HP Innovation Journal Issue 12: Summer 2019 | Page 67

Developing algorithms and dealing
with errors by hand is expensive.
simple problems can be trained quickly, and problems that
are otherwise impossible just take longer.
VIRTUAL MACHINES AND FACTORIES
The rise of cyber-physical systems with volumes of
machine-generated data enable virtual machines (VMs).
A VM, also called Digital Twin in the industry, is a
holistic, model-based representation of a physical system
with all of the functional physical attributes that mimic
the operation of the machinery. An example of a physical
system is a 3D Multi Jet Fusion printer.
During development, VM system-scale simulations can
reduce time for product development by 33%. 5 VM facto-
ry-scale simulations of deployment environments, such
as additive manufacturing operations, can help customers
predict the total cost of ownership (TCO) based on vari-
ous workflows and service level agreements.
At runtime, data collected in the factory and VM models
enable customers to drive real-time efficiency, strive to main-
tain the TCO for various material and energy flows that are
part of manufacturing, and get to root causes of anomalies.
COMPUTE’S BRAVE NEW WORLD
As an industry, we quest tirelessly
for technologies that enable us to
work more efficiently, sustainably,
and productively. Innovation does
not happen in a vacuum, and
often advancements in one area
(IoT, sensors, monitors) cause
complexities downstream
(the resulting deluge of data).
HP’s ability to research and tackle challenges from
multiple angles at once often results in branched but com-
plimentary approaches—like the three areas of innovation
described above. Turning raw data into actionable infor-
mation has always been important, but perhaps never in
our history has the available data been vaster and the need
to process it more critical. With rising energy demands,
we’ll increasingly turn our focus toward lowering con-
sumption and maximizing efficiency. In this effort, and in
many others, new compute architectures and technologies
are essential building blocks for our data-driven future.
1. A network edge is defined as one or fewer network hops
from the source of the data.
2. Evidence of this includes delays in Intel’s 10nm (nanometer) CPUs.
3. Next-generation 4nm transistors are hardly wide enough for 20 silicon atoms.
4. IDC
5. Reductions are achieved by reducing physical iterations, and VMs enable “what
if” analysis of new future components.
65