BSC Researchers Shrink Floating Point Formats to Accelerate Deep Neural Network Training

By Ken Strandberg

April 15, 2019

Sometimes calculating solutions as precisely as a computer can wastes more CPU resources than is necessary. A case in point is with deep learning. In early stages of training a Deep Neural Network (DNN), a lot of guesswork goes on. The algorithm assigns random values to the weights and computes the error. But the error is enormous in the beginning, and the values of the weights are a long way from the ones selected at the end. Representing weights as a 32-bit floating-point number is costly in terms of processing, yet most of the bits of the mantissa are not needed in early training. As training progresses and it hones the value of the weights, then greater precision becomes important in order to optimize the solution.

Using reduced precision floating point number formats offers benefits in memory footprint and bandwidth and in processing time, which can translate to power savings. These savings can possibly be significant if the benefits can be scaled out to accommodate training of massive DNNs. But will less precision affect overall accuracy of the training?

A lot of research in reduced precision for AI training and inferencing has gone on over the last year. Across Europe and the U.S., industry, academia, and research institutions are looking at this aspect of AI, including U.S. National Labs, Google, and Microsoft. Thus far, the work has resulted in papers, proposals, and some code. Google’s experiments with DNNs have shown that reducing the mantissa in 32-bit floating point numbers for certain calculations of DNNs is okay, “as long as you can represent tiny values closer to zero as part of the summation of small differences during training” (https://en.wikichip.org/wiki/brain_floating-point_format).

Google integrated the bfloat16 format, which provides the same size exponent as the IEEE standard 32-bit FP (float32) but with a smaller mantissa, into some of its products. Bfloat16 is being implemented in a range of future Intel processors for AI deep learning applications.

Intel has integrated a reduced representation format into the Vector Neural Network Instruction (VNNI), a part of Intel Deep Learning Boost (DL Boost), added to the Intel Advanced Vector Extensions 512 instruction set in 2nd generation Xeon Scalable processors.

But the jury is still out on which numbering format or code is best to use at different stages of training and for inferencing. What are the benefits to be gained, in terms of processing performance and power, for the different formats used? And what conditions tell a developer the best format or code to use and when? These are all areas of great interest to Marc Casas, Senior Researcher, at Barcelona Supercomputing Center (BSC).

“We believe dynamic numerical precision approaches offer the best benefit to training and inferencing,” stated Casas. “We are evaluating the applications of many formats and codes, including Intel DL Boost (such as VNNI and others), 32-bit and 64-bit floating point, Flexpoint, and integer formats, at various phases of training neural networks and inferencing.” Flexpoint is a format proposed by Intel for tensors and will be integrated in its Nervana Neural Network processors.

Casas and his team, including John Haiber Osorio Rios and Marc Ortiz of BSC, expect to identify at what phase of training it is best to apply different numerical presentations and how they benefit the network evolution without loss of accuracy. They will also study their impact on processor performance and power consumption on Intel hardware. But, understanding when to use an appropriate format and the impact on the hardware is only one aspect.

“We propose to not only develop innovative ways to exploit the potential of DL Boost and these numerical representations, but to dynamically adjust the Flexpoint/Bfloat16 formats to determine which DL Boost instructions to apply at different phases of training,” add Casas. “We will develop an algorithm to drive these dynamic adjustments based on different proxies describing the network evolution. These adaptive and dynamic schemes used for learning or inferencing phases of DNNs will make it possible to switch across different precisions on runtime.”

Casas says their baseline models are built on Alexnet and Resnet using the Imagenet data set. The project will use software emulations and eventually be applied and evaluated on Intel hardware designed to implement the numerical formats as the next-generation Intel silicon becomes available.

In 2017, BSC installed MareNostrum4, a large supercomputing cluster from Lenovo built on Intel Xeon Scalable processors and Intel Omni-Path Architecture fabric. Casas and his team will use MareNostrum4 to help them answer these questions.

“Understanding the use of dynamic numerical formats and developing schemes to apply them with will change the way industry is training networks,” concluded Casas. “Our work will shed light on allowing a more flexible training mechanism. We will look for ways to apply it to DNN Frameworks, like Intel’s version of Caffe and TensorFlow, so everyone can use it.”

