From 1044a084aa5206ee1c8a0f29edcc4afd5568e095 Mon Sep 17 00:00:00 2001 From: Lin Yuan Date: Tue, 30 Oct 2018 15:02:37 -0700 Subject: [PATCH] document env variable MXNET_ENFORCE_DETERMINISM --- docs/faq/env_var.md | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/docs/faq/env_var.md b/docs/faq/env_var.md index 0664d7907415..cf784d923638 100644 --- a/docs/faq/env_var.md +++ b/docs/faq/env_var.md @@ -173,6 +173,12 @@ When USE_PROFILER is enabled in Makefile or CMake, the following environments ca - Flag to enable or disable MKLDNN accelerator. On by default. - Only applies to mxnet that has been compiled with MKLDNN (```pip install mxnet-mkl``` or built from source with ```USE_MKLDNN=1```) +* MXNET_ENFORCE_DETERMINISM + - Values: 0(false) or 1(true) ```(default=0)``` + - If set to true, MXNet will only use deterministic algorithms in forward and backward computation. + If no such algorithm exists given other constraints, MXNet will error out. This variable affects the choice + of CUDNN convolution algorithms. Please see [CUDNN developer guide](https://docs.nvidia.com/deeplearning/sdk/cudnn-developer-guide/index.html) for more details. + Settings for Minimum Memory Usage --------------------------------- - Make sure ```min(MXNET_EXEC_NUM_TEMP, MXNET_GPU_WORKER_NTHREADS) = 1```