Metis PCIe

Putting this up for folk who may hit a similar issue.If you have a machine which has a single root PCIe device, where the output of lspci -tv looks something like this;alsutton@svr204:~$ lspci -tv-[0000:00]-+-00.0 Intel Corporation 8th Gen Core Processor Host Bridge/DRAM Registers +-02.0 Intel Corporation CoffeeLake-S GT2 [UHD Graphics 630] +-08.0 Intel Corporation Xeon E3-1200 v5/v6 / E3-1500 v5 / 6th/7th/8th Gen Core Processor Gaussian Mixture Model +-14.0 Intel Corporation Cannon Lake PCH USB 3.1 xHCI Host Controller +-14.2 Intel Corporation Cannon Lake PCH Shared SRAM +-16.0 Intel Corporation Cannon Lake PCH HECI Controller +-1b.0-[01]----00.0 Micron/Crucial Technology P2 [Nick P2] / P3 / P3 Plus NVMe PCIe SSD (DRAM-less) +-1f.0 Intel Corporation B360 Chipset LPC/eSPI Controller +-1f.3 Intel Corporation Cannon Lake PCH cAVS +-1f.4 Intel Corporation Cannon Lake PCH SMBus Controller

I’m putting an idea out to see if anyone has some data to back a theory.I’ve been talking with ​@Spanner about some problems I was having with a card, and, after looking over everything I’ve done, I’m wondering if a 75W slot is *required* rather than recommended, which would mean you need to use a PCIe x16 slot.One issue I had with the card was a small v shaped cut-out in the full height PCIe back-plate, which meant that the card wouldn’t sit fully into any PCIe x16 slot that I had (photo attached so you can see what I mean). This meant I had to use a 8x or 4x slot.Looking through the PCIe spec, only 16x slots are supplied with 75W, meaning that the 8x and 4x slots I tried were limited to 25W (which I confirmed), which I’m thinking could be the cause of the problems I saw.I have a 2nd hand machine turning up next week which I can test this theory on more easily (eBay still is useful for some things), but I’m wondering whether the PCIe card should be sized for a 16x slot in order to ens

Hi, I’m trying to deploy and inference with a yolov8 segmentation model. I tried to follow the tutorial, but I just get:INFO: deploying model yolov8sseg-coco for 4 cores. This may take a while… How long should this run? I left it on for 3.5 hours and nothing happened (except the loading bar). This seems to be a model included in your library: yolov8sseg-coco.Any idea if I can turn on more extensive logging to see what is happening? Thank you in advance!  

I’m also not seeing the card recognised in `lspci`. I’m adding a post to this thread to keep all the info in one place for others, but let me know if it should be a different thread.I’ve enabled `pci-check.service` and get the following when I check the status;alsutton@svr220:~$ sudo systemctl status pcie-check.service● pcie-check.service - Check for PCIe devices with vendor ID 1f9d and reboot once if not found     Loaded: loaded (/etc/systemd/system/pcie-check.service; enabled; vendor preset: enabled)     Active: active (exited) since Wed 2025-08-06 09:33:01 UTC; 11min ago    Process: 1709 ExecStart=/usr/local/bin/check_pcie_device.sh (code=exited, status=0/SUCCESS)   Main PID: 1709 (code=exited, status=0/SUCCESS)        CPU: 6msAug 06 09:33:01 svr220 systemd[1]: Starting Check for PCIe devices with vendor ID 1f9d and reboot once if not found...Aug 06 09:33:01 svr220 check_pcie_device.sh[1709]: Reboot already performed; skipping further checks.Aug 06 09:33:01 svr220 systemd[1]: Finish

Hey everyone, I’m having trouble getting my Axelera Metis PCIe AI Accelerator to be recognized by `lspci`(and my system in general). I tested the card on two different systems. On my AMD setup, I am using an AMD 5950X with an ASUS B550-F. I tried installing the card in the slot I normally use for my RTX 3080 as well as in another slot that meets the specifications. In both cases, `lspci` does not list the card even though the fan spins and I know it is getting power. Also Voyager SDK dosn’t recognize the card. I also tried it on an Intel system with an Intel i5-8500T on a Supermicro X11SCA-F motherboard. The same issue occurs. The card is powered (fan spin) but not recognized. I noticed that the boot time increases significantly when the accelerator is installed. This makes me think that UEFI might be attempting to detect something, even though no error messages are shown. As I already tried a lot and I couldn’t get it running, I wonder if you have any ideas what I can test.Could there

