In our recap of the final day of International Supercomputing 2016, we explore how OpenPOWER members are working in the HPC industry and continue the conversation around cognitive computing, deep learning, and machine learning.
2. During the third
and final day of
ISC, we learned
about the role that
our OpenPOWER
members play in
Industry, and what
they can enable
for businesses.
3. Customers need choice, and
Samsung works with ecosystems
like OpenPOWER so HPC users
have the ability to choose between
implementation options, and
aren’t locked into single source
solutions.
ThomasArenz
DirectorofStrategicMarketing, EMEA
Samsung Semiconductor
4. These systems thrive by
getting data into the
hardware, and our goal as a
systems provider is to feed
the beast.
ScottSoutter
Product ManagerHPCSystems,
IBM
6. The combination of POWER8's
processing strengths,
OpenPOWER technology, and
SoftLayer's flexibility allows clients
to rapidly deploy Linux workloads
on bare metal systems.
BradBrech
PowerSystemsDistinguished Engineer
IBM
8. Hadoop’s infrastructure is
similar to what we’re doing in
HPC, but it’s doing high
performance data analytics for
industry.
JanWender
HPCServicesandSoftware,TeamLeader
Atos
9. But the growth of
cognitive
computing, deep
learning, and
machine learning
remained a key
driver of
conversations.
10. The ability to choose from
platforms, runtimes and
deployment models
allows the creation of
purposeful architectures
that will define digital and
cognitive success.
CalistaRedmond
President
OpenPOWER Foundation
11. We believe
OpenPOWER will be
very important for the
future of cognitive
computing.
Alessandro Curioni
VicePresident, EuropeandDirector
IBMResearch
12. We mean for cognitive
computing to be a
revolution, and for a
revolution you need an
ecosystem.
Alessandro Curioni
VicePresident, EuropeandDirector
IBMResearch
13. When you have your data you start
training your deep learning
network. It’s an iterative process
that takes a long time and you
need powerful hardware.
JeroenBedorf
SeniorSystemArchitect
Minds.ai
14. The biggest application of
supercomputing will not be in dense
matrix floating point operations or large
data processing for pattern recognition,
it will in fact be symbolic computing,
the creation of machines that think.
More importantly, machines that
understand.
ThomasSterling
ProfessorofIntelligent SystemsEngineering
Indiana University
15. But as we continue to ask
more and more questions
about our world, and the
problems keep changing,
even greater challenges
for HPC are on the
horizon that will need
technologies we haven’t
even dreamt of… yet.
16. 118
10billion
The years that it will
take an Exaflop
supercomputer to
solve a binary
optimization problem
with 512 variables.
Rupak Biswas
DirectorofExploration Technology Directorate
NASA AmesResearchCenter
17. This concludes our coverage of ISC
2016. Please be sure to follow us
on Twitter at @OpenPOWERorg
and share our coverage with your
colleagues in the HPC industry.
Until next year: Auf Wiedersehen!