# Technical Support **URL:** https://community.memryx.com/c/technical-support/10.md [Latest](https://community.memryx.com/latest.md) · [Categories](https://community.memryx.com/categories.md) · [Tags](https://community.memryx.com/tags.md) --- ## [About the Technical Support category](https://community.memryx.com/t/about-the-technical-support-category/24)
**Author:** [@sam.fuller](https://community.memryx.com/u/sam.fuller)\ **Replies:** 0\ **Last updated:** [November 13, 2024, 12:13am UTC](https://community.memryx.com/t/about-the-technical-support-category/24 "2024-11-13T00:13:52Z")
For questions about the MX3 AI accelerator hardware check here. Post here if you can’t find the answers to your questions. MemryX staff monitors this category and will respond in a reasonable time frame. --- ## [MX3 wedges under concurrent streaming with some DFPs — reproducer, workarounds, and a way to recover without rebooting](https://community.memryx.com/t/mx3-wedges-under-concurrent-streaming-with-some-dfps-reproducer-workarounds-and-a-way-to-recover-without-rebooting/232)
**Author:** [@JESUSROYETH](https://community.memryx.com/u/JESUSROYETH)\ **Replies:** 2\ **Last updated:** [August 17, 2026, 1:19pm UTC](https://community.memryx.com/t/mx3-wedges-under-concurrent-streaming-with-some-dfps-reproducer-workarounds-and-a-way-to-recover-without-rebooting/232 "2026-08-17T13:19:52Z")
Hi @lindsay.morel and team: Update, same day: I kept measuring and the cause is not what this post says. These DFPs are not broken — they run 3000 frames straight with one inference in flight. The failure needs concurre… --- ## [Cannot compile YOLO11](https://community.memryx.com/t/cannot-compile-yolo11/229)
**Author:** [@ahmet](https://community.memryx.com/u/ahmet)\ **Replies:** 4\ **Last updated:** [July 29, 2026, 7:18pm UTC](https://community.memryx.com/t/cannot-compile-yolo11/229 "2026-07-29T19:18:24Z")
Hi, mx\_nc --version 2.2.2 I want to compile the YOLO11 model I exported from Ultralytics using the Neural Compiler, but I’m encountering an error. memryx.errors.OperatorError: Matmul (‘/model.10/m/m.0/attn/MatMul’) w… --- ## [Request: RF-DETR (Roboflow) Neural Compiler Extension](https://community.memryx.com/t/request-rf-detr-roboflow-neural-compiler-extension/226)
**Author:** [@Miguel](https://community.memryx.com/u/Miguel)\ **Replies:** 1\ **Last updated:** [July 16, 2026, 2:50pm UTC](https://community.memryx.com/t/request-rf-detr-roboflow-neural-compiler-extension/226 "2026-07-16T14:50:17Z")
We have an RF-DETR face detection model (exported to ONNX) we’d like to compile to DFP on SDK 2.2. The compiler fails on GridSample, TopK, GatherElements, Where (ops from the multi-scale deformable attention decoder). Gi… --- ## [Unable to install MemryX runtime v2.2 on Windows](https://community.memryx.com/t/unable-to-install-memryx-runtime-v2-2-on-windows/200)
**Author:** [@barkas](https://community.memryx.com/u/barkas)\ **Replies:** 1\ **Last updated:** [April 20, 2026, 7:52pm UTC](https://community.memryx.com/t/unable-to-install-memryx-runtime-v2-2-on-windows/200 "2026-04-20T19:52:11Z")
I was upgrading a previous 2.0 MemryX runtime & driver installation, which was working just fine. Here is the steps I followed: Downloaded MemryXInstaller.exe; Ran it as Admin; Checked in Device Manager and saw accele… --- ## [Bug in MemryX runtime v2.2](https://community.memryx.com/t/bug-in-memryx-runtime-v2-2/199)
**Author:** [@barkas](https://community.memryx.com/u/barkas)\ **Replies:** 1\ **Last updated:** [April 20, 2026, 2:21pm UTC](https://community.memryx.com/t/bug-in-memryx-runtime-v2-2/199 "2026-04-20T14:21:05Z")
Trying to run MemryX in local mode (by setting local\_mode = true and ignore\_server\_ = false in MX::Runtime::MxAccl constructor leads to a crash with vector::\_M\_range\_check: \_\_n (which is 0) \>= this-\>size() (which is 0… --- ## [SDK 2.1 question: placement signals for multi-device / multi-DFP setups](https://community.memryx.com/t/sdk-2-1-question-placement-signals-for-multi-device-multi-dfp-setups/186)
