|
42 | 42 | "cell_type": "markdown",
|
43 | 43 | "metadata": {},
|
44 | 44 | "source": [
|
45 |
| - "## Install environment" |
46 |
| - ] |
47 |
| - }, |
48 |
| - { |
49 |
| - "cell_type": "markdown", |
50 |
| - "metadata": {}, |
51 |
| - "source": [ |
| 45 | + "## Setup environment\n", |
| 46 | + "\n", |
52 | 47 | "Loading data directly from disk to GPU memory requires the `kvikio` library. In addition, this tutorial requires many other dependencies such as `monai`, `torch`, `torch_tensorrt`, `numpy`, `ignite`, `pandas`, `matplotlib`, etc. We recommend using the [MONAI Docker](https://docs.monai.io/en/latest/installation.html#from-dockerhub) image to run this tutorial, which includes pre-configured dependencies and allows you to skip manual installation.\n",
|
53 | 48 | "\n",
|
54 | 49 | "If not using MONAI Docker, install `kvikio` using one of these methods:\n",
|
|
113 | 108 | "from monai.inferers import sliding_window_inference\n",
|
114 | 109 | "from monai.networks.nets import SegResNet\n",
|
115 | 110 | "import matplotlib.pyplot as plt\n",
|
116 |
| - "import torch\n", |
117 | 111 | "import gc\n",
|
118 | 112 | "import pandas as pd\n",
|
119 | 113 | "from timeit import default_timer as timer\n",
|
| 114 | + "from utils import prepare_test_datalist, prepare_model_weights, prepare_tensorrt_model\n", |
120 | 115 | "\n",
|
121 | 116 | "print_config()"
|
122 | 117 | ]
|
|
284 | 279 | }
|
285 | 280 | ],
|
286 | 281 | "source": [
|
287 |
| - "from utils import prepare_test_datalist, prepare_model_weights, prepare_tensorrt_model\n", |
288 |
| - "\n", |
289 | 282 | "root_dir = \".\"\n",
|
290 | 283 | "torch.backends.cudnn.benchmark = True\n",
|
291 | 284 | "torch_tensorrt.runtime.set_multi_device_safe_mode(True)\n",
|
|
465 | 458 | "outputs": [],
|
466 | 459 | "source": [
|
467 | 460 | "# collect benchmark results\n",
|
468 |
| - "all_df = pd.read_csv(os.path.join(root_dir, f\"time_original.csv\"))\n", |
| 461 | + "all_df = pd.read_csv(os.path.join(root_dir, \"time_original.csv\"))\n", |
469 | 462 | "all_df.columns = [\"file_name\", \"original_time\"]\n",
|
470 | 463 | "\n",
|
471 | 464 | "for benchmark_type in [\"trt\", \"trt_gpu_transforms\", \"trt_gds_gpu_transforms\"]:\n",
|
|
0 commit comments