MetaRAM was founded in early 2006 to develop ICs that improve memory performance. The company received first round funding in January 2006, secured Series B funding in May 2007, demonstrated working samples in July 2007 and released its first chipset into production in November 2007. Investors include Kleiner Perkins Caufield and Byers, Khosla Ventures, Storm Ventures, and Intel Capital. The company has raised $20 million to date and is not currently looking for additional capital. MetaRAM employs 35 people.

Processing power doubles every 18 months, while memory capacity lags, doubling only every 36 months – a gap that is limiting processor performance. Meanwhile, virtualization, 64-bit computing, multi-core CPUs, and in-memory databases are driving demand for more and more memory. In typical virtualization scenarios, CPUs are under-utilized while memory is at capacity.

Today’s servers typically have 4 DIMMs per channel and 2 channels per CPU, placing an upper limit on the amount of memory that can be added to a server. And increasing memory speeds will lead to fewer DIMMs per channel, going down to three, teo and eventually one.

Until now, the industry has addressed this gap by adding higher capacity, but not readily available and exponentially more expensive DRAM to each DIMM on the motherboard. MetaSDRAM is focused on providing cost effective solutions to the persistent gap between memory capacity and processing improvements.

MetaRAM recently launch its first product, DDR2 MetaSDRAM, a new memory technology that can significantly increase server and workstation performance. MetaSDRAM enables two to four times the amount of standard memory to be placed into existing systems without any modifications. Using DDR2 MetaSDRAM, a quarter-terabyte, four-processor server with 16 cores can cost as little as $50,000, up to an eight-fold reduction in system cost – all without any system modifications.

The MetaSDRAM chipset, which sits between the memory controller and the DRAM, solves the memory capacity problem cost effectively by enabling up to four times more mainstream DRAMs to be integrated into existing DIMMs without the need for any hardware or software changes. The chipset makes multiple DRAMs look like a larger capacity DRAM to the memory controller. The result is “stealth” high-capacity memory that circumvents the normal limitations set by the memory controller.

MetaSDRAM, which is underpinned by more than 50 pending patents, solves the memory capacity problem affordably by enabling multiple mainstream DRAMs to look like a larger capacity DRAM to the CPU. For instance, the MetaSDRAM chipset, makes four separate 1Gb DDR2 SDRAMs look and act like a single 4Gb DDR2 SDRAM.

A dynamic command scheduler ensures that the MetaSDRAM is compatible with the JEDEC DDR2 protocol. WakeOnUse power management improves the power efficiency of DRAMs, thus enabling twice the memory to fit into a typical system’s cooling capacity. Low latency circuit design and an innovative clocking scheme allow MetaSDRAM-based DIMMs to fit into existing memory controller designs. Unique split-bus stack designs enable flexible access of the multiple DRAMs in a stack.

The MetaSDRAM MR08G chipset enables 2-rank 8GB DIMMs and is capable of functioning at speeds up to 667MT/s. It consists of an AM150 Access Manager and 5 FC540 Flow Controllers working as a group. The chipset is currently in full production.

The MetaSDRAM MR16G chipset enables 2-rank 16GB DIMMs and is capable of functioning at speeds up to 667MT/s. It consists of two AM160 Access Managers and 9 FC540 Flow Controllers. The chipset is qualified for production. All devices are fabricated by TSMC in 180nm CMOS.

MetaRAM products are designed for high performance rack-mount servers and workstations that run compute-intensive applications such as CAD/EDA simulations, database transaction processing (OLTP), business intelligence, digital content creation, and virtualization. With MetaSDRAM, two dual processor, quad core Barcelona servers at 50% CPU capacity can be consolidated into one server with double the memory capacity, resulting in 34% power savings and 50% space savings.

AMD is currently validating MetaSDRAM for use with the Dual-Core AMD Opteron processor and the Quad-Core AMD Opteron processor. MetaSDRAM is also compatible with the Dual-Core and the Quad-Core Intel Xeon processors with the 5100 MCH. Modules are currently available from Hynix and SMART Modular Technologies. Servers and workstations are expected in Q1 from Colfax International, Rackable Systems and Verari Systems. Future products will address higher memory speeds and capacities as the industry evolves.

Fred Weber, Founder, President and CEO (previously CTO at AMD where he spearheaded the development of the Opteron)

Suresh Rajan, Founder and VP, Marketing and Business Development (previously Director of Marketing for NVIDIA’s nForce platform processors and held engineering and management positions at Rambus, SGI, and Schlumberger)

Mike Smith, Ph.D., Chief Architect (previously a Chief Architect at NVIDIA and CTO of iReady)

Dhiraj Mallick, VP Silicon Engineering (most recently AMD’s director of Microprocessor CAD Engineering)

Barry Cornell, VP of Operations (previously VP of Operations for Cradle Technologies, and VP of Operations and Quality for Oak Technology)

Kashi Anoosheh, VP Worldwide Sales (most recently NVIDIA’s Senior Director of Sales)

Ron Nikel, VP Systems Engineering (most recently Director of R&D at Synopsys)