In today’s increasingly AI-assisted, software-defined IT world, it’s understandable how federal enterprise leaders might overlook some of the revolutionary developments reshaping the hardware landscape that powers all that software.
There’s little question: the land rush to leverage AI has cast a spotlight on the rising use of graphic processing chips in enterprise computing. What has gone less noticed, however, is the emergence of a new generation of scalable, high-performance microprocessors featuring built-in accelerators to improve memory, data streaming, balance database and security. These accelerators, developed by Intel, can match the computational performance of more expensive GPUs — and do so at significantly lower operating costs.
That development is important for many reasons, especially as enterprises and federal agencies, in particular, prepare to refresh their data center server fleets. Chief among those reasons: As enterprises enter this new era of hyper-intensive, AI-driven data processing, the marriage of software and hardware — and choosing the proper hardware for the job — will become more critical.
The bottom line is that as agencies map out their future IT investment strategies, choosing “fit-for-purpose” hardware will become essential to balancing performance demands with total operating costs.
Moving beyond cores
Until recently, the evolution of making CPU chips perform faster involved building more and more processing cores into each central processing unit. The advent of data streaming accelerator engines, like those built into the 4th Gen Intel Xeon Scalable processors, allows the CPU to offload the most common data movement tasks to a combination of general-purpose processors and specific-purpose processors.
The result is the ability to process multiple tasks within different processors simultaneously, translating into a massive increase in processing speed and a lower workload per processor. Compared to the prior generation of Intel Xeon processors, for instance, the 4th Gen Xeon processors deliver up to a 10x increase in AI workload performance, a 3x improvement in performance per watt, a 75% reduction in the total cost of operation and the ability to consolidate server footprints by a ratio of 5:1.
More fundamentally, the ability to deploy a variety of tailored CPU accelerators to meet today’s expanding data center and edge-computing workloads will give federal agencies far greater control of their overall IT operating costs while also delivering significantly more powerful computing performance.
For example, Intel’s series of built-in “fit-for-purpose” accelerator engines can handle:
- Advanced Matrix Extensions-– capable of natural language processing.
- Data Streaming – capable of speeding up data movement across memory caches, storage and network devices.
- In-Memory Analytics – that increases query throughput for big data analytics workloads.
- Advanced Vector Extensions – facilitating complex scientific simulations, financial analysis and 3D modeling.
- Crypto Acceleration – improving server utilization and performance of pervasive encryption-sensitive workloads.
Other Intel accelerator engines in the series are also available to help agencies manage fluctuating computing demands, protect sensitive or regulated data, and respond to system memory attacks that long evaded software-only solutions.
Planning for the future
Fit-for-purpose accelerator engines offer a new way of thinking about meeting current and future computing needs. Federal enterprises like the Defense Department, where computing at the edge is a way of life, stand to see significant new benefits by matching the correct hardware with computing needs at sea, in the air and the battlefield. So does the Intelligence Community, where surveillance data and image processing volume has reached unprecedented dimensions.
Virtually every federal agency, however, can expect to see escalating growth in data processing demands — and with it, a concurrent need to optimize IT hardware operating costs. The stakes are hard to understate. The federal government spends more than $100 billion on IT and cyber-related investments each year, with much of that going to maintain aging and increasingly inefficient hardware.
Making the most of this new generation of microprocessor power will also require working with experienced partners. Specialized IT solution providers like Sterling — which has more than two decades of experience building relationships, not just modern infrastructure systems, can provide federal agencies a head start in optimizing their data strategies by leveraging Intel’s latest microprocessor innovations.
Sterling’s highly skilled and certified engineering staff, its network of 1500 partners and broad federal contracts, can provide agencies with insightful consulting services, including the ability to provide advanced data engineering assessments and visionary design and architecture for state-of-the-art data exchange platforms. Sterling also has the expertise to fully execute data center refreshes, automation orchestration, and multi-cloud solutions to optimize performance. Sterling also provides expert managed services to accelerate deployment and fortify security.
While AI is likely to capture plenty of technology headlines in the months to come, don’t overlook the transformational changes taking shape right now that promise a world of new capabilities on the hardware side of IT.
Gretchen Stewart is the Chief Data Scientist at Intel Public Sector. Christopher Cyr is Chief Technology Officer at Sterling.