**Author:** [@Jerem6](https://community.memryx.com/u/Jerem6)\ **Replies:** 3\ **Last updated:** [April 1, 2026, 9:32pm UTC](https://community.memryx.com/t/sdk-2-1-question-placement-signals-for-multi-device-multi-dfp-setups/186 "2026-04-01T21:32:09Z")
Hello all, I’ve been reading through the SDK 2.1 docs and a few related threads, and I had a question about placement in multi-device or multi-DFP setups. When compile-time capabilities and runtime deployment constrain… --- ## [Quad M.2 Adapters?](https://community.memryx.com/t/quad-m-2-adapters/172)
**Author:** [@Amiejay](https://community.memryx.com/u/Amiejay)\ **Replies:** 9\ **Last updated:** [March 13, 2026, 11:22pm UTC](https://community.memryx.com/t/quad-m-2-adapters/172 "2026-03-13T23:22:45Z")
Hello… Are there any quad M.2 adapters that support 4 MX3 devices? Are there any devices that will support 4 MX3 devices with a Raspberry Pi 5? For my particular application, even something connected with USB3/USB4 wo… --- ## [Unable to compile a model](https://community.memryx.com/t/unable-to-compile-a-model/174)
**Author:** [@yan12125](https://community.memryx.com/u/yan12125)\ **Replies:** 6\ **Last updated:** [March 13, 2026, 6:26am UTC](https://community.memryx.com/t/unable-to-compile-a-model/174 "2026-03-13T06:26:25Z")
I have an ONNX model that causes mx\_nc to report “the model must have a uniform batch dimension throughout the entire model” error. However, the model already have batch size=1 for all layers. How to fix that? The model… --- ## [MX3 on Intel N5105 SBC - illegal instruction](https://community.memryx.com/t/mx3-on-intel-n5105-sbc-illegal-instruction/171)
**Author:** [@andrey](https://community.memryx.com/u/andrey)\ **Replies:** 1\ **Last updated:** [March 5, 2026, 9:00pm UTC](https://community.memryx.com/t/mx3-on-intel-n5105-sbc-illegal-instruction/171 "2026-03-05T21:00:21Z")
Hi folks, I’m setting up a Frigate instance on Odroid H3 + Debian 13. SDK install went smooth until the very last step - mx\_nc –version. mx\_nc crashes with illegal instruction and initial digging points to vmovq (in neu… --- ## [DFP exceution failur](https://community.memryx.com/t/dfp-exceution-failur/160)
**Author:** [@ralmusharraf](https://community.memryx.com/u/ralmusharraf)\ **Replies:** 1\ **Last updated:** [January 13, 2026, 4:17pm UTC](https://community.memryx.com/t/dfp-exceution-failur/160 "2026-01-13T16:17:57Z")
Iam currently running benchmarks on the MXA device using the mx\_bench command with 1,000 frames of random data. This setup was working previously, however, at the moment mx\_bench is no longer running successfully, and a… --- ## [I think I opened this under the wrong category -- general instead of technical](https://community.memryx.com/t/i-think-i-opened-this-under-the-wrong-category-general-instead-of-technical/158)
**Author:** [@Aragon476](https://community.memryx.com/u/Aragon476)\ **Replies:** 0\ **Last updated:** [December 20, 2025, 5:15am UTC](https://community.memryx.com/t/i-think-i-opened-this-under-the-wrong-category-general-instead-of-technical/158 "2025-12-20T05:15:21Z")
--- ## [MX3 failed upgrade](https://community.memryx.com/t/mx3-failed-upgrade/146)
**Author:** [@ronne991](https://community.memryx.com/u/ronne991)\ **Replies:** 4\ **Last updated:** [December 5, 2025, 6:18am UTC](https://community.memryx.com/t/mx3-failed-upgrade/146 "2025-12-05T06:18:55Z")
Hi everyone, I have an MX3 M.2 accelerator module (purchased a month plug ago from Mouser). It was working normally until I tried to upgrade the MemryX SDK. Below is the log. The accelerator no longer appear under lspci… --- ## [Missing Simulator in SDK 2.1 – Clarification on Tutorial Compatibility](https://community.memryx.com/t/missing-simulator-in-sdk-2-1-clarification-on-tutorial-compatibility/152)