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

Quantinuum Reports 99.9% 2-Qubit Gate Fidelity, Caps Eventful 2 Months

April 16, 2024

March and April have been good months for Quantinuum, which today released a blog announcing the ion trap quantum computer specialist has achieved a 99.9% (three nines) two-qubit gate fidelity on its H1 system. The lates Read more…

Mystery Solved: Intel’s Former HPC Chief Now Running Software Engineering Group 

April 15, 2024

Last year, Jeff McVeigh, Intel's readily available leader of the high-performance computing group, suddenly went silent, with no interviews granted or appearances at press conferences.  It led to questions -- what's Read more…

Exciting Updates From Stanford HAI’s Seventh Annual AI Index Report

April 15, 2024

As the AI revolution marches on, it is vital to continually reassess how this technology is reshaping our world. To that end, researchers at Stanford’s Institute for Human-Centered AI (HAI) put out a yearly report to t Read more…

Crossing the Quantum Threshold: The Path to 10,000 Qubits

April 15, 2024

Editor’s Note: Why do qubit count and quality matter? What’s the difference between physical qubits and logical qubits? Quantum computer vendors toss these terms and numbers around as indicators of the strengths of t Read more…

Intel’s Vision Advantage: Chips Are Available Off-the-Shelf

April 11, 2024

The chip market is facing a crisis: chip development is now concentrated in the hands of the few. A confluence of events this week reminded us how few chips are available off the shelf, a concern raised at many recent Read more…

The VC View: Quantonation’s Deep Dive into Funding Quantum Start-ups

April 11, 2024

Yesterday Quantonation — which promotes itself as a one-of-a-kind venture capital (VC) company specializing in quantum science and deep physics  — announced its second fund targeting €200 million. The very idea th Read more…

Exciting Updates From Stanford HAI’s Seventh Annual AI Index Report

April 15, 2024

As the AI revolution marches on, it is vital to continually reassess how this technology is reshaping our world. To that end, researchers at Stanford’s Instit Read more…

Intel’s Vision Advantage: Chips Are Available Off-the-Shelf

April 11, 2024

The chip market is facing a crisis: chip development is now concentrated in the hands of the few. A confluence of events this week reminded us how few chips Read more…

The VC View: Quantonation’s Deep Dive into Funding Quantum Start-ups

April 11, 2024

Yesterday Quantonation — which promotes itself as a one-of-a-kind venture capital (VC) company specializing in quantum science and deep physics  — announce Read more…

Nvidia’s GTC Is the New Intel IDF

April 9, 2024

After many years, Nvidia's GPU Technology Conference (GTC) was back in person and has become the conference for those who care about semiconductors and AI. I Read more…

Google Announces Homegrown ARM-based CPUs 

April 9, 2024

Google sprang a surprise at the ongoing Google Next Cloud conference by introducing its own ARM-based CPU called Axion, which will be offered to customers in it Read more…

Computational Chemistry Needs To Be Sustainable, Too

April 8, 2024

A diverse group of computational chemists is encouraging the research community to embrace a sustainable software ecosystem. That's the message behind a recent Read more…

Hyperion Research: Eleven HPC Predictions for 2024

April 4, 2024

HPCwire is happy to announce a new series with Hyperion Research  - a fact-based market research firm focusing on the HPC market. In addition to providing mark Read more…

Google Making Major Changes in AI Operations to Pull in Cash from Gemini

April 4, 2024

Over the last week, Google has made some under-the-radar changes, including appointing a new leader for AI development, which suggests the company is taking its Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

Leading Solution Providers

Contributors

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Eyes on the Quantum Prize – D-Wave Says its Time is Now

January 30, 2024

Early quantum computing pioneer D-Wave again asserted – that at least for D-Wave – the commercial quantum era has begun. Speaking at its first in-person Ana Read more…

GenAI Having Major Impact on Data Culture, Survey Says

February 21, 2024

While 2023 was the year of GenAI, the adoption rates for GenAI did not match expectations. Most organizations are continuing to invest in GenAI but are yet to Read more…

Intel’s Xeon General Manager Talks about Server Chips 

January 2, 2024

Intel is talking data-center growth and is done digging graves for its dead enterprise products, including GPUs, storage, and networking products, which fell to Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire