# Compute time in DLA slower than expected **URL:** \ **Category:** Jetson AGX Orin\ **Tags:** dla\ **Created:** [July 26, 2023, 2:49pm UTC](https://forums.developer.nvidia.com/t/compute-time-in-dla-slower-than-expected/261089 "2023-07-26T14:49:20Z")\ **Posts on this page:** 6\ **Page:** 1
**Author:** ![vinayakah](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@vinayakah](https://forums.developer.nvidia.com/u/vinayakah)\ **Post date:** [July 26, 2023, 2:49pm UTC](https://forums.developer.nvidia.com/t/compute-time-in-dla-slower-than-expected/261089/1 "2023-07-26T14:49:20Z")
## Description Hi, we tried to run Resnet50 pretrained inference model on the DLA of Jetson AGX Orin using this code: [GitHub - NVIDIA-AI-IOT/jetson\_dla\_tutorial: A tutorial for getting started with the Deep Learning Accelerator (DLA) on NVIDIA Jetson](https://github.com/NVIDIA-AI-IOT/jetson_dla_tutorial). The resnet50 pretrained model is used from torhvision.models module. We have used the same format for creating the TRT builder, creating the TRT engine, creating IO buffers and doing the inference using async\_exec\_v2 function. But, the slowdown in compute time for DLA, when compared to the GPU is much higher than expected. For batch\_size = 1, The compute time when GPU is used = 1.2 ms and the compute time when DLA is used = 19ms, which is 15x degradation. Similarly, for batch\_size = 16, The compute time when GPU is used = 8 ms and the compute time when DLA is used = 333ms, which is 40x degradation. But, recent papers show a 3-5x degradation in DLA when compared to GPU. Can we please know why are we observing such a large degradation in DLA when compared to GPU ? ## Environment **Device** : Jetson AGX Orin **TensorRT Version** : 8.4.0 **Nvidia Driver Version** : V11.4.239 **CUDA Version** : 11.4 **Jetpack + Version** : 5.0.1-b118 **Python Version (if applicable)**: 3.8.10 **Baremetal or Container (if container which image + tag)**: Baremetal ## Relevant Files > **[GitHub - NVIDIA-AI-IOT/jetson\_dla\_tutorial: A tutorial for getting started with the Deep...](https://github.com/NVIDIA-AI-IOT/jetson_dla_tutorial)** > > A tutorial for getting started with the Deep Learning Accelerator (DLA) on NVIDIA Jetson ---
**Author:** ![AakankshaS](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/aakankshas/32/14047_2.png) [@AakankshaS](https://forums.developer.nvidia.com/u/AakankshaS)\ **Post date:** [July 26, 2023, 6:37pm UTC](https://forums.developer.nvidia.com/t/compute-time-in-dla-slower-than-expected/261089/2 "2023-07-26T18:37:31Z")
Hi, Please check the below links, as they might answer your concerns. > **[Developer Guide :: NVIDIA Deep Learning TensorRT Documentation](https://docs.nvidia.com/deeplearning/tensorrt/developer-guide/index.html#dla_topic)** > > This is the revision history of the NVIDIA TensorRT 8.6 Developer Guide. > **[Developer Guide :: NVIDIA Deep Learning TensorRT Documentation](https://docs.nvidia.com/deeplearning/tensorrt/developer-guide/index.html#dla_layers)** > > This is the revision history of the NVIDIA TensorRT 8.6 Developer Guide. > **[Developer Guide :: NVIDIA Deep Learning TensorRT Documentation](https://docs.nvidia.com/deeplearning/tensorrt/developer-guide/#restrictions-with-dla)** > > This is the revision history of the NVIDIA TensorRT 8.6 Developer Guide. Thanks! ---
**Author:** ![vinayakah](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@vinayakah](https://forums.developer.nvidia.com/u/vinayakah)\ **Post date:** [July 27, 2023, 4:59am UTC](https://forums.developer.nvidia.com/t/compute-time-in-dla-slower-than-expected/261089/4 "2023-07-27T04:59:59Z")
Hi @AakankshaS thanks for the reply. We have already gone through the documentation and the related links that you’ve shared. But, none of them answers our queries as to what is causing the slowdown in the DLA. Request you to please forward this query to the relevant team. Thanks ! ---
**Author:** ![spolisetty](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/spolisetty/32/14047_2.png) [@spolisetty](https://forums.developer.nvidia.com/u/spolisetty)\ **Post date:** [July 27, 2023, 4:21pm UTC](https://forums.developer.nvidia.com/t/compute-time-in-dla-slower-than-expected/261089/5 "2023-07-27T16:21:12Z")
Hi, We are moving this post to the Jetson AGX Orin forum to get better help. Thank you. ---
**Author:** ![AastaLLL](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/aastalll/32/14043_2.png) [@AastaLLL](https://forums.developer.nvidia.com/u/AastaLLL)\ **Post date:** [July 28, 2023, 3:25am UTC](https://forums.developer.nvidia.com/t/compute-time-in-dla-slower-than-expected/261089/7 "2023-07-28T03:25:19Z")
Hi, Have you maximized the device performance first? ```auto $ sudo nvpmodel -m 0 $ sudo jetson_clocks ``` More, could you try it with INT8 mode? Orin’s DLA FP16 mode is expected to run slower by design. More details can be found in the below topic: > [@DLA-v2 is slower than DLA-v1](https://forums.developer.nvidia.com/t/dla-v2-is-slower-than-dla-v1/215468): > > Hi, I’m testing AGX Orin’s NVDLA. The neural network inference results are significantly slower than Xavier-AGX and Xavier-NX. I’m following the instructions for neural networks on building the engines: [GitHub - dusty-nv/jetson-inference: Hello AI World guide to deploying deep-learning inference networks and deep vision primitives with TensorRT and NVIDIA Jetson.](https://github.com/dusty-nv/jetson-inference). I’m ready to provide more details if necessary. System details: Xavier AGX ORIN Developer KIT Power mode: MAXN nv\_tegra\_release … Thanks. ---
**Author:** ![system](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/system/32/68080_2.png) [@system](https://forums.developer.nvidia.com/u/system)\ **Post date:** [August 23, 2023, 1:52am UTC](https://forums.developer.nvidia.com/t/compute-time-in-dla-slower-than-expected/261089/9 "2023-08-23T01:52:41Z")
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.