**Author:** [@swells](https://community.memryx.com/u/swells)\ **Replies:** 1\ **Last updated:** [November 24, 2025, 9:17pm UTC](https://community.memryx.com/t/missing-simulator-in-sdk-2-1-clarification-on-tutorial-compatibility/152 "2025-11-24T21:17:21Z")
I’ve been following the tutorial Classification using Accelerator and noticed that the provided Python script imports: Python from memryx import NeuralCompiler, Simulator Show more lines However, in SDK 2.1 (installe… --- ## [MX3 support for Yolo-v11-nano-pose](https://community.memryx.com/t/mx3-support-for-yolo-v11-nano-pose/136)
**Author:** [@paspf](https://community.memryx.com/u/paspf)\ **Replies:** 3\ **Last updated:** [November 13, 2025, 7:46pm UTC](https://community.memryx.com/t/mx3-support-for-yolo-v11-nano-pose/136 "2025-11-13T19:46:24Z")
Hi memryX-Team, I want to use the Yolo-v11-nano-pose on the MX3. When converting the model I’m getting the error: Converting Model : (Done) memryx.errors.OperatorError: Matmul (‘/model.10/m/m.0/attn/MatMul’) with a… --- ## [Significantly reduced confidence values when using yolov10 on mx3](https://community.memryx.com/t/significantly-reduced-confidence-values-when-using-yolov10-on-mx3/134)
**Author:** [@ScytheEngineering](https://community.memryx.com/u/ScytheEngineering)\ **Replies:** 6\ **Last updated:** [November 6, 2025, 5:03pm UTC](https://community.memryx.com/t/significantly-reduced-confidence-values-when-using-yolov10-on-mx3/134 "2025-11-06T17:03:07Z")
Hello! I am running a trained YOLOv10 model on the MX3, and I’m testing between direct run of the .pt file generated by YOLO and running a compiled model on the MX3. It seems that both are generating very similar box det… --- ## [Support on Apple Silicon hosts](https://community.memryx.com/t/support-on-apple-silicon-hosts/123)
**Author:** [@eli](https://community.memryx.com/u/eli)\ **Replies:** 1\ **Last updated:** [October 15, 2025, 9:14am UTC](https://community.memryx.com/t/support-on-apple-silicon-hosts/123 "2025-10-15T09:14:19Z")
Hey everyone, I just bought my Memryx MX3 and I’m struggling to get the device to be recognized by my machine. I have a Macbook Air M1 (Apple Silicon) and I’m using lima-vm (running debian) as suggested in the FAQs sect… --- ## [Found 1 instance(s) of unsupported configuration for operator 'ReduceMax'](https://community.memryx.com/t/found-1-instance-s-of-unsupported-configuration-for-operator-reducemax/122)
**Author:** [@YuriB](https://community.memryx.com/u/YuriB)\ **Replies:** 4\ **Last updated:** [September 23, 2025, 1:41am UTC](https://community.memryx.com/t/found-1-instance-s-of-unsupported-configuration-for-operator-reducemax/122 "2025-09-23T01:41:11Z")
We’re attempting to convert a custom model for benchmarking with a 2-chip configuration: mx\_nc -v -m octonet.onnx -c 2 -so The model itself is a MobileNetV3 backbone with custom heads, as well as a secondary branch for… --- ## [USB-to-PCIe adapter?](https://community.memryx.com/t/usb-to-pcie-adapter/119)
**Author:** [@YuriB](https://community.memryx.com/u/YuriB)\ **Replies:** 8\ **Last updated:** [September 20, 2025, 5:10am UTC](https://community.memryx.com/t/usb-to-pcie-adapter/119 "2025-09-20T05:10:12Z")
Hi there, we’re looking to evaluate MemryX against a few other accelerators (like Hailo-8). MemryX looks very promising from the perspective of floating point math, which we want to leverage for regular vision ML as well… --- ## [Integrating with the rpicam-apps framework](https://community.memryx.com/t/integrating-with-the-rpicam-apps-framework/118)
