Intel's Xeon Phi 14nm 'Knights Landing' Co-Processors Detailed - OmniPath Architecture 100 Series and 16GB HMC on a 2.5D Interposer

Intel has already revealed most of the big points of the upcoming Knights Landing platform earlier this year. However, Intel will be volunteering more information near the OmniPath Architecture used in its Xeon Phi products at the International Supercomputing Conference 2022 (ISC 2022). The Xeon Phi family unit of products is Intel's (plainly successful) foray into the GPGPU sector or the globe of extensively parallel data processing. The Co-Processors will exist deploying commercially later this yr.

Intel talks about Knight'due south Landing Omni Path 100 Series Architecture

Knight's Landing was built on the 14nm Process and uses modified silvermont cores (x86 ofcourse). It is as well one of the starting time mass produced components developed for this marketplace segment that features stacked DRAM (curt of AMD's Fury lineup, which doesn't actually count here). Since the GPU grade factor is commonly express (to heck) by the PCI-E lane, an culling CPU class factor is also provided which gives super-low latency and almost no bottlenecks. The on-packed stacked DRAM will come up in a whooping sixteen GBwhile at that place are connections for boosted DDR4 2400Mhz memory. Proceed in mind that we are talking about a socket-able, packaged processor here, not only a GPU sized product.

The HMC volition non actually be placed or stacked upon the dice. Using the 2.v D stacking that should now be pretty familiar for our readers, Intel volition actually be surrounding the Xeon Phi die using a Micron-Intel custom made, super-high bandwidth, parallel path interface that will brand the HMC announced as if its on the die. Infact it will deed more than or less similar an L3 cache worth 16GB. The Hybrid Memory Cube used in the Knights Landing Xeon Phi bundle will characteristic upto 2000 TSVs and an ASIC at the base of operations of the HMC to manage the DRAM package. Information technology promises more than 5 times the bandwidth of DDR4 RAM and more than than 15 times the bandwidth of DDR3 Ram. Because Intel is using a customized Micron 16GB HMC solution (they already have 2GB and 4GB variants) and a customized interface the bandwidth volition exist 500GB/s.

IntelOmniPath100

The Knight'southward Landing 'Xeon Phi' Co Processor will take upto 72 cores, something Intel has over again confirmed. This is really pretty decent news, because for once Intel seems to be having skillful luck when information technology comes to Yield. Its predecessor were but able to chug out Knight's Corner with around 61 cores. While I am sure there will be variants with cut down cores, there will also be variants with the full 72 cores. A signal I might add together is the number of cores can be misleading. These are cores based on the Silvermont architecture and nowhere near as powerful equally the the cores deployed in the mainstream desktop market. Nonetheless, they volition be able to pack a punch of upto 3 TeraFlops which is efficient enough given the electric footprint.

And ofcourse, the biggest advantage is that this is a co-processor. Unlike a GPU running GPGPU applications, which requires a CPU to direct its workload, a Co-Processor requires no such direction and is completely independent. The downside is that it needs more than customized software to effectively utilise it. The Omni Path 100 Series as Intel at present calls it is an endeavour at vertical integration and expanding into the market occupied past Infiniband. As you can conspicuously tell from the PR cloth, Intel is showcasing its Omni-Path interconnect every bit a significantly superior alternative to InfiniBand. Each Knight's Landing product will accept ii ports for the Omni-Path connectors. The exact release date of this product is not known but information technology is thought to be sometime later in this twelvemonth.