default 75%. Lowering the amount preallocated can fix OOMs that occur when the JAX program starts. (You may see the old name `XLA_PYTHON_CLIENT_MEM_FRACTION` in existing code; it is deprecated in favor of `XLA_CLIENT_MEM_FRACTION`.) **`XLA_PYTHON_CLIENT_ALLOCATOR=platform`** This makes JAX allocate exactly what is needed on demand, and deallocate memory that is no longer needed (note that this is the only configuration that will deallocate GPU memory, instead of reusing it). This is very slow, so it's not recommended for general use, but may be useful for running with the minimal possible GPU memory footprint or debugging OOM failures. ## Common causes of OOM failures **Running multiple JAX processes concurrently.** Either use `XLA_CLIENT_MEM_FRACTION` to give each process an appropriate amount of memory, or set `XLA_PYTHON_CLIENT_PREALLOCATE=false`. **Running JAX and GPU TensorFlow concurrently.** TensorFlow also preallocates by default, so this is similar to running multiple JAX processes concurrently. One solution is to use CPU-only TensorFlow (e.g. if you're only doing data loading with TF). You can prevent TensorFlow from using the GPU with the command `tf.config.experimental.set_visible_devices([], "GPU")`. Alternatively, use `XLA_CLIENT_MEM_FRACTION` or `XLA_PYTHON_CLIENT_PREALLOCATE`. There are also similar options to configure TensorFlow's GPU memory allocation; see [Using GPUs: Limiting GPU memory growth](https://www.tensorflow.org/guide/gpu#limiting_gpu_memory_growth). **Running JAX on the display GPU.** Use `XLA_CLIENT_MEM_FRACTION` or `XLA_PYTHON_CLIENT_PREALLOCATE`. **Poor choices by the automatic rematerialization pass.** Sometimes it's better to disable XLA's automatic rematerialization HLO pass, to avoid poor remat choices by the compiler. The pass is enabled by default; disable it with `jax.config.update('jax_compiler_enable_remat_pass', False)`. Enabling or disabling it produces different trade-offs between compute and memory; the algorithm is basic, and you can often get a better trade-off by disabling the pass and rematerializing manually with {func}`jax.checkpoint` (also known as `jax.remat`; see {ref}`jax-301-remat`). ## Experimental allocator features Features here are experimental and must be tried with caution. **`XLA_PYTHON_CLIENT_ALLOCATOR=vmm`** This uses CUDA's virtual memory management (VMM) allocator (`cudaDeviceAddressVmmAllocator`). This is a CUDA-only experimental allocator that provides fine-grained virtual memory control. It does not preallocate memory; use `XLA_CLIENT_MEM_FRACTION` to control the fraction of GPU memory used. **`TF_GPU_ALLOCATOR=cuda_malloc_async`** This replaces XLA's own BFC memory allocator with [`cudaMallocAsync`](https://docs.nvidia.com/cuda/cuda-runtime-api/cuda_runtime_api/group__CUDART__MEMORY__POOLS.html). This removes the big fixed preallocation and uses a memory pool that grows. The expected benefit is no need to set `XLA_CLIENT_MEM_FRACTION`. The risks are: - memory fragmentation is different, so if you are close to the limit, the exact OOM case due to fragmentation will be different; - the allocation time won't all be paid at the start, but incurred as the memory pool needs to grow, so you could experience less speed stability at the start (and for benchmarks it will be even more important to ignore the first few iterations). You can mitigate these risks, while keeping the benefit of a growing memory pool, by preallocating a significant chunk with `TF_CUDA_MALLOC_ASYNC_SUPPORTED_PREALLOC=N`. If `N` is `-1`, it preallocates the same amount as the default; otherwise it's the size in bytes to preallocate.