emerging solutionss22.q4cdn.com/909968316/files/doc_presentations/2015/analyst_da… · major...
TRANSCRIPT
Emerging Solutions Laura Stark
Senior Vice President and General Manager
50
Moving the world’s data from memory through interfaces
Memory
& Interfaces
Reinventing embedded security from silicon to cloud
Cryptography
Research
Reimagining computing from sensor to cloud
Emerging
Solutions
What is ESD?
Foundational R&D Invention and harvesting
Methods and architecture Incubation from concept to market
51
0
500
1000
1500
2000
2500
3000
3500
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Sum of Granted
Sum of Pending
Invention Happens Here…
Source: Rambus
Large worldwide portfolio with >2400 active patents and applications
YTD
52
Rambus Portfolio is Widely Cited
Source: Innography (Aug-2015)
53
In Technologies that …
Source: Innography (Aug-2015)
54
Key Programs
Next-Gen Memory Computational Sensing
New technologies and
architectures to improve memory performance and
bring compute closer to memory
New smart sensors that
enable low cost, low power light weight
sensing everywhere
55
Challenges Facing the Industry
Limits of DRAM scaling
Number of CPU cores scaling faster than
memory can support
Low CPU Utilization Big Data Analytics
Cost per bit of DRAM no longer scaling with
process
Massive and growing data sets are straining
data center architecture
90nm
0.25um
0.18um
0.13um 65nm 45nm
0.35um
1990 1995 2000 2005
Gate Oxide
Thickness (nm)
10
1
The end of traditional scaling
56
• Continued development of DDR5+ concepts
• Development of Gen n+2 ideas for buffer chips
• Key contributions to IP portfolio for future licensing
Advanced Development for Future Memory
Processing
Memory
Storage
57
• Faster than flash; lower cost and more reliable than DRAM
• Addresses slow-down in DRAM cost reduction
• Alternative bit-cell technology to improve memory endurance and power efficiency
• Rambus crosspoint ReRAM IP and technology
Emerging Storage Class Memory
Processing
Memory
Storage
58
• Exploring alternatives to accelerate the delivery and computation of data
System architectures
Application and Platform Software
FPGA acceleration
• Compatible with existing data center hardware and software infrastructure
New Memory Architecture: Smart Data Acceleration
Processing
Memory
Storage
59
Rambus Reveals Smart Data Acceleration Research Program
Advanced development platform slated to improve performance and power efficiency for data centers
SUNNYVALE, Calif. – September 14, 2015 – Rambus Inc. (NASDAQ:RMBS) today announced the Smart Data Acceleration (SDA) Research Program designed to tackle major issues facing data centers in the age of Big Data. The SDA Research Program has been exploring new architectures for servers and data centers that are optimized for rack-level Big Data computation, targeting significant improvements in performance and power efficiency, as the industry brings computing closer to data. “Modern servers are out of balance with today’s needs – data centers are under stress due to escalating demands of real-time access to large amounts of information driven by Big Data and new applications,” said Laura Stark, senior vice president and general manager of the Emerging Solutions division at Rambus. “This research platform focuses on architectures that offload computing closer to very large data sets at multiple points in the memory and storage hierarchy.” As part of the program, Rambus has created a platform to investigate system architectures that include software, firmware, FPGAs and large amounts of memory. The platform can be used to test new methods to optimize and accelerate data analytics for extremely large data sets.
With use cases including real-time risk analytics, ad serving, neural imaging, transcoding and genome mapping, the Rambus SDA Research program is a key technology investment for the next generation of data centers.
NEWS RELEASE
Smart Data Acceleration Research Program
60
The Key Building Block of the SDA Research Program
61
Focus Areas for Improving System Performance P
erfo
rman
ce
DRAM vs. SSD • Lower latency under load improves
application performance; benefits increase as workload demands rise
Minimizing Data Movement • High memory capacity reduces the time
and power spent moving data back and forth across networks
System with SSD
System with SDA
Compute Offload and Acceleration • FPGAs enable parallel offloading and
acceleration of compute tasks next to the memory storing the data
62
Lensless Smart Sensor
0.055mm ~1.5mm
Commercial Sensor Lens LSS Spiral Grating
Top view
Cross-section view
63
“Rambus on-chip camera will lead to finer gesture recognition”
- CNET
“Tiny sensors such as these are going to be used in always-on, always-connected smart devices that are going to improve every part of our daily lives”
- Tom’s Hardware
“An itty-bitty camera could bring sight to the Internet of Things”
- MIT Technology Review
“A microscopic lens-free image sensor could turn anything into a camera”
- Gizmodo
64
Partners in Open Development
• Partnering with the industry to accelerate and amplify discovery of applications for our technology
• Collaboration with partners to develop MVP (minimal viable product) prototypes
POD Program
Universities
frog IXDS
65
• Hardware development kits featuring LSS test chip built on open-source hardware and software
• Raspberry PI platform
• POD kits include:
Maker boards with grating/sensor, program brief & instructions, apps + firmware, SDK, imaging toolkit, tutorials and documentation
POD Development Kits
66
POD in Action
67
• Image change detection
• Point tracking
• Range finding
• Sophisticated gesture recognition
• Object recognition
• Image capture
• Video streaming
Breadth of Capabilities
68
• From inventing to prototyping to incubating, ESD is growing our technology foundation
• Growth in Big Data and the number of connected devices is creating difficult challenges and tremendous opportunities
New memory types and architectures addressing the performance and capacity gaps in memory hierarchy
Fundamentally new approach to capture data for ubiquitous sensing
• Collaborative engagement models open new doors for technology creation and adoption
Supporting Growth through Emerging Solutions