Support memory semantic operations
Another pivotal underline is they support memory semantic operations, so they can be accessed like DRAM instead of being treated like storage, that reads and writes information indiscriminate in vast blocks.
Existing interfaces for hooking adult storage are not designed for this mode of operation, miss a compulsory bandwidth, and would deliver too most latency.
Gen-Z aims to residence this with a new interface that can scale to several hundred gigabytes per second, with latency subsequent 100ns. The standard, due for execution by a finish of this year, specifies sequence links that can be connected together for some-more bandwidth, in a identical conform to lanes in a PCI Express element bus.
Gen-Z is being designed to use Ethernet signalling and encoding, that doesn’t meant it is formed on Ethernet, only that it allows a use of existent Ethernet line motorist silicon for electrical or visual connections.
Because Gen-Z supports memory semantic operations, it could also be used to bond DRAM to a processor, though it is not dictated to reinstate a processor’s internal memory bus, during slightest not during first.
“There is still a need for top speed memory to be connected directly to compute, and that’s going to come with DDR5 in a subsequent era [of servers],” says Kurtis Bowman, Gen-Z Consortium boss and executive of Server Solutions for Dell EMC. “This [Gen-Z] is a element that allows for tiering to start to start with memory like we’ve seen with storage.”
Gen-Z permits memory to be connected directly to a horde processor as normal, or to be connected around a switched fabric as partial of a incomparable memory pool common between many processors. The consortium talks about “rack scale apparatus pools”, that means storage-class memory could be deployed in a apart enclosing in a datacentre rack, in a identical approach to storage arrays, permitting for a incomparable pool than could be propitious inside a singular server.
Article source: http://www.computerweekly.com/feature/Industry-giants-collaborate-on-open-standard-interconnects-to-meet-growing-data-processing-challenge