Key Notes of IDF 2016 in San Francisco, CA

[SergeyKostrov]

*** Key Notes of IDF 2016 in San Francisco, CA *** This year I've attended IDF 2016 and for me it was the most productive and interesting. So, here are my notes and I hope it will be interested for everybody to learn what topics are the hottest for Intel.

[SergeyKostrov]

Intel related: - Intel 7th Generation CPU ( Xeon Phi / Knights Mill ) - Intel Stratix 10 FPGA 14nm Tri-Gate process ( Sampling Q4 2016 / ~30 billion transistors ) - Intel Silicon Photonics - Intel Systmem Studio for Microcontrollers - Intel Instrumentation and Trace API ( IIT API ) - Intel MKL DNN - Intel Deep Learning SDK - Intel Performance SnapShot - Intel SNAP FPGA related: Types of FPGAs: - Discrete - Co-Package - On-Die Vendors: - Altera FPGA - Microsoft Catapult V1 Card FPGA - Google TPU Card ASIC Other: - VeriLog / VHDL Types of processing: - CPU-bound - Memory-bound - I/O-bound ( aka Disk-bound ) Abbreviations: - DC - Data Center - HBM - High Bandwidth Memory - HDF5 - Hierarchical Data Format v5 - NN - Neural Network - DNN - Deep Neural Network - ML - Machine Learning - DL - Deep Learning Storage related: - Capacity / Low Latency / Power Consumption Generic: - CentOS v7.2 - Code-Based Algorithm - Data-Driven Algorithm - SpGEMM - Numerical Library PETSC - Rack Scale / QCT / Inspur - Nervana - TensorFlow Web-links: . http://www.intel.com/idfsessionsSF http://github/hfp/libsxmm

[RBang]

Sergey Kostrov wrote:

Intel related:

- Intel 7th Generation CPU ( Xeon Phi / Knights Mill )
- Intel Stratix 10 FPGA 14nm Tri-Gate process ( Sampling Q4 2016 / ~30 billion transistors )
- Intel Silicon Photonics
- Intel Systmem Studio for Microcontrollers
- Intel Instrumentation and Trace API ( IIT API )
- Intel MKL DNN
- Intel Deep Learning SDK
- Intel Performance SnapShot
- Intel SNAP

FPGA related:

Types of FPGAs:
- Discrete
- Co-Package
- On-Die

Vendors:
- Altera FPGA
- Microsoft Catapult V1 Card FPGA
- Google TPU Card ASIC

Other:
- VeriLog / VHDL

Types of processing:

- CPU-bound
- Memory-bound
- I/O-bound ( aka Disk-bound )

Abbreviations:

- DC - Data Center
- HBM - High Bandwidth Memory
- HDF5 - Hierarchical Data Format v5
- NN - Neural Network
- DNN - Deep Neural Network
- ML - Machine Learning
- DL - Deep Learning

Storage related:

- Capacity / Low Latency / Power Consumption

Generic:

- CentOS v7.2
- Code-Based Algorithm
- Data-Driven Algorithm
- SpGEMM
- Numerical Library PETSC
- Rack Scale / QCT / Inspur
- Nervana
- TensorFlow

Web-links:
.
http://www.intel.com/idfsessionsSF

http://github/hfp/libsxmm

Hi Sergey,

Thanks for sharing your notes. I feel :

Intel Deep Learning SDK

Intel Performance SnapShot

Intel Aero Ready-to-Fly Drone

Intel Project Alloy VR headset and

Intel Joule Board

are the hot topics for Intel Dev community.

Moreover, wanted to point out the link you shared - http://github.com/hfp/libsxmm , lands on a 404 error page.

[SergeyKostrov]

>>Moreover, wanted to point out the link you shared - http://github.com/hfp/libsxmm, lands on a 404 error page. I confirm that and there is nothing I can do regarding it. I also tried: . https://github.com/hfp/libsxmm . and there is the same problem - web page is Not avalable.

[SergeyKostrov]

Updated list of hot topics of IDF 2016 related to Intel from the feedbacks of two IDZ users - Rishabh Banga and Sergey Kostrov. Intel related: - Intel 7th Generation CPU ( Xeon Phi / Knights Mill ) - Intel Stratix 10 FPGA 14nm Tri-Gate process ( Sampling Q4 2016 / ~30 billion transistors ) - Intel Silicon Photonics - Intel Deep Learning SDK - Intel MKL DNN - Intel Aero Ready-to-Fly Drone - Intel Project Alloy VR headset - Intel Joule Board - Intel System Studio for Microcontrollers - Intel Instrumentation and Trace API ( IIT API ) - Intel Performance SnapShot - Intel SNAP

[Janko__Bayncore_]

Hi there, the correct link to matrix library on GitHub is https://github.com/hfp/libxsmm (xs switcheroo) and it is actually highly optimized and recommended.

Cheers!

[Hans_P_Intel]

Janko (Bayncore) wrote:

Hi there, the correct link to matrix library on GitHub is https://github.com/hfp/libxsmm (xs switcheroo) and it is actually highly optimized and recommended.

Cheers!

That's the correct URL. Thank you Janko!

Related to the IDF'16 topic, I also want to point out that we changed our summary:

Library targeting Intel Architecture (x86) for small dense and sparse matrix multiplications as well as small convolutions.

The "small convolutions" may help with emerged workloads such as deep learning and more specifically Convolutional Neural Networks (CNN). The latter are pushing the limits of today's hardware, and one of the expensive kernels is a small convolution with certain kernel sizes (3, 5, or 7) such that calculations in the frequency space are not the most efficient method when compared with direct convolutions. LIBXSMM's current support for convolutions aims for an easy to use invocation of small (direct) convolutions, which are intended for CNN training and classification.

[SergeyKostrov]

>>...it is actually highly optimized and recommended... Simply to let you know that IPP library is in "business" for more than 20 years and optimized for almost all Intel CPUs released during that period of time. I really Do Not see any reason why another 3rd party dependency should be introduced to a project if it is, for example, uses Intel IPP's convolution functions. My point of view is Modern Software Systems are over-complicated by 3rd party dependencies. Just take a look at dependencies of a latest Intel's "software miracle" MKL DNN.