1 igolgi Inc .
Dec 29, 2015
1igolgi Inc .
2
igolgi has developed native multi-core parallel video compression and processing solutions igolgi solutions are 100% software based architectures allowing customers to define optimized features that meet their video workflow needsigolgi solutions run on standard, low-cost, off the shelf, IT computers enabling leading price/performance for any video compression application
igolgi Inc .
3
We believe IT equipment can provide a significant price/performance
advantage for video applications for Broadcast, Internet, and Mobile
Markets
Video Equipment is traditionally built with inflexible hardware platforms
Such as FPGAs, ASICs, DSPs, etc.
Traditional platforms provide static features, limited video quality,
and very limited flexibility
igolgi is building IT platform based Digital Video Processing Solutions
that provide a breakthrough in price/performance, flexibility and novel
features
igolgi Inc .
4igolgi Inc .
igolgi Inc. was founded in early 2007, self-funded to date
Management Jeffrey Cooper, Chairman and Chief Executive Officer
Over 18 years design experience of video compression technology MSEE Purdue Prior Experience
Compression Research Director and GM R&D Thomson. Princeton, NJ Key contributor to MPEG2 receiver and IC implementations for several industry firsts
including satellite, DVD , and HDTV. Key contributor to early H.264 implementations in the industry and several standard
contributions Over 40 pending or issued patents in the fields of video compression and signal processing
Dr. Kumar Ramaswamy, President Over 17 years of product and business development PhD, Rensselaer Polytechnic Institute Prior Experience
CTO, Thomson Inc’s $4 Billion Systems Division including Grass Valley and Consumer Products
Head of Corporate Research, Thomson Inc, a 440 person organization 66 US patents, over 80 patents pending Strategic Advisor and Board Member of a number of start ups
5igolgi Inc .
6
Video DistributionContinues to expand into larger formats and increased compression complexity such as H.264, 1080p, and 3D
At the same time, the IT industry continues to evolvemore powerful CPUs that have the capability to supplant traditional dedicated hardware solutions for video
Multi-Core finally provides a software platform that can provide massive benefits for video equipmentigolgi
Inc .
7
Reduce power requirements
Reduce cooling cost (less heat and less airflow)
Reduce weight Decreased footprint Decreased
footprint
Reduce the ‘churn’ needed to bring on new
technology
Reduce points of failure and increase RAS
(reliability, availability, serviceability)
N + 1 redundancy offers large savings over N+N
Speed deployment and future scalability
Drive out cable complexity
(e.g. remove 128 cables/chassis)Drive out costs to reduce Total Cost of
Ownership (TCO)
Blades are significantly more efficient form factor :
igolgi Inc .
8igolgi Inc .
9
1.00
10.00
100.00
0.7 0.75 0.8 0.85 0.9 0.95 1
Amdahl's LawPerformance increase for a 256 core
processor vs algorithm parallization percentage
Modern Computer Architectures are multi-core (less than 10 cores)
Over the next 5 years, multi-core will become many-core” with 10s of cores
It is a technical challenge to optimize video Performance on many core processors.
According to Amdahl’s law from computer science, a high percentage of the algorithm must be able run in parallel. The performance of applications that do not have a high level of parallelism is poor due to the limit of sequential operations
In addition, overhead of memory accesses and core to core communication must be minimized to achieve high performance on many core processors
In early 2007, igolgi began designing parallel algorithms for video processing that will scale from
Today’s multicore chips to tomorrow’s many core chips. This work has resulted in highly efficient
Parallel execution of complex video functions necessary for media compression
igolgi Inc .
10
Parallel Video Compression
We have developed novel video compression algorithms in combination with an advanced software architecture built for dense multi-core platforms.
Motion estimation, mode decision, rate control, and other advances in algorithms
Software Architecture from ground up for parallel processors
Scalable Codec's supporting from HD to Mobile apps
Scales to 100s of cores
Intelligent Load Balancing to enable multi-format/multi-codec simultaneous processing
Completely flexible and dynamic
Novel pre-analysis technologies to optimize encoder performance
Video segmentation, feature recognition and algorithmic tuning
30%+ more efficient compared to traditional software architectures
Patent disclosures filed in the fields of parallel video compression algorithms and unique product functionalities
igolgi Inc .
