Jump to content

User:Isadamune/sandbox

From Wikipedia, the free encyclopedia

History (New)

NVIDIA is an American based technology company founded in 1993 that revolves around the design of graphic processing units (or GPU’s). GPU's became a coined phrase when NVIDIA released their GeForce 256 marketing it as the world's first GPU.[1] Their primary product line they are producing is labeled GeForce that makes use of their latest micro architecture technology. The previous GPU micro architecture in their line was Pascal-architecture, which saw extensive use in the Titan-X, a highly advanced gaming computer able to run the most demanding of games. However, the current successor to the GeForce line is the Volta, their newest GPU micro architecture. Announced in March 2013, the Volta is named after Alessandro Volta, the inventor of the electrical battery.  

Details (Existing)

The Volta includes numerous memory innovations in relation to its counterparts. It features non-reliance on off-chip memory, the conventional way to process information for GPU’s, and opts to share its silicon substrate with DRAM.[2] Silicon substrate (also known as a wafer) is an extremely thin slice of semi conductive material that is used as the foundation for integrated circuits to be deposited. It’s use with DRAM is important since DRAM uses a high bandwidth interface allowing for high performance calculations. In comparison to its counter parts the DDR4 or GDDR5, High Bandwidth Memory (HBM) achieves a higher bandwidth at less power consumption by a substantial margin.[3] In addition HBM up to 4 GB per package which allows for smoother frame rates when gaming. 

The size of the architecture is also significant since generally speaking the smaller a GPU is, the less power it consumes and the faster it can run. Several rumors have been going about on the exact size of the GPU, mostly correlating a smaller size to a later release date.[4] If the GPU was to be released with a 12 nm for example a release in later 2017 was to be expected since they were hinting towards it.[4] A smaller size like 10 nm would require more time since the technology had no public announcements as of yet. For comparison, the Pascal GPU's are built at around 16 nm, where every nano meter reduction in size significantly lessens power consumption. A large step up from its predecessor, the Volta will sport a 1TB/s bandwidth.[4] Hothardware.com stated that one of the other reasons to believe that Volta could stick its head out in 2017 is because IBM references the GPU architecture when outlining Summit, which is its next generation high performance computing (HPC) system.[5] 

Use (New)

The Volta is already being integrated in next generation super computer such as the Summit being developed at Oak Ridge National Laboratory that they say will release some time in 2017.[5] In addition, among the features include that each node will have over half a terabyte of coherent memory (high bandwidth memory + DDR4) which will be addressable by all CPUs and GPUs plus 800GB of non-volatile RAM that can be used as a burst buffer or as extended memory.[5] Also, the Summit computer will also make use of both IBM GPU's and NVIDIA GPU's connected together with NVIDIA's high speed NVlink. Their NVlink is a highly energy efficient, high-bandwidth path between GPU's and CPU's as well as between GPU's themselves at bandwidths of 80-200GB/s. The advantage of having the GPU and CPU linked is that the user has the ability to choose which processor is right for the job at high throughput, a certain way of processing data called a Heterogeneous computing model.[6] The output of the computer in FLOPS per second is 150-300 petaFLOPS, or 10^15 floating operation points per second (FLOPS). For comparison the newest Microsoft console, the Xbox One, runs at 1.24 terFLOPS per second which is 10^12. The Summit will replace the Titan at Oak Ridge National Laboratory as their current super computer, with varying uses in research and simulations. In short the new computer will make use of multiple technological innovations made in the past years made complete with the newest GPU technology by NVidia.

Short graph of objects that contribute to the Summit Volta incorporated super computer

Although most of NVIDIA's products are used for gaming computers bought by gamers, super computers using their technology such as the Summit allow highly demanding calculations and research to take place. For example a common usage of super computers is to predict the weather. Meteorologists can use them to work on their research as well as monitor signs of global warming. Scientists can make use of them through molecular modeling soft wares, allowing for construction and observance of complex biological compounds. Simulation wise, the military can re create and reflect on the effect of nuclear weapons they would have on an environment. Other militaristic uses include providing a simulator for high cost vehicle operators to experience what it would be like piloting their machine.[7] Additionally, The Volta GPUs will connect to the POWER9 CPUs via NVLink 2.0 which is expected to support cache coherency and therefore improve GPGPU performance.[8]

  1. ^ "NVIDIA Launches the World's First Graphics Processing Unit: GeForce 256". http://www.nvidia.com. August 31, 1999. {{cite web}}: |first= missing |last= (help); External link in |website= (help)
  2. ^ Gasior, Geoff (March 19, 2013). "Nvidia's Volta GPU to feature on-chip DRAM". https://techreport.com. {{cite web}}: External link in |website= (help)
  3. ^ Handy, Jim (April 18, 2014). "Where Are DRAM Interfaces Headed?". http://www.eetimes.com. {{cite web}}: External link in |website= (help)
  4. ^ a b c James, Dave (February 21, 2017). "Nvidia's next-gen Volta GPU appears in the latest driver update". https://www.pcgamesn.com. {{cite web}}: External link in |website= (help)
  5. ^ a b c Lilly, Paul (January 25, 2017). "NVIDIA 12nm FinFET Volta GPU Architecture Reportedly Replacing Pascal In 2017". http://hothardware.com. {{cite web}}: External link in |website= (help)
  6. ^ "Summit and Sierra Supercomputers: An Inside Look at the U.S. Department of Energy's New Pre-Exascale Systems" (PDF). November 1, 2014. {{cite web}}: |first= missing |last= (help)
  7. ^ Bell, Thomas (November 5, 2004). "What are Supercomputers used for?". https://mason.gmu.edu. {{cite web}}: External link in |website= (help)
  8. ^ Smith, Ryan (November 17, 2014). "NVIDIA Volta, IBM POWER9 Land Contracts For New US Government Supercomputers". http://www.anandtech.com. {{cite web}}: External link in |website= (help)