Unanswered questions

Buongiorno a tutti, Mi chiamo Vito e sono il titolare di una società Italiana, una start-up. Insegno anche in un istituto tecnico di Brescia Robotica, Elettronica e Informatica. Abbiamo avviato un progetto per la realizzazione di un robot umanoide nella mia società con alcuni dei miei studenti, qualche mese fa nell'ambito dell'assistenza. Ho visto l'intervista dell'Ing. Del Maffeo e sono rimasto colpito. Spanner, ​@Spanner dobbiamo implementare il robot con AI e quindi di una scheda potente, possiamo collaborare? Il potenziale è grande..  come potremmo approfondire l'argomento? Vi ringrazio a tutti in anticipo.. 

Hi,I'm having trouble installing the metis-dkms driver. Can you help me?The device is a NanoPC-T6 from Friendly Elec, and the operating system is Ubuntu 22.04, also provided by Friendly Elec.I'm leaving the error message and OS information below. If there is any more information I should provide, please let me know. Thank you. The Error:$ sudo dpkg -i metis-dkms_0.07.16_all.deb(Reading database ... 155441 files and directories currently installed.)Preparing to unpack metis-dkms_0.07.16_all.deb ...Deleting module metis-0.07.16 completely from the DKMS tree.Unpacking metis-dkms (0.07.16) over (0.07.16) ...Setting up metis-dkms (0.07.16) ...Loading new metis-0.07.16 DKMS files...Building for 6.1.99Building for architecture aarch64Building initial module for 6.1.99ERROR (dkms apport): kernel package linux-headers-6.1.99 is not supportedError! Bad return status for module build on kernel: 6.1.99 (aarch64)Consult /var/lib/dkms/metis/0.07.16/build/make.log for more information.dpkg: error pro

Hello,I’m running inference with model  unet_fcn_512-cityscapes with pipe torch-aipu on the aetina eval board. It runs at 1.8 fps system and with 500ms of latency although device fps shows  11.5 fps capability. In the doc it is also mentioned that it should reach 18fps. I originally thought it was an issue with time wasted to load and decode png images from the SD card so I put them in shared memory but result are identical.I also tested yolov5s-v7-coco that should reach 805 fps but I can only achieve 214fps. Here are the output of: AXELERA_USE_CL_DOUBLE_BUFFER=0 ./inference.py yolov5s-v7-coco media/traffic3_720p.mp4 --show-stats --no-displayINFO : Deploying model yolov5s-v7-coco for 4 cores. This may take a while...|████████████████████████████████████████| 12:41.1 arm_release_ver: g13p0-01eac0, rk_so_ver: 9========================================================================

Hi all,I am trying to compile a high-resolution model (1504×1504) on a Metis PCIe card, but the compilation fails due to a memory constraint:ERROR : RuntimeError: Could not find a tiling factor that fits the memory constraints l1_constraint=4011520 l2_constraint=7894528. After attempt=7 and h_size=1 and adj_factor=1, memory usage still is memory_usage={L1: {190: 141376, 193: 5017600, 191: 143360}} and per-pool memory usage {L1: 5302336}.The error indicates a required L1 size (≈5.3 MiB) exceeds the constraint (≈4.0 MiB).My questions are: Voyager SDK Solution: Is there a Voyager SDK flag or setting to adjust the tiling to overcome this ≈4 MiB L1 limit? Hardware Solution: Are there any Metis card variants (or multi-AIPU cards) that feature a larger per-core L1 cache? Any guidance on compiling this high-resolution model is appreciated. Thank you!

