Scribd Logo
Upload

Welcome to Scribd!

  • 11 views

    BRKFP292

    Uploaded by

    Steph Deloison

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd

    BRKFP292

    Uploaded by

    Steph Deloison
    11 views15 pages

    Copyright

    © © All Rights Reserved

    Available Formats

    PPTX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    11 views15 pages

    BRKFP292

    Uploaded by

    Steph Deloison

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    of 15

    Microsoft Build

    AI Everywhere –
    Accelerate your development
    from the PC to cloud
    Session: BRKFP292
    This session will be recorded and posted for on
    demand viewing on the public event website.

    By joining this session, you are consenting to be


    recorded and your name, image, likeness, voice,
    or other personal indicia could be part of the
    recording for future use by Microsoft as described
    in this event’s Terms and Conditions, available
    for review from your registration record.

    Should you not wish to be recorded,


    you are welcome to leave the session
    and view it on demand later.
    Todays panel speakers

    Brian Rogers Brian Ehlert Lior Kamrat Ken Koyanagi


    Solutions Architect Director, Product Mgmt. Principal Product Mgr. lead AI Solutions Engineer
    Intel F5 NGINX Microsoft Intel
    The Xeon AMX instructions for AI

    The first step is to check if Intel AMX is enabled


    on your hardware. On the bash terminal, enter
    the following command: cat /proc/cpuinfo

    Alternatively, you can use: lscpu | grep amx

    Check the “flags” section for amx_bf16 and


    amx_int8. If you do not see them, consider
    upgrading to Linux* kernel 5.17 or newer.
    Your output should look something like this:
    Real World AI
    How this works in ‘real world AI’

    Areas of Growth
    Efficiency. Real world AI performance isn’t just about TOPS and FLOPS – it’s about bringing
    the right assets to the right place at the right time – and that’s what Intel does. For example,
    with AI PC it’s the combined power of all 3 xPUs working together.

    Security. AI without security isn’t AI – it’s just risk. Intel brings a higher level of security to AI solutions.

    Developer ready. With an AI PC you can buy an off the shelf laptop, download Pytorch
    and developers can start coding – we’ve invested in developing all the drivers and other elements needed.

    Hybrid AI – AI solutions don’t just live in the cloud or the client or the edge. Comprehensive solutions can span
    all three areas. Intel supports all of these areas. From the latest Azure Instances with Deep Learning Boost and
    AVX-512 to support for PyTorch, Direct ML, WebGPU, Transformer.js, and driving the WebNN standards – Intel AI
    is everywhere you need to be.

    Ecosystem – we haven’t started our AI journey alone. With partners like Microsoft and our decade
    of work together on adding AI to Office to partners like F5 that help customers lift and shift workloads
    onto the cloud, we’ve created a strong ecosystem of partners who help make your AI journey on Intel easier.
    Store cameras

    Store AKS Edge Essentials cluster (Arc-enabled)

    Checkout Queue
    Queue Monitoring Local PostgreSQL
    AI API service database server

    Monitoring service
    architecture 4 3

    Queue Monitoring Queue Monitoring


    frontend service backend service
    1 2

    Contoso
    Supermarket store
    manager
    Leverage ONNX
    Runtime with DirectML
    and OpenVINO™ EP to
    accelerate Gen AI on PC
    What is an AI PC?
    A PC with new NPU silicon that brings new AI
    experiences in productivity, creativity, and security
    through a combination of the CPU, GPU, and the NPU.

    Three AI Engines

    GPU NPU CPU


    Heterogenous execution of AI
    workloads embraces the best High Throughput Low Power Fast Response
    Ideal for AI- Ideal for sustained AI Ideal for
    practices in AI software design. accelerated digital workloads and AI low-latency AI
    content creation offload for battery life workloads

    Intel, the Intel logo, and Intel Core are trademarks of Intel Corporation or its subsidiaries.
    Demo
    Session Resources

    Next steps
    Resources
    Related sessions

    Find all of this and more


    on the session details page

    aka.ms/build/seattle/sessions
    © Copyright Microsoft Corporation. All rights reserved.

    You might also like