I'm currently setting up the Axelera Voyager SDK on Ubuntu 22.04.5 LTS with a verified PCIe Metis AIPU card (rev 02 properly detected via lspci, and metis module loaded). However, during installation with install.sh --all --media, several symbolic link warnings occurred (e.g., media: invalid symlink), and although the runtime reports as installed, the file libaxruntime.so is missing entirely from runtime/lib/. This results in the inference script failing with: Failed to create device manager: libaxruntime.so: cannot open shared object file. Additionally, the AxInference.cpp build fails due to missing axruntime/axruntime.hpp. I’ve already tried clearing ~/.cache/axelera/ and reinstalling the runtime, but the issue persists. Would appreciate any guidance to resolve the missing runtime files and complete the setup.Thanks Pravin Kumar R

Hey!PCIe card did run out of the box with a first running demo in minutes with no issues!Question: Is there a way to control the fan speed to reduce the noise of the card?Thanks!

Hi all,I am trying to enable the the usage of the Metis PCIE rev02 on RISC-V Hosts, in particular the Sifive P550. Currently I could successfully (at least I think so), install the drivers and the device gets correctly identified and mapped in the system. I attach the loggings just to be sure: dmesg (after rescan):[  858.500785] pci 0000:01:00.0: [1f9d:1100] type 00 class 0x120000[  858.500843] pci 0000:01:00.0: reg 0x10: [mem 0x04380000-0x04380fff 64bit][  858.500865] pci 0000:01:00.0: reg 0x18: [mem 0x08000000-0x09ffffff][  858.500922] pci 0000:01:00.0: reg 0x30: [mem 0x00000000-0x0000ffff pref][  858.500943] pci 0000:01:00.0: Max Payload Size set to 512 (was 128, max 512)[  858.501115] pci 0000:01:00.0: supports D1[  858.501121] pci 0000:01:00.0: PME# supported from D0 D1 D3hot[  858.516166] pci_bus 0000:01: busn_res: [bus 01] end is updated to 01[  858.516197] pcieport 0000:00:00.0: BAR 14: assigned [mem 0x42000000-0x44ffffff][  858.516209] pci 0000:01:00.0: BAR 2: assigned [mem 0x

Hi folks,after I resolved some of the issues I had with my Metis PCIe device, I stopped working with it since it was so loud that I got headache after 30min of runtime.Now I realized that there is a new FW which silences the fan and I upgraded the SDK and the FW:(venv) root@holodeck7:/voyager-sdk# axdevice -vINFO: Found PCI device: 01:00.0 Processing accelerators: Axelera AI Metis AIPU (rev 02)INFO: Found AIPU driver: metis 90112 0INFO: Firmware version matches: v1.3.1INFO: Using device metis-0:1:0Device 0: metis-0:1:0 4GiB pcie flver=1.2.0-rc2 bcver=1.0 clock=800MHz(0-3:800MHz) mvm=0-3:100% device_runtime_firmware=v1.3.1 board_controller_board_type=matterhorn sw_throttling: 200°C, hysteresis 5°C, throttle rate:12% hw_throttling: 105°C, hysteresis 10°C pvt_warning_threshold: 95°CBUT the fan keeps on spinning at max although the Metis is not being used at all.Tried to find means to read-out the actual temperature, which does not seem to work:(venv) root@holodeck7:

I wanted to try the new SDK version `v1.3.1` but I now run into authentication issues with the installer.Authentication is required for the package index URL: https://software.axelera.ai/artifactory/axelera-dev-pypi/ - for the subset developmentChecking credentials with: pip install --dry-run --force --no-deps --no-cache-dir --break-system-packages --index-url https:/<email>:<token>@software.axelera.ai/artifactory/axelera-dev-pypi/ axelera_compilerERROR: Authentication failed for the supplied email address (florianzaruba@gmail.com) and token.When trying to run the step manually it succeeds though: ➜ voyager-sdk git:(v1.3.1) ✗ pip install --dry-run --force --no-deps --no-cache-dir --break-system-packages --index-url https://florianzaruba@gmail.com:<token>@software.axelera.ai/artifactory/axelera-dev-pypi/ axelera_compilerDefaulting to user installation because normal site-packages is not writeableLooking in indexes: https://florianzaruba%40gmail.com:****@software.axele