Hello guys;I did not follow the instructions in the link below on the Orangepi 5 device, but it does not see the device.“ERR DEVCICE NOT FOUND”help! https://support.axelera.ai/hc/en-us/articles/27059519168146-Bring-up-Voyager-SDK-in-Orange-Pi-5-Plus

Hello everyone, I’m trying to install some packages using APT with my Axelera Metis Compute Board with Voyager Linux 1. However the /etc/apt/sources.list is empty. What kind of repositories should I use?Are there any tutorials?Thanks.Best regards

Hi there,i´m working with an Axelera M.2 chip at uni. For some research on green AI i´m doing i want to track the power usage of the chip. As far as i know i need to track the power through the PCIe connection. Is there a way to track it with the AIPU directly or do i need to use the PCIe? And if there is a way can i save the usage in an array easily or is this a rather tricky situation?Thanks in advance Morruebe

Hello, I’m new to voyager SDK and everything related to AXELERA and i need help deploying a custom onnx model that has 2 inputs and 20+ outputs, all i want for now is to manage to run an inference with my own dataset/image and to save to disk the raw outputs of the model. The input is a pair of 2 files. I have data in yuv format and it’s saved in input_y.npy and input_uv.npy. Do you have any idea how can i do this or if it’s possible from the voyager 1.3 or do i need to add a split layer so i need just 1 input and only then i can deploy it?  Or to upgrade the voyager version to 1.4 

Will Metis M.2 ( AXE-BME20M1AR01B02) work with QNAP NAS?

Does anyone have any suggestions on how to implement SAHI inference with Yolo via voyager-sdk to run on metis m2?

Hello,I am currently running performance tests on the Metis M.2 card with different models (e.g., YOLOv8n).I observed a behavior that I would like to clarify: When running with 1 AIPU core, I get lower latency and the same FPS as when using multiple cores (4). However, with 4 AIPU cores, the latency per image is higher, even though the FPS does not increase. In addition, CPU utilization is noticeably higher with 4 cores (around 17%) compared to only ~0.8% with 1 core. I would like to know: Is this expected behavior (due to synchronization overhead between AIPU cores)? Is there a recommended configuration to optimize either latency or throughput (FPS) depending on the use case? Is it normal that CPU usage increases significantly when using 4 cores, given that processing is supposed to be largely handled by the AIPU? Thank you in advance for your clarifications.Best regards,

Hello everyone,I have been trying to compile the PRL Tracker model (see link below) for about a month using both Hailo and Vitis AI, but so far I have not been able to fully meet the requirements.I am now considering using the Metis Accelerator, but I am not sure whether I would run into the same issues.Has anyone here successfully deployed this model on the Metis Accelerator, or could you confirm if it is supported?Repository: https://github.com/vision4robotics/PRL-Track/tree/main Any guidance or shared experience would be greatly appreciated.Thanks in advance! 

Hello everyone, I am trying to use the Metis PCIe (4GB) in combination with the firefly ITX-3588J.Issue: I can not load the Metis kernel module (driver) into the Linux kernelSetup: I have the Metis PCIe with 4GB of ram  the Firefly ITX-3588J with 16GB of ram with a RockChip. OS: Ubuntu 22.04.4 LTS. I have a Gen3 Pcie with 4 lanes, details below```        LnkCap:    Port #0, Speed 8GT/s, Width x4, ASPM L0s L1, Exit Latency L0s <4us, L1 <16us            ClockPM- Surprise- LLActRep- BwNot- ASPMOptComp+``` So far:I installed the ubuntu OS as described here: https://support.axelera.ai/hc/en-us/articles/25556437653138-System-Imaging-Guide-Firefly-RK3588 I installed the latest 1.3.3 voyager sdk release.  i can see the metis board from lspci``` firefly@firefly:~$ lspci | grep accelerators01:00.0 Processing accelerators: Axelera AI Metis AIPU (rev 02)```however, i do not detect any drivers for the board, even after installing the sdk. Both the following commands do not return anything:```

Hi,Bit of a weird issue with the M.2 Evaluation System (SBC model AIB-MR1B-A1), and I’m worried it could be a hardware fault. I somehow got to a state where I would see AXR_ERROR_CONNECTION_ERROR: No target device found in lspci output".lspci didn’t show the accelerator card, but instead listed “Non-VGA unclassified device”:00:00.0 PCI bridge: Rockchip Electronics Co., Ltd RK3588 (rev 01)01:00.0 Non-VGA unclassified device: Synopsys, Inc. DWC_usb3 / PCIe bridgedmesg showed:Tue Sep 2 19:51:14 2025] rk-pcie fe170000.pcie: PCIe Linking... LTSSM is 0x3 [Tue Sep 2 19:51:16 2025] rk-pcie fe170000.pcie: PCIe Link Fail [Tue Sep 2 19:51:16 2025] rk-pcie fe170000.pcie: failed to initialize hostAfter trying a few things, I resorted to re-imaging the SBC. Then, on the first log-in from adb shell, I still saw Non-VGA unclassified device, but after a reboot, SSH’ing into the Eval System, I saw it as:00:00.0 PCI bridge: Rockchip Electronics Co., Ltd Device 3588 (rev 01)01:00.0 Processing accelerators

