This site may earn affiliate commissions from the links on this page. Terms of use.

Servers

The entire computing industry has a memory trouble, and a new consortium of industry partners, dubbed Gen-Z, hopes to solve it. For decades, DRAM has driven almost every segment of the computing market, from smartphones to supercomputers, but new classes of memory devices already threaten that dominance. What's needed is a new retentivity interface that can necktie these various components together, and that's where Gen-Z comes in.

Interface-1

Some of the problems with DRAM performance scaling are long-standing, well-known bug. By and large speaking, the amount of bandwidth available per cadre has continued to decrease, despite the advent of DDR4. Consider the deviation between Intel'south Core i7-6950X, with 10 CPU cores and a total bandwidth of 76.8GB/due south when using DDR4-2400 versus the Core i7-4960X, with half-dozen cores and 59.7GB/s of DDR3-1866. The total bandwidth available to the Core i7-6950X is college, by nearly 30% — simply the 6950X also has 10 cores and 20 threads, compared with the 4960X'south six cores and 12 threads. Total bandwidth per core has indeed gone down — from 9.95GB/south per core for the 4960X to 7.68GB/s per core for the 6950X.

This difference persists even if we presume the user steps outside Intel's official specs and uses the highest-cease RAM realistically available. A quad-channel Core i7-4960X with DDR3-3100 would offer 99.2GB/south of bandwidth (xvi.5GB/s of bandwidth per core) while a Core i7-6950X with DDR4-4266 offers 136.51GB/s of bandwidth, or 13.65GB/south per core. No thing which components you choose, the amount of bandwidth bachelor per core is going down.

And so instead of just beating our heads against that fundamental limit, Gen-Z wants to beef up the performance of side by side-generation interconnects that might exist used to tap these emerging types of memory — some of which need to exist connected in means not covered by current standards.

DRAM-SCM

Today, the majority of systems incorporate DRAM and some type of storage, be that HDD or SSD. That'south going to start changing in the not-too-distant future, as High Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), and Managed DRAM are all more widely adopted. Other technologies, like Resistive RAM (RRAM), 3D XPoint (aka Intel's Optane), magnetic RAM (MRAM) and low-latency NAND will all be deployed in various systems and components. The goal of Gen-Z as stated is to build a "retentivity semantic fabric" that handles communications as memory operations with sub-microsecond latencies, from the time the CPU issues a load command to the time data is actually stored in a annals.

Gen-Z-4

That Gen-Z is talking about sub-microsecond latencies leaves a lot of room for speculation equally far as final performance is concerned. DRAM is technically a sub-microsecond retention, but we typically measure DRAM latency in tens to hundreds of nanoseconds (the exact number depends on the type of operation being measured, the DRAM's timing, and the speed of the integrated memory controller on-board the CPU). Gen-Z could offer substantially faster functioning for sure kids of attached hardware scenarios than equivalent standards today — particularly when compared with existing interconnect standards, which are oftentimes much slower than DRAM.

Gen-Z-3

The long-term goal of Gen-Z is to necktie the entire fastened ecosystem of products together on a single open up standard that can support sub-100ns load-to-use memory latencies in at least some cases. This will be at to the lowest degree partly adamant past which type of retentiveness is being discussed, which is probably one reason why the Gen-Z presentation doesn't comprise a lot of hard figures. A number of significant companies are backing the initiative, including AMD, ARM, Broadcom, Cray, Dell, HP, Huawei, Micron, Samsung, SK Hynix, and Xylinx.

The one major visitor missing is one you might expect to atomic number 82 such an endeavor: Intel. With upwards of 98% of the server and enterprise markets, Intel hardware is what you would expect these new retentiveness standards to all be compatible with. Even if you lot think AMD and ARM are poised to seize pregnant chunks of the data center marketplace, such growth takes years to build. Enterprise giant Cisco is also nowhere to be plant. More details and specification information are expected before the end of the yr, implying this project has already been in the works for quite some time.