11
igolgi codec algorithms and architecture allows for immediate scaling when more cores are available
igolgi Inc .
12
Technologies Features
igolgi Inc .
13igolgi Inc .
14igolgi Inc .
15
igolgi Video Compression Products are native software applications offering ultimate in flexibility and scalability
Native Parallel codecs that scale with the evolution of multi-core density
Real Time and Non-Real Time Applications
Dynamic codec control allows to change codec parameters instantly
Mix/match the channels and resolutions as needed
Integrate Customer specific interfaces or functions as desired to fit your system requirements
Over Time, upgrade to next gen multi-core and seamlessly increase your channel density
Leverage Multi-Core CPU Trends
igolgi Inc .
16
Blade Chassis
Ideal for High Channel Density installations Variety of Vendors and Densities Standard 2 Gig Ethernet per Blade Optional IB, 10Ge I/O Lowest TCO Redundant Power Supplies Integrated Switching Modules – reduced cabling N+1 Redundancy - big cost savings Upgradeable to next gene multi-core blades
1U , 2U, 4U Servers
Ideal for Small Footprint Installations Variety of Vendors Standard 2 Gig Ethernet PCI expandability Integrated HDD options Optional I/O HD-SDI , composite, ASI, others Portable Redundant Power
igolgi Inc .
17
IBM Blade Chassis
Ideal for High Channel Density Installations 14 Blades in a 9U Chassis Standard 2 Gig Ethernet per Blade Optional IB or 10 Ge Lowest TCO Redundant Power Supplies Integrated Switching Modules – reduced cabling N+1 Redundancy - big cost savings Upgradeable to next gen multi-core blades – re-use I/O, power, etc
Mercury 1U Server
Ideal for Low Channel Density Installations Standard 2 Gig Ethernet Optional IB or HD-SDI I/O Portable
igolgi Inc .
18igolgi Inc .
19igolgi Inc .
20igolgi Inc .
21igolgi Inc .
22igolgi Inc .
23igolgi Inc .
24igolgi Inc .
25igolgi Inc .
26igolgi Inc .
27
H.264 Encoder Features
Baseline, Main, High ProfileAny Resolution Frame Support
I,P,B , Hierarchical GOPS Max Multiple reference frames WP implicit mode Adaptive De blocking
Mode Decision/Motion Estimation All Block Sizes RDO based mode and motion estimation
algorithms All intra and inter modes supported down to
4x4 Search to 1,1/2,1/4 pel accuracy Motion Vectors over picture boundaries 8x8 transform
Pre-Analysis
2 Pass mode (programmable look-ahead)
Scene detection Image Analysis
– Complexity analysis– Skin Detection– Texture analysis – Edge detection
Rate Control HVS based quantization
control Load balanced rate control
with mode/me decision Programmable look ahead
delay
igolgi Inc .
28
Igolgi codec algorithms provide Very High Quality Performance with the ability to create Visually Lossless Quality demanded for some applications
1st Pass Look Ahead Period Programmable to allow small or large latency
igolgi Inc .
29
Igolgi two pass architecture provides smooth bit rate resulting in high video quality and accurate rate control
First pass also provides essential video characteristic data that improves second pass mode decisions
igolgi Inc .
30
Igolgi proprietary image segmentation and analysis identifies macro block level characteristics that enable intelligent mode, motion, and rate control decisions.
Result is higher video quality tuned to human perception
igolgi Inc .
31
H.264 High Profile provides tools above Main Profile that enables 15-30%
video quality improvement at the same bit rate
igolgi Inc .
32
Hierarchical GOPS provide typically 10% improvement in video compression performance by leveraging more B frames per GOP on average (less P frames)
Temporal scalability is also enabled
igolgi Inc .
33
Igolgi High Profile and Hierarchical GOP performance gains compared to Main Profile
igolgi Inc .
34
Igolgi weighted prediction can provide up to 60% improvement during a fade,
and 20% improvement during a dissolve video segment
igolgi Inc .