Hi everyone,I'm currently developing a web application on top of the Axelera PCIe card and Voyager SDK, and I’ve hit a roadblock.Use Case Overview: I want to relay this annotated stream back to my web frontend for live viewing. I’ve tried: Using GStreamer pipelines for HLS (e.g., with hlssink) Embedding video using <video> tag on the frontend pointing to HLS or .mp4 System Details: Hardware: Axelera Metis PCIe Card Software: Voyager SDK (latest) YOLOv8n-based custom model Python 3.10 + GStreamer on Ubuntu 22.04

Hi,Recently I tried running some ROI tracking algorithms (the ones included with opencv, such as CSRT, MOSSE, etc) on an embedded ARM board and found that the tracking performance and latency is not as good as the application needs. One reason is that all these algorithms run on the CPU, which explains why they are not the fastest. Searching around, I found the VOT (Visual object tracking) challenge related to visual object tracking and it seems that most of the state-of-art trackers now use AI/ML. I guess that’s a good thing since in theory it should be easier to run these on dedicated AI accelerator hardware? Anyway, I want to try running a visual tracking algorithm on the Metis hardware to see how fast and well it can work. However, there are so many algorithms and I reckon the Metis may not support all the operators used in those networks? So before I embark on this task, i wanted to ask the Axelera experts for some guidance on what to go for and what to avoid in order to find a tr

Hi everyone,I’m new to working with the Axelera system. I see that there is a CenterCrop style preprocessing function, but in my use case I need more control. Specifically, I’d like to crop an image (for RTSP streams or training data) by specifying the x-offset, y-offset, width, and height of the region of interest.Is there a built-in preprocessing function in the Axelera SDK/YAML configs for this kind of custom crop? Or is the recommended way to handle this via GStreamer (e.g., videocrop/videobox) before passing frames to axinferencenet?Any help would be greatly appreciated.Thank you! 🙏

Thank you for your continued support.Could you please tell me the performance (fps) under the following conditions?1. METIS COMPUTE BOARD2. YOLOv8n model3. USB WEB CAMERA (1080P@60FPS)Thank you in advance.

Hello.I installed the SDK on Ubuntu 22.04, butit stopped with the following error:[172/194] Install libgstreamer-plugins-bad1.0-dev[172/194] E: Error, pkgProblemResolver::Resolve stopped. This is probably due to a non-modifiable package.ERROR: Failed to install libgstreamer-plugins-bad1.0-devI would appreciate any advice.

Hello,I’m deploying a model onnx (opset 17) and got a large bunch of warnings of unsupported operations. One of them is the unsupported Transpose:WARNING : Configuration of node '/backbone/3/0/Transpose' may not be supported. 'Transpose' parameters:WARNING : perm: [0, 3, 1, 2]WARNING : data: Metadata(shape=(1, 48, 64, 160), is_constant=False)WARNING : transposed: Metadata(shape=(1, 160, 48, 64), is_constant=False)WARNING : Unsatisfied constraint: perm == [0, 1, 2, 3]I guess the rule should have been perm != [0, 1, 2, 3] because here it is requesting a non no-op operation perm = [0, 3, 1, 2].I think this is also causing later the errorERROR : RuntimeError: The expanded size of the tensor (192) must match the existing size (160) at non-singleton dimension 1. Target sizes: [1, 192, 256, 160]. Tensor sizes: [1, 160, 1, 1]as it looks that indeed the transpose has not been applied.Could someone shade some light on this behavior?Thanks in advance