Andy Patrizio is a freelance technology writer based in Orange County, California. He's written for a variety of publications, ranging from Tom's Guide to Wired to Dr. Dobbs Journal, and has been on staff at IT publications like InternetNews, PC Week and InformationWeek.
The company is offering a range of generative AI solutions across its hardware, software, and services offerings.
The two companies are building a new facility to speed the fabrication and deployment of modular data centers.
For data center operators, available power supply is a growing problem and it's driving development into smaller markets, according to research from CBRE.
The SuperNODE system combines 32 AMD Instinct MI210 accelerators in a single server using the GigaIO’s FabreX low-latency PCIe memory fabric.
New ThinkSystem enterprise storage arrays are designed for high-performance use.
Cloud sales are expected to grow over the next four years while on-prem sales will diminish.
Memory isn’t going to get more expensive overnight, but it’s not going to get much cheaper, either.
In a benchmark meant to measure the performance of training machine-learning models, Nvidia came out on top.
A pair of reports from Omdia paints a positive picture of the impact of AI on business as well as a decline in overall server shipments as buyers prioritize GPU-driven systems.
The demands of AI are too much for current IT infrastructure, according to 42% of respondents to an Equinix survey.
HPE's GreenLake for Large Language Models is supported by HPE supercomputers and AI software hosted in the cloud with LLM provided by a partner.
Almost half of data center operators surveyed have had an outage because they couldn’t get needed parts.
New Lenovo ThinkEdge servers will deliver more processing power where data is being generated and enable enterprises to run real-time inferencing at the edge.
A single Instinct MI300X can do the work of several GPUs, said AMD CEO Lisa Su.
The provider had been trying to find a buyer or reduce its debt load for months.
A minor bug can cause a system crash after 1,044 days of uninterrupted uptime. Be sure to reboot before then.
As popular locations such as Northern Virginia become saturated, data center developers are eyeing other US regions, including Atlanta, Las Vegas, Salt Lake City, and Denver.
Nvidia CEO gives the strongest indication yet it will have Intel make some of its chips.
Nvidia's Grace Hopper CPU/GPU combo underpins its supercomputer the company claims can crank out nearly an exaFLOP of AI performance.
Intel is switching up its data-center processor roadmap, and its proposed combination CPU and GPU, code-named Falcon Shores, will now be a GPU chip only.
The new processor is built around processing network workloads to take the load off the CPU.
Its North Carolina facility may be the most valuable asset that Bed, Bath & Beyond has left to liquidate.
AmpereOne chips are the first built around cores designed in-house and promise two to three times the power savings over x86.
The US Department of Energy is funding 15 projects aimed at developing energy-efficient cooling technologies for data centers.
NEO is developing a 3D DRAM architecture that uses memory stacking to solve the DRAM capacity bottleneck.
The A3 supercomputer's scale can provide up to 26 exaFlops of AI performance, Google says.
Demand for data center infrastructure persists amid power limitations, supply chain issues, and rising costs.
Joint offering combines 5G connectivity, edge computing infrastructure, and multi-cloud support in an on-demand service.
IBM LinuxONE Bare Metal Servers let enterprises get the hardware of a mainframe for their own Linux environments via an off-premises IaaS model.
The new Jericho chip can connect up to 32,000 GPUs concurrently and promises shorter job completion times for AI workloads.
Three announcements from Intel pertain to chip manufacturing and divesting its prebuilt server business.
A software update for AI benchmarking and a new networking chip are the latest developments in AI speeds and feeds.
Secure