RAMBUS Dram

[Jorge-Vieira]

Rambus developing new DRAM with low latencies, increased capacities

Rambus has announced that it is working on a new dynamic random access memory (DRAM) architecture that will shrink latencies and increase capacities of computer memory chips. The company did not reveal many details on the matter, but indicated that at least two major producers of DRAM are interested in the new technology.
“While we are still in stealth mode on the project, I can tell you that we are again trying to make our semiconductor IP as consumable as possible and that we are trying to help the industry solve the enormous problems in the data center around latency and capacity for memory,” said Ron Black, chief executive officer of Rambus, during a conference call with investors and financial analysts.
According to Rambus, it has already signed a memorandum of understanding with a major customer (who is allegedly a DRAM maker) regarding licensing of the new tech. Besides, another client of Rambus got interested in the technology.

Although Rambus’ XDR, XDR2 and mobile XDR types of memory have a lot of advantages compared to industry-standard solutions, they have never managed to become popular on the market due to various reasons. A key motive why makers of DRAM and developers of microprocessors and application processors decided not to support memory architectures from Rambus was their reluctance to support proprietary technologies.
Rambus implies that its new memory technology is supported by at least two DRAM makers, which means that the technology has wider support than the XDR-series of technologies.
“The project is not custom, but broadly applicable to the industry,” said Mr. Black.
Rambus plans to formally unveil its new memory technology later this year and expects the industry to adopt the new architecture in 2016.

Noticia:
http://www.kitguru.net/components/me...ed-capacities/

[Jorge-Vieira]

Rambus to unveil brand-new DRAM tech in the coming weeks

Rambus this week said that it will officially introduce its all-new datacenter-class dynamic random access memory (DRAM) architecture in the coming weeks. The new DRAM is expected to shrink latencies and increase capacities of computer memory chips, something that is tremendously important for servers working with big data and in-memory databases.
“We are gearing up for a new product announcement just ahead of the Intel developer forum next month,” said Ron Black, chief executive officer of Rambus, during a conference call with investors and financial analysts. “This product pertains to two of the strategic memory programs we have discussed previously.”

Earlier this year Rambus disclosed that it was working on a new DRAM architecture aimed specifically at datacenters. The new DRAM tech was supposed to solve challenges associated with capacities and latencies of DRAM for servers. Rambus also indicated that at least two major producers of memory were interested in its new DRAM type.
This time the company did not reveal many additional details about the new memory architecture, but it did reiterate that the new type of memory will not replace NAND flash, but is designed to be a new DRAM technology.
“We have been working for the past few years to make our IP consumable taking our intuitive technologies and expertise in the memory space and packaging it into a product that will help improve server-based memory performance,” said Mr. Black. “Our product while not related to storage class memory architectures, as some had speculated, will improve both bandwidth and capacity requirements to meet the growing needs of the datacenter.”

It is noteworthy that Rambus will discuss the new technology ahead of Intel developer forum, not at the IDF, which may indicate that its new type of DRAM is not exactly a part of Intel’s roadmap.
Although Rambus’ RDRAM, XDR, XDR2 and mobile XDR memory architectures sported a lot of advantages compared to industry-standard types of DRAM, they have never managed to become popular on the market due to multiple reasons. Therefore, if Rambus wants its datacenter memory type to become popular, it will need to make it a breakthrough. Moreover, somehow Rambus will have to convince Intel Corp. to support its new DRAM architecture. Intel’s upcoming Xeon “Skylake” processors will support over 6TB of memory per socket, hence, they will hardly need to support a proprietary type of memory.

Noticia:
http://www.kitguru.net/components/me...-coming-weeks/


Tal como aconteceu há uns anos atrás na geração do Pentium 4 em que apareceram as memórias Rambus (bem caras na altura), não estou a ver este novo formato vingar face ao DDR3/DDR4.

[Jorge-Vieira]

Rambus To Go Into Fabless Chip Production, Announces RB26 DDR4 DIMM Chipset

Since its inception in 1990, DRAM technology company Rambus’s business model has been an unusual one, focused on the creation and licensing of technology as opposed to selling finished chips. It’s this emphasis on technology design and licensing that has led to Rambus holding a number of major DRAM patents, and also their resulting infamous legal fights with DRAM manufacturers over DDR technology. That said, with their legal battles settled earlier this decade we haven’t heard much from Rambus in the DRAM space as of late; after a run in the PC space with RDRAM and a design win in the Playstation 3 for XDR DRAM, the company has seen little success licensing further DRAM designs.
Meanwhile with the DRAM market having unified behind JEDEC standards – DDR4, GDDR5, and HBM – Rambus’s influence has been limited to that of a smaller player, though one still holding a selection of major DRAM patents. As a result there has been some speculation over just what the future would hold for a company whose newer DRAM designs have been ignored. With the 2015 Intel Developer Forum kicking off this week we finally have the answer to that question: Rambus will be making the transition from an IP licensing business to a true fabless semiconductor firm, designing and selling their own products.

For their entry into the world of chip design and sales, at this week’s IDF Rambus is announcing that they will be designing and selling DDR4 DIMM chipsets. The chips, which will trade under the R+ chipset family, will be for Registered DIMMs (RDIMMs) and Load Reduced DIMMs (LRDIMMs) for server usage, with Rambus producing both the Register Clock Driver (RCD) chip for RDIMM/LRDIMM, and the data buffer chips for LRDIMMs.

As we have covered in the past, RDIMMs and LRDIMMs allow servers to expand the amount of memory they can address and to do so at higher speeds. By interfacing with an RDIMM’s RCD to better organize address and command signals, a CPU memory controller can handle more DIMMs and more memory per DIMM than standard unbuffered memory. LRDIMMs in turn take this one step further by attaching buffers to the data bus of each RAM chip on a DIMM – essentially changing DIMM bus operation to a type of serial mode – yet again increasing the amount of memory servers can address and the speeds they can work at. The tradeoff is of course DIMM cost due to extra chips, and depending on the configuration performance as well, but in return servers can accept far more memory than what a standard unbuffered bus would allow.

