Whenever new hardware is released, they always come with cool names and Intel’s latest Xeon Phi chip’s don’t disappoint with the name Knights Landing (any Game of Thrones fan spot the possible reference?). While not designed for desktops the next step of Colfax’s Ninja desktops will make sure of these supercomputing chips.
Be warned the extra power will come at a cost, with costs from Colfax’s website starting at $4,983 (around £3,508) for the base configuration. Featuring a 240GB SSD, a 4TB hard drive and a staggering 96GB of DDR4 memory the computer could easily let you get on with your daily YouTube and emailing while loading up the computer with two 1.6TB SSDs and two 6TB hard drives will jump the price to $7,577. With everything liquid cooled and two-gigabit ethernet ports, you don’t need to worry about overheating or slow network traffic.
Workstations are typically used for graphically intense operations such as film editing, graphics manipulation or engineering applications but with process heavy software coming out with the likes of virtual and augmented reality, people are looking at getting greater computing power like those offered by workstations for everyday use.
At the Supercomputing Conference ’15, Intel revealed more information about their new 72-core chip, Knight’s Landing (codenamed KNL), as well as showing off the 14nm die. Rumors surrounding the specifications of the Knight’s Landing chip, but it was good to have official confirmation of the details, courtesy of Anandtech.
The most surprising detail of the KNL chip was visible from the wafer that Intel had to showcase at the event. Close examination showed that each individual Knight’s Landing die could be as large as ~683mm^2, while being on a 14nm node, which is huge for a chip. Knight’s Landing will also be available in 2 layouts, a PCIe card, like past Xeon Phi chips that required a separate internal OS, or as a socket form factor, which will act as a regular processor. Further, the socket form factor will pack 36 PCIe lanes, which allow it to handle usage of two Knight’s Corner PCIe co-processor units.
The main cores of Knight’s Landing will be based on modified x86 Silvermont cores, generally used in low-power processors such as Intel Atom and Celeron lines. Both of KNL’s form factors will carry stacked DRAM, a first for its type. The socket form factor, which is not bottlenecked by the PCIe lane will have super low latency connections, making it able to accept additional 2400Mhz DDR4 memory, like a typical processor. The Knight’s Landing chips will also carry an onboard hybrid memory cube (HMC). And while the HMC will not actually be part of or stacked upon the die, instead a fully custom HMC unit will surround the die, acting like an L3 cache and appearing like it is actually part of the die.
Intel also showcased a basic reference motherboard for the KNL socket factor. The board was longer than a typical ATX motherboard but also thinner. Interestingly the motherboard did not come with an Intel Omni-Path connector, which would limit its expandability. This could be rectified with either the addition of additional KNL units or PCIe Omni-Path cards.
If that isn’t all, Intel also published a compiled list of all the disclosures they have made regarding the Knight’s Landing architecture, which for those who want to know the hard figures behind the chip can find here.
Typically larger and more expensive than a typical desktop, workstations are typically used for processing-intensive tasks such as high-quality computer generated graphics, film editing and for computations and modelling in the science and engineering fields. Due to these professional applications, they also require more processing power, often using high-end desktop chips or even server chips like Intel’s Xeon. This new Knight’s Landing chip will be based on the Xeon Phi architecture, of which the current generation chips are used in systems such as Tianhe-2, the world’s most powerful supercomputer.
The aim of bringing these powerful chips to workstations as well as supercomputers is an experiment in making supercomputing available to researchers without access to a full-scale supercomputer to run computation on, or to allow writing and testing of code intended for Xeon Phi-based supercomputers before deployment to the supercomputer itself. And while current workstations make use of discrete coprocessors alongside production CPUs in order to supplement their power, Knight’s Landing will run both the main processor and coprocessing units on one chip, in concert, the system will be able to provide over 3 teraflops at peak.
While the idea of 72-cores on a processor may boggle the mind of most PC users used to between 2 and 8 cores, Knight’s Landing runs more like a modern graphics card, of which the top end chips have multiple thousands of single purpose cores. Further, Knight’s Landing possesses 16GB of MCDRAM, of which is claims has 5-times the bandwidth of consumer DDR4 RAM as well as lower power draw and higher density than GDDR5.
Intel will be handling initial distribution of these new workstations themselves, hoping to extend sales of the workstations and maybe even desktop variants through other partner companies. These machines will be far more limited than typical PCs, however due to the chip being highly integrated into the rest of the system and the OS and other tools being pre-loaded by Intel. And while this seems like it could bring a new face to desktop computing, Intel claims that currently the rollout is more of an experiment than an attempt to do so. The ambition is definitely there, though, after the dropping of its Larrabee chip back in 2010. This is just the start, Intel already has plans for the successor to Knight’s Landing, Knight’s Hill.