**Author:** [@smat32](https://community.memryx.com/u/smat32)\ **Replies:** 1\ **Last updated:** [August 25, 2025, 6:34pm UTC](https://community.memryx.com/t/integrating-with-the-rpicam-apps-framework/118 "2025-08-25T18:34:12Z")
Any examples on how to efficiently inference on the mapped arrays in picamera 2 or the request buffers in rpicam apps c++ framework? --- ## [Setup MX\_Bench --hello](https://community.memryx.com/t/setup-mx-bench-hello/84)
**Author:** [@Joe](https://community.memryx.com/u/Joe)\ **Replies:** 12\ **Last updated:** [June 27, 2025, 9:07pm UTC](https://community.memryx.com/t/setup-mx-bench-hello/84 "2025-06-27T21:07:16Z")
Get the following response on trying to verify install on a raspberry pi5 - removed directory and user info leaving just the mx/ (mx) ~/mx $ mx\_bench --hello Traceback (most recent call last): File “mx/bin/mx\_bench”, … --- ## [Performance Metrics](https://community.memryx.com/t/performance-metrics/72)
**Author:** [@brise0117](https://community.memryx.com/u/brise0117)\ **Replies:** 3\ **Last updated:** [March 23, 2025, 2:50pm UTC](https://community.memryx.com/t/performance-metrics/72 "2025-03-23T14:50:17Z")
I’ve scoured the support site and most areas online looking for any way to grab performance metrics for monitoring. Is this possible? Ideally any of the following: Utilization percentage of the whole card (or individ… --- ## ["Hello, MobileNet!" example hangs forever](https://community.memryx.com/t/hello-mobilenet-example-hangs-forever/49)
**Author:** [@AlbertaBeef](https://community.memryx.com/u/AlbertaBeef)\ **Replies:** 3\ **Last updated:** [February 19, 2025, 2:38pm UTC](https://community.memryx.com/t/hello-mobilenet-example-hangs-forever/49 "2025-02-19T14:38:35Z")
Hello, I was able to assemble the heatsink on the MX3 module (thanks for the great video). I was able to install the drivers (thanks for the great instructions). Here is what “lspci -k” reports: (mx) ~/memryx$ lspci … --- ## [MxAccl::stop() takes very long time even when all inferences are done](https://community.memryx.com/t/mxaccl-stop-takes-very-long-time-even-when-all-inferences-are-done/54)
**Author:** [@vlad-nn](https://community.memryx.com/u/vlad-nn)\ **Replies:** 5\ **Last updated:** [February 3, 2025, 2:54pm UTC](https://community.memryx.com/t/mxaccl-stop-takes-very-long-time-even-when-all-inferences-are-done/54 "2025-02-03T14:54:58Z")
Typical use case: we want to switch from one model to another model. To do it gracefully, the documentation recommends to call MxAccl::wait() then MxAccl::stop(), and then we can load new dfp. What we noticed, that MxAc… --- ## [Device info and device selection](https://community.memryx.com/t/device-info-and-device-selection/53)
**Author:** [@vlad-nn](https://community.memryx.com/u/vlad-nn)\ **Replies:** 5\ **Last updated:** [January 29, 2025, 12:59am UTC](https://community.memryx.com/t/device-info-and-device-selection/53 "2025-01-29T00:59:59Z")
It would be nice to provide API to query device info for all installed MX devices: at least how many devices are available. Then, it would be nice to be able to do inference on a particular device, specifying the device… --- ## [About USB 3, maximum number of interconnection, documentation](https://community.memryx.com/t/about-usb-3-maximum-number-of-interconnection-documentation/45)
**Author:** [@AliUzel](https://community.memryx.com/u/AliUzel)\ **Replies:** 3\ **Last updated:** [January 6, 2025, 11:14am UTC](https://community.memryx.com/t/about-usb-3-maximum-number-of-interconnection-documentation/45 "2025-01-06T11:14:10Z")
Hi, Are there any cons using USB 3 interface rather than PCIE? Why up to 16-chips (96 TOPS/TFLOPs) can be interconnected? What is the reason for this limit? What about documentation of MX3? Datasheet, reference schemat…