0000:01:00.0 : Axelera AI Metis AIPU (rev 02)ERROR:axelera.runtime:No AIPU driver found in lsmod outputERROR: AXR_ERROR_CONNECTION_ERROR: No AIPU driver found in lsmod output 

Dear all,I am trying to get my hands on the Metis PCIe card on my current working PC, beingmainboard: Asus Prime X570-P CPU: AMD Ryzen 9 3900X OS: Ubuntu 24.04.2 LTS, 6.8.0-58-generic x86_64The first issue I face is that the warm-reset does not seem to work for me, i.e. if I issue a reboot without re-powering the system, the Metis device does not show up in lspci, while after a power-down (with physical PSU switching off) it is there. This is reproducible and based on my experience points to either my mainboard BIOS issuing a wrong reset sequence to the PCIe slot, or the device itself not issuing a full reset sequence on warm-boot. Since I know how to get the card detected, this is only an annoyance - but I saw that also another user of AMD64 system in the forum has similar issues and the problem might go deeper.Which leads to the second issue I am observing and which I filed already a support request but realized here would be the better place to discuss it.According to David M. to ch

After finally getting my metis card properly accessible through PCI, it turned out I need to upgrade the FW to the latest version:(venv) root@zefir-PC:~/voyager-sdk# axdevice [libdmabuf.c:1720] Device command returned an error code.[libdmabuf.c:1721] Consider inspecting the device log via `triton_trace --slog`.[...]Device 0: metis-0:a:0 board_type=pcie fwver='v1.2.0-rc2+bl1-stage0' clock=800MHz(0-3:0MHz) mvm=0-3:-1%(venv) root@zefir-PC:~/voyager-sdk# triton_trace --slogVersion mismatch!Actual="v1.2.0-rc2+bl1" Expected="v1.2.5"Please update device firmware before continuing.I proceeded like described in FW flashing tutorial at github(venv) root@zefir-PC:~/Axelera.ai/firmware_release_public_v1.2.0# ./flash_update.sh flash_bundle.img>>> loading new AIPU runtime stage0>>> the following operations updates the flash content Make sure to only run this operation if really needed Also make sure to not power cycle the devices or it might be rendered non-functional.do y

Hi all,not sure if my issue is related to this one, but symptoms are at least similar, so I hope it matches best here. Otherwise sorry for derailing.I am trying to get my hands on the PCIe metis card on my AMD64 based system driven by an ASUS PRIME X570-P mainboard under Ubuntu-24.04 (6.8.0-58-generic x86_64). First issue was that the card did no show up under lspci, no matter which kernel boot options I tried. Then I looked into the mainboard BIOS to double-check the peripheral config and noticed that the PCIe slots are not fully independent. In fact, the board has two PCIex16 slots and three PCIex1, of whose the latter are meant for expansion cards and the former for graphic cards. They are pre-configured for ‘auto’ mode, which means that the BIOS detects how the slots are populated and splits available lanes between the two x16 slots.I changed this ‘auto’ setting to manually split the lanes to ‘x8+x8’ and after rebooting the card is now visible to the kernel:zefir@zefir-PC:~$ lspci

Hey!Is there any support planned for 24.04? I don’t have a machine available with the older Ubuntu version. It says native support on 22.04 here but not that it is completely unsupported. Thanks ➜  voyager-sdk git:(release/v1.2.5) ./install.sh --all --media --user florianzaruba[at]gmail.com --token <token>Install/update prerequisite packages required by the installer itself (y/n): yInstall/update prerequisite packages required by the installer itself (y/n): yERROR: cfg/config-ubuntu-2404-amd64.yaml: File not found 

Hi! I represent a large organization and am interested in conducting a feasibility study using your technology. Before proceeding, could you kindly provide a white paper detailing your technology, along with technical documentation showcasing the use cases the Metis chip supports.Kind regards,G.

hi there,I was wondering if you have a planned feature to enables the usage of multiple PCIe cards in the same system. if so, is there any feature to use large models over multiple cards?