rasp pi 3 B+ with best compute stick in 2022?

[bartlino]

What is the current model for best compute stick for a rasp pi B+ running a Buster image and pi camera?

Is this tutorial outdated for installing the Intel Movidius Neural stick?

https://pyimagesearch.com/2018/02/12/getting-started-with-the-intel-movidius-neural-compute-stick/

Thanks

[Megat_Intel]

Hi Bartlino,

Thank you for reaching out to us.

The current Intel® Neural Compute Stick model is the Intel® Neural Compute Stick 2 (Intel® NCS2). However, Intel® is discontinuing the Intel® Neural Compute Stick 2 (Intel® NCS2) as mentioned here.

The suggested immediate replacement product for NCS2 is MX HDDL M.2 cards.

https://www.intel.com/content/www/us/en/developer/topic-technology/edge-5g/hardware/vision-accelerat...

On the other hand, the tutorial for Intel® Movidius™ Neural Compute Stick that you provided uses the Movidius API which was replaced by the OpenVINO™ Toolkit software.

The latest compatible Raspbian OS OpenVINO™ toolkit package for Intel® Movidius™ Neural Compute Stick is the 2020.3.2 LTS release. You can download the l_openvino_toolkit_runtime_raspbian_p_2020.3.355.tgz package here and check out the installation guide.

Regards,

Megat

[bartlino]

Hopefully this doesn't sound like a silly question but would there be anything that could assist a Windows OS on CPU? or are the neural compute sticks only meant for IoT applications running Linux?

Is there an Intel IoT device that can be ordered as a package with some sort of compute stick?

I see there is also data science work stations:

https://www.intel.com/content/www/us/en/products/systems-devices/workstations/data-science-workstations.html

What would be the most cost effective computer listed and where to buy?

Thank you for the time in response.

[Megat_Intel]

Hi Bartlino,

Intel® Neural Compute Stick is a plug & play AI device for deep learning inference at the edge. It can be used to run inference on Windows OS. To assist the CPU, check out Multi-device execution which shares the inference workload across multiple devices and helps to increase performance stability. 

For Intel® IoT devices, you can explore Intel® Solutions Marketplace to discover products and solutions for your needs.

Regarding the workstations, you can refer to our Benchmark results of OpenVINO™ inferencing on different Intel® hardware as a reference for performance. You can compare the price based on your required configuration and click on the respective URL for each brand. It will take you to the brand's external websites for purchasing.

Regards,

Megat

[bartlino]

Hi,

Would there be any chance you could give me a link where to buy the Intel® Neural Compute Stick to run deep learning inference at the edge that is operating on a Windows OS? Would the same stick run on Ubuntu server or Ubuntu desktop 20.04 LTS?

I just happened to be testing some concept ideas on a Windows OS but interested in converting it Linux if the results would be better, like faster inference and FPS.

Is it this product?

https://www.amazon.com/dp/B08G1CCWN5?tag=digitrend08sub4-20&th=1&_keeptag=1&ascsubtag=16667915762577oq8&geniuslink=true

Thank you for your time in response....

[Megat_Intel]

Hi Bartlino,

Intel® Neural Compute Stick 2 will run on Ubuntu 20.04 LTS but it is not validated to run on Ubuntu Server.

For purchasing, Intel® is discontinuing the Intel® Neural Compute Stick 2 (Intel® NCS2) with the Last Product Discontinuance Shipment Date on 30 June 2022. You can read more about the Discontinuation Notice here. We apologize for the inconvenience.

For your information, the product you mentioned is not the Intel® Neural Compute Stick 2. I share the Intel® Neural Compute Stick 2 image below.

On the other hand, you can refer to the Intel® Distribution of OpenVINO™ Toolkit System Requirements to check out other Intel® hardware that is supported for OpenVINO™.

Regards,

Megat

[bartlino]

Sorry to be confused about if all VPU's are being discontinued and no longer being manufactured by Intel? Is that correct?

But here on Intel's website for VPU and in the screenshot below is this referring to one model being discontinued (Intel® Movidius™ Neural Compute Stick 2) and another model still being supported on Windows 10? (Intel® Movidius™ Myriad™ X VPU) Which stick could I buy that is not to be discontinued?

[Megat_Intel]

Hi Bartlino,

The Discontinuation Notice is only for Intel® Neural Compute Stick 2 (Intel® NCS2). You can check out Intel® Vision Accelerator Design for other available VPU boards powered by Intel® Movidius™ Vision Processing Unit (VPU).

For your information, Intel® NCS2 is also powered by Intel Movidius Myriad X Vision Processing Unit (VPU).

Please note that while Intel® NCS2 uses USB 3.0 Type-A connectivity, other VPU boards have different connectors such as M.2 2230, M-PCIe, and single-slot PCIe*. To find out which board suits your needs, refer to the detailed Specifications of Intel® Vision Accelerator Design.

To purchase the VPU boards from the respective vendors, the details are available in the Suppliers section.

Regards,

Megat

[bartlino]

Hi @Megat_Intel ,

So just to triple confirm, thank your bearing with me....an edge device AI accelerator with the USB 3.0 (NCS2) type connections, these VPU's are all discontinued and the only other AI accelerator not discontinued is physically connected with only M.2 2230, M-PCIe, and single-slot PCIe*.

Sorry if this sounds like a silly question but do any of the non-USB VPU's work directly with a rasp pi? Or any of these embedded devices I also noticed programmable with the Arduinio IDE:

https://www.intel.com/content/www/us/en/developer/topic-technology/edge-5g/hardware/fpga-xlr8.html

https://www.intel.com/content/www/us/en/developer/topic-technology/edge-5g/hardware/fpga-hinj.html

Also curious how one could work with the openVINO toolkit and embedded devices like the links above. Thank you

[Megat_Intel]

Hi Bartlino,

Yes, you are correct. Intel® Neural Compute Stick 2 (Intel® NCS2) which uses USB connectivity is discontinued while some other Intel® VPU boards using M.2 2230, M-PCIe, and single-slot PCIe* connectors are still available.

The VPU boards might not work directly with Raspberry Pi since Raspberry Pi does not have M.2, M-PCIe, or Single-slot PCIe.

On the other hand, the embedded devices you mentioned are using Arduino Integrated Development Environment (IDE) which is not supported on OpenVINO™ Toolkit.

Regards,

Megat

[bartlino]

Thank you so much!!

[Megat_Intel]

Hi Bartlino,

This thread will no longer be monitored since we have provided a suggestion. If you need any additional information from Intel, please submit a new question.

Regards,

Megat