high-performance computing
precision medicine is helping to save the lives of children and adults who have exhausted the possibilities of traditional medicine,’ stated Conway. ‘We will soon be adding precision medicine
to the new HPC market segments that we track which require the power of HPC.’
A second moon-shot In his 2016 State of the Union Address, president Obama called on vice president Biden to lead a new, national ‘Cancer Moonshot’ to dramatically accelerate efforts to prevent, diagnose, and treat cancer – to achieve a decade’s worth of progress in five years. In pursuit of this mission, president Obama
➤ treating complex diseases does not offer much respite to the less common forms of such diseases. Dr Walter Kibbe, director at the Center
for Biomedical Informatics and Information Technology (CBIIT), stated: ‘I think we have all been touched by cancer in different ways, but to put some numbers on it – this year, the American Cancer Society (ACS) estimates that there will be 1,700,000 new cancer cases and 600,000 cancer deaths in the United States alone.’ Kibbe explained that while the mortality
rates for cancer are falling we can still do more to accelerate the treatment further.‘We need to understand more about the interplay between biology, cancer and the data we generate in labs across the whole country. Tat is the essence, in my view, of precision medicine.’ Te arrival of post-petascale HPC systems
alongside machine learning and big data is opening up new avenues for research that were not previously within the scope of producing real-world treatments because of the huge reliance on computation. ‘We are now on the path, early in the next
decade, to produce computers that will compute 1,018 operations per second’ said Moniz. ‘We are also looking at applications of computers of that scale to big problems. Cancer is one of those big problems.’ Moniz stated that the scientists that apply HPC to the complex challenges facing humanity today will be ‘central to cracking this problem in our society.’ ‘We think this can be a revolution in how
cancer is treated,’ concluded Moniz. During the plenary session, Steve Conway,
research vice president in IDC’s high- performance computing group and chair of the session highlighted the rising importance of precision medicine and its role within the wider HPC industry. ‘Te IDC is already tracking dozens of initiatives around the world where
20 SCIENTIFIC COMPUTING WORLD
established the Cancer Moonshot Task Force charged with leveraging federal investments, targeted incentives, private sector efforts, patient engagement and other initiatives to support cancer research. Tis represents the largest effort to date to tackle the problem of cancer on patient health – never before have so many government agencies come together to tackle the challenges along the spectrum of cancer research and care to improve outcomes for patients. Te Cancer Moonshot project aims to
accelerate cancer research through five specific project aims. Te first is to catalyse scientific breakthroughs through the use of interdisciplinary approaches for elucidating the biological mechanisms underlying cancer onset and treatment. Te project also aims to align research and care as a seamless and iterative process and maximise the collection and research use of longitudinal data and biospecimens. Once this has been accomplished, the project
will move on to unleashing the power of the data that is available through the creation of a seamless data environment for clinical and research data through shared policies and technologies. Te project also aims to unlock scientific advances through open publication, storage platforms and next-generation computer architectures – this is where HPC will be crucial to driving the project forward, as huge amounts of data need to be processed and analysed in order to achieve the kind of progress that has been laid out by president Obama. Once these goals have been reached, the next
stage will be to use these tools to create new medicines and treatments, strengthen prevention and diagnosis and improve patient care.
Technology at the heart of precision medicine Underlying these new precision medicine initiatives are machine learning tools and HPC hardware. In recent years machine learning has been an increasingly popular topic in HPC
because machine learning uses technology created for HPC but applies it to applications that are not typically seen as traditional HPC verticals. Tese application areas can range from medicine to consumer spending habits – opening up new application areas for HPC research. At the heart of machine learning is the GPU.
GPU technology has been particularly suited to machine learning applications because of the inherent parallelisation of GPU technology couple with the increasing memory bandwidth which has been seen in many HPC technologies in recent years. GPUs are powerful tools for machine learning
because they are particularly well suited to training deep neural networks used in machine learning algorithms. Nvidia has now started to develop products that are specifically optimised for deep learning such as the Pascal P100 and the DGX-1. Te DGX-1, heralded by Nvidia as a ‘deep
learning supercomputer in a box’, comprises eight Pascal P100 GPUs connected through Nvidia’s proprietary interconnect NV Link. In addition to technologies provided by Nvidia
many other companies are now developing products specifically aimed at deep learning and machine learning technology. At SC16 Xilinx announced a new range of 16nm FPGA
IN RECENT YEARS MACHINE LEARNING HAS BEEN AN INCREASINGLY POPULAR TOPIC IN HPC
products that include HBM memory and the latest 16nm fabrication process to significantly increase performance of these accelerators. While adoption of FPGAs is not as widespread
as Nvidia GPUs, FPGA technology has focused on the difficulty with coding the technology; however, this could change if Xilinx can help accelerate adoption through libraries and other tools to facilitate easier programming. Tis is particularly important as machine learning workloads are fairly monolithic when compared to general purpose HPC so FPGAs could become highly optimised machine learning accelerators very quickly. Te US government has invested $210 billion
in recent years through its Precision Medicine initiative, and earlier this year announced the Cancer Moonshot initiative in an effort to improve cancer treatments. While it is not yet clear which hardware will be the standard for machine learning, it is clear that this technology will be instrumental in treating complex diseases. l
www.scientific-computing.com
f11photo/
Shutterstock.com
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32