Could the future be a supercomputer in all of our homes be more real than we think? If Knight’s Landing succeeds in its experimental release, we could be seeing chips of this calibre on the consumer market all-too-soon; an exciting idea for sure!
Supercomputer manufacturer, Cray, is said to help the US guard its arsenal of nukes after winning a $174 million contract to provide a new supercomputer to the National Nuclear Security Administration (NNSA).
The current supercomputer, a Cray XE6 called “Cielo”, is said to have 107,152 cores and a theoretical peak performance of a little over 1028 TFlops. The new supercomputer, which is a Cray XC super model going by the name of “Trinity”, is said to be connected to the company’s Sonexion storage at Los Alamos and is expected to provide 8x the power of the current XE6. The new supercomputer is said to be a joint project between “the New Mexico Alliance for Computing at Extreme Scale (ACES) at the Los Alamos National Laboratory and Sandia National Laboratories as part of the NNSA Advanced Simulation and Computing Program (ASC)”.
Trinity is said to be based on Intel’s Xeon Haswell processors and the upcoming “Knights Landing” Xeon Phi processors, boasting a 82 PB capacity and a design throughput of 1.7 TB per second. Its main purpose is to test the nuke arsenal’s safety, security, reliability and performance, in addition to conducting simulations of the US nuke stockpile in order to understand the weapons’ integrity as they age, while avoiding the need for underground detonations of devices.
Thank you The Register for providing us with this information
Intel has announcement the ‘Knights Landing’ Xeon Phi Coprocessor late last year, having released very few details about the lineup back then. As time passes, details are bound to be revealed and Intel is said to start shipping the series next year. This is why Intel apparently has decided to reveal some more details regarding the ‘Knights Landing’ Xeon Phi Coprocessor.
The announcement from last year points to the Knights Landing taking the jump from Intel’s enhanced Premium 1 P54C x86 cores and moving on to the more modern Silvermont x86 cores, significantly increasing the single threaded performance. Furthermore, the cores are said to incorporate AVX units, allowing AVX-512F operations and provide bulk Knight Landing’s compute power.
Intel is said to offer 72 cores in Knight Landing CPUs, with double-precision FP63 performance expected to reach 3 TFLOPS, having the CPUs boasting the 14nm technology. While this is somewhat old news, Intel revealed some more insights at the ISC 2014.
During the conference, Intel stated that the company is required to change the 512-bits and GDDR5 memory present in the current Knights Corner series. This is why Intel and Micron have apparently struck a deal to work on a more advanced memory variant of Hybrid Memory Cube (HMC) with increased bandwidth.
Also, Intel and Micron are said to be working on a Multi-Channel DRAM (MCDRAM) specially designed for Intel’s processors, having a custom interface best suited for Knights Landing. This is said to help scale its memory support up to 16 GB if RAM while offering up to 500 GB/s memory bandwidth, a 50% increased compared to Knights Corner’s GDDR5.
The second change made to Knights Landing is said to include replacing the True Scale Fabric with Omni Scale Fabric in order to offer better performance compared to the current fabric solution. Though Intel is currently keeping this information on a down-low, traditional Xeon processors are said to benefit from this fabric change in the future as well.
Lastly, compared to Intel’s Knights Corner series, the Knights landing will be available both in PCIe and socketed form factor, mainly thanks to the MCDRAM technology. This is said to allow the CPU to be installed alongside Xeon processors on specific motherboards. The company has also emphasised that the Knights Landing version will be able to communicate directly with other CPUs with the help of Quick Patch Interconnect, compared to current PCIe interface.
In addition to the latter, having the Knights Landing socketed would also allow it to benefit from the Xeon’s NUMA capabilities, being able to share memory and memory spaces with the Xeon CPUs. Also, Knights Landing is said to be binary compatible with Haswell CPUs, having the company considering writing programs once and running them across both types of processors.
Intel is expected to start shipping the Knights Landing Xeon Psi Coprocessor somewhere around Q2 2015, having the company already lining up its first Knights Landing supercomputer deals with National Energy Research Scientific Computing Center with around 9300 Knights Landing nodes.
While the awesome Xeon Phi offers some impressive performance options, it is a co-processor, that means that it needs a normal CPU to function, which is why the Xeon Phi normally comes installed on an add-in card, much like a modern GPU.
Intel are looking to address this issue with the launch of the upcoming, next-generation Knights Landing model which Intel revealed this week at the Supercomputing Conference.
Designed using a 14 nanometer process, the new chip will be made available as a stand-alone model that can run software much like any traditional CPU. However, since the new independent co-processor will no longer need to rely on a main CPU to shuttle data to it, it will be faster, more efficient and cheaper to get up and running than ever before. Intel are pushing this integrated performance even further by adding integrated high-speed memory on the chip.
Due for release in late 2014 or early 2015, the new Knights Landing models could be quite the revolution in the world of massively parallel computing.
Thank you Engadget for providing us with this information.