Rambus for their part sees a place for themselves in the current DIMM market for supplying the register and buffer chips used for DDR4 RDIMM/LRDIMMs, as DDR4 imposes further limitations in order to reach its greater speeds. This in turn is where Rambus is looking to capitalize on their experience with high speed memory, as they believe it gives them a leg up in producing DDR4 chipsets for both speed and reliability. The company’s first chipset, the RB26, will be compliant with DDR4-2666 specifications, and eventually Rambus wants to take it to DDR4-2933.
Meanwhile, although Rambus is producing their own DIMM chipsets, they will not be producing their own DIMMs or DRAM. Rather the company will be offering their chipsets for sale to the DIMM vendors – Hynix, Micron, Samsung, etc – for those companies to use in building their respective RDIMMs and LPDIMMs. Ultimately such DIMMs would end up in the hands of server manufacturers and operators for their respective systems, with Intel’s latest Xeon processors being the catalyst.

Going forward, Rambus is aiming to turn this into a regular business for the company. Besides the RB26 chipset, which is sampling low and will be in production in Q4, the company is also already in the process of developing future generations of chipsets to further boost performance and reduce power consumption.
Finally, as for Rambus’s foray into fabless manufacturing, given the difficulties the company has seen in getting their DRAM technology adopted, the move into making chipsets for DDR4 is an interesting and somewhat unexpected one for a company who has traditionally only licensed IP, but also a logical one in their current situation. Though the company officially hasn’t thrown in the towel on their own DRAM standards, embracing the JEDEC standards rather than competing with them is the safer move given the momentum behind DDR4 and other standards. Meanwhile transitioning from pure licensing to selling their own chips is going to be a big – and undeniably risky – step for Rambus, but again owing to their background and history, it’s the move that makes the most sense as it means they aren’t wholly reliant on licensing to other companies to bring their designs to life, for all of the benefits and challenges that entails.

Noticia:
http://www.anandtech.com/show/9534/r...4-dimm-chipset

[Jorge-Vieira]

Rambus Explores Future Memory Systems

Oh great, more endless lawsuits and patent infringements from Rambus ? Rambus announced it will collaborate with Microsoft researchers in the exploration of future memory requirements for quantum computing. The expertise of Rambus in high-bandwidth, power-efficient memory architectures, combined with Microsoft researchers' knowledge of advanced system and data center design will be applied to drive new technology platforms.
"Existing computer architectures are reaching limits due to the ever increasing demands of real-time data consumption, which is driving the need to explore new high-performance, energy-efficient computer systems," said Gary Bronner, vice president of Rambus Labs. "By working with Microsoft on this project, we can leverage our vast expertise in memory systems to identify new architectural models."
"We've been investing in projects that advance our understanding of quantum computing along with its applications and implementation," said Douglas Carmean, Architect, Quantum Architectures and Computation Microsoft Research. "We are focusing on exploring theoretical and experimental approaches to creating quantum computers, designing software, hardware and other elements that support our research and direction. In this process, we are engaging with partners that have critical expertise, such as Rambus, to build new types of computing platforms."

Rambus has a long history of researching and exploring emerging platforms, developing meaningful technologies with broad applicability. As part of this program, Rambus and Microsoft will pool resources to further examine potential architectures that can greatly enhance memory capabilities in various settings to improve overall system performance. Specific details of the project are confidential.

Noticia:
http://www.guru3d.com/news-story/ram...y-systems.html

[Jorge-Vieira]

Microsoft and Rambus pool resources on quantum computing

Quantum computing research and development continues to pick up pace. Now we hear that software (services and devices) giant Microsoft will be collaborating with memory specialist Rambus on developing memory systems for future architectures, specifically for quantum computing.
Rambus notes that "existing computer architectures are reaching their limits," due to the tasks they are faced with in our fast paced digital world. Increasing demand for real-time data consumption is cited as one of the driving factors pushing the limits.
Companies need to "explore new high-performance, energy-efficient computer systems," said Rambus Labs VP Gary Bronner in a blog post. "By working with Microsoft on this project, we can leverage our vast expertise in memory systems to identify new architectural models," he explained.
So what is Microsoft's side of the bargain? The Rambus blog reminds us that Microsoft has previously invested in understanding quantum computing. Drilling down on what that means, it is clarified that Microsoft is focusing on "exploring theoretical and experimental approaches to creating quantum computers, designing software, hardware and other elements that support the company's research and direction". Douglas Carmean, Architect, Quantum Architectures and Computation Microsoft Research said that Microsoft was engaging with partners, including Rambus, thanks to their critical expertise in certain fields. Going forward Microsoft and Rambus will be pooling resources, aiming to enhance memory performance for quantum computers.

Last year brought with it quite a few advancements in quantum computing research and development. We have seen new fabrication techniques, optical chips with quantum capabilities, and the likes of Intel, Google and NASA plough money into research. Most interestingly we saw some of the first tantalising proof of potential in December when Google boasted that a D-Wave 2X quantum computer boosted certain calculation speeds by 100-million times.

Noticia:
http://hexus.net/tech/news/ram/89585...tum-computing/

[Enzo]

O pouco ou nenhum sucesso tem andado atrás destes senhores que ainda assim, não desistiram de verem as suas rams vingarem num futuro próximo. A provar isso, está esta aquisição, que os ajudará a criar e tornar consistente o seu roadmap já para este ano.
Para mais info sobre este tipo de ram, carrega aqui