For many years we have developed and marketed high performance memory products,
including “Very Fast” static random access memory, or SRAM, and
low latency dynamic random access memory, or LLDRAM, that are incorporated primarily
in high-performance networking and telecommunications equipment, such as routers,
switches, wide area network infrastructure equipment, wireless base stations
and network access equipment. We sell these products to leading original equipment
manufacturer, or OEM, customers including Nokia and Cisco Systems. In addition,
we serve the ongoing needs of the military, industrial, test and measurement
equipment, automotive and medical markets for high-performance SRAMs. Based
on the performance characteristics of our products and the breadth of our product
portfolio, we consider ourselves to be a leading provider of Very Fast SRAMs.
We utilize a fabless business model, which allows us both to focus our resources
on research and development, product design and marketing, and to gain access
to advanced process technologies with only modest capital investment and fixed
costs.
Virtually all types of high-performance electronic systems incorporate some
form of volatile memory. An SRAM is a memory device that retains data as long
as power is supplied, without requiring any further user intervention. In contrast,
dynamic random access memory, or DRAM, is a memory device that requires user
intervention in the form of refresh operations to retain data while power is
supplied, due to the capacitive nature of its memory cell. However, a DRAM memory
cell is much smaller than an SRAM memory cell, so several times more DRAM bits
than SRAM bits can be implemented in any given unit area of silicon. The fundamentally
different characteristics of SRAM and DRAM memory cells have resulted in the
emergence of markedly different architectures for SRAM-based and DRAM-based
memory products, and the two types of memory serve different applications. Classically,
SRAM-based products have served high performance requirements while DRAM-based
products have been used in cost-optimized applications. Today, SRAM- and DRAM-based
products serve both performance and cost-based applications. As the volatile
memory market fragments into a variety of specialized products, more meaningful
distinctions between volatile memory products can be made.
There is an increasingly broad variety of volatile memory products on the market,
characterized by a number of attributes, such as speed, memory capacity, or
density, I/O interface and power consumption. There are several different industry
measures of speed:
latency, which is the delay between the request for data and the delivery
of such data for use and is measured in nanoseconds, or ns, or when used to
describe performance of synchronous memory products may be described in terms
of numbers of clock cycles required between the load of an address and the delivery
of valid data;
random access time, which is the minimum amount of time required between accesses
to random locations within the memory array, typically measured in nanoseconds,
or ns;
bandwidth, which is the rate at which data can be streamed to or from a device
and is often measured in megabits or gigabits per second (Mb/s or Gb/s);
clock frequency, which is the cycle rate of a clock within a synchronous device
and is often measured in megahertz or gigahertz (MHz or GHz); and
transaction rate, which is the rate at which new commands can be executed
by the memory device, and is often measured in millions or billions of transactions
per second (MT/s or BT/s).
Historically, SRAMs have been utilized wherever other lower price-per-bit memory
technologies have been inadequate. SRAMs demonstrate lower latency and faster
random access times relative to DRAMs and other types of memory technologies,
but at a higher price-per-bit. Historically, the volatile memory market has
had three price-performance points, DRAM at the low end, Fast SRAM at the high
end, and slow SRAM in the middle. Gartner Dataquest divides the SRAM market
into segments based on speed. The highest performance segment is comprised of
SRAMs that operate at speeds of less than 10 nanoseconds, which we refer to
as “Very Fast SRAMs.” Very Fast SRAMs are predominantly utilized
in high-performance networking and telecommunications equipment. Over the past
two decades, alternative memory technologies have been introduced to address
certain applications that formerly used slow SRAMs. For example, new types of
DRAM have displaced slow SRAM in applications such as cell phones. However,
in the networking memory market a technology vacuum formed between Fast SRAMs
on one end and commodity DRAMs at the other, with no high bandwidth, high transaction
rate, moderate capacity, moderate latency, and moderate cost volatile memory
product to fill the void. In the past decade, low latency DRAMs, or LLDRAMs,
have been developed to fill that void. Like the slow SRAMs that came before
them, LLDRAMs have a much higher price-per-bit than commodity DRAMs (in order
to deliver higher transaction rates) but demonstrate slower random access times
and longer latencies than Fast SRAMs.
The need for increasingly greater capacity, data bandwidth and transaction
rates from the various memory technologies continues unabated as the networking
market begins to make preparations for Terabit networking in the latter half
of the current decade. We believe that Fast SRAM and LLDRAM, optimized for networking
applications, will continue to play an essential role in enabling continued
improvements in network performance.
We believe the key success factors for a networking memory vendor are the ability
to offer a broad catalog of high-performance, high-quality and high-reliability
networking memory products, to maintain timely availability of prior generations
of products for several years after their introductions, and to provide effective
logistic and technical support throughout their OEM customers’ product
development and manufacturing life cycles.
Memory Requirements for “Big Data” Applications
With the vast amount of data currently being generated and the demand for faster
processing of that data, processor speeds are continuing to increase. However,
existing systems that move data back and forth between the processor and memory
are inadequate to address the fast response times required by “big data”
applications (including machine learning, CNNs and natural language processing).
Faster response times are also needed to meet the demands of developers in such
markets as cyber security and computer vision. For example, in the automotive
market, advanced driver assistance systems (“ADAS”) require a tremendous
amount of image processing to be accomplished in real time
The GSI Solution
Continue Leadership in the High Performance Memory Market
We endeavor to address the overall needs of our OEM customers, not only satisfying
their immediate requirements for our latest generation, highest performance
networking memory, but also providing them with the ongoing long-term support
necessary during the entire lives of the systems in which our products are utilized.
Accordingly, the key elements of our solution include:
Product Performance Leadership. Through the use of advanced architectures and
design methodologies, we have developed high-performance SRAM and LLDRAM products
offering superior high speed performance capabilities and low power consumption,
while our advanced silicon process technologies allow us to optimize yields,
lower manufacturing costs and improve quality.
Product Innovation. We believe that we have established a position as a technology
leader in the design and development of Very Fast SRAMs. We are believed to
have the industry’s highest density RadHard SRAM, the SigmaQuad-II+, which
is an example of our industry-leading product innovation.
Broad and Readily Available Product Portfolio. We have what we believe is the
broadest catalog of Very Fast SRAM products.
Master Die Methodology. Our master die methodology enables multiple product
families, and variations thereof, to be manufactured from a single mask set
so that we are able to maintain a common pool of wafers that incorporate all
available master die, allowing rapid fulfillment of customer orders and reducing
costs.
Customer Responsiveness. We work closely with leading networking and telecommunications
OEMs, as well as their chip-set suppliers, to anticipate their requirements
and to rapidly develop and implement solutions that allow them to meet their
specific product performance objectives.
0.42% 
8.8 %