Today Samsung announced that its chips are used exclusively to make up the 324-terabytes of DRAM in Germany’s new Leibniz Supercomputing Centre SuperMUC supercomputer.
Samsung’s release tells us that the SuperMUC, the most powerful supercomputer system in Europe, is an IBM System x iDataPlex dx360 M4 server built using over 18,000 Intel Xeon CPUs and over 80,000 4GB DRAM modules from Samsung. (Simple math makes this out to be 82,944 modules.)
That looks like a lot of silicon! Let’s see how much that might be.
A 4GB parity DRAM module would use nine 4Gb DRAM chips, which Samsung appears to Continue reading “Samsung DRAMs in Massive Leibniz SuperMUC”
In a surprise announcement Toshiba has said that it will immediately cut NAND flash production by approximately 30%. The company explains that this is being done “to reduce inventory in the market and improve the overall balance between supply and demand.” Toshiba’s release implies that this move is expected to improve prices, which have dropped as low as $0.31/GB recently.
By common measures of market share, which typically leave out SanDisk (for reasons too complex to discuss here) Toshiba holds a share of roughly 30% of the NAND flash market. By cutting its output by 30% Toshiba would be reducing overall NAND supply by 10%. If we were to include SanDisk, then that percentage would decrease to about 7.5%. Either one of these is significantly more than Continue reading “Toshiba to Cut NAND Production by 30%”
After years of prototyping Micron Technology claims to be the first to introduce production volumes of Phase-Change Memory, or PCM. This memory, also known as PRAM, has long been positioned as a contender to replace flash once flash reaches its scaling limit. Rather than use electrons to store a bit, PCM uses a type of glass that is conductive when in a crystalline state and resistive when amorphous, two states that are relatively easy to control. The size of the bits can shrink to a very small dimensions, allowing PCM to scale into the single-digit number of nanometers, which most folks today believe to be beyond the realm of flash.
This product began its life at Intel, then followed the Numonyx spin-off, and was taken over by Micron when it acquired Numonyx. In fact, Intel got into PCM very early on – this post’s graphic is the cover of an Electronics Magazine from September 1970 with an Intel story, written by Gordon Moore, telling about a 128-bit PCM research chip.
So far only three companies have produced samples Continue reading “Micron PCM Enters Mass Production”
An acquaintance recently brought to my attention an article in R&D Magazine about some pioneering research on phase-change memories or PCM. The researchers’ findings hold a lot of promise. (R&D Magazine’s article is based upon an original paper in the journal Science.)
A team led by Ritesh Agarwal, associate professor at the University of Pennsylvania, was trying to develop a better understanding of the mechanism behind the phase changes in PCM. The team found that existing programming algorithms that involve melting the material could be replaced with pulses of electrical current that not only would program the cell without heat, but provided an “On” to “Off” resistance ratio of 2-3 orders of magnitude, which renders the cell significantly easier to read, especially in the presence of noise. This effectively makes memory chip design Continue reading “A New Way to Build Phase-Change Memory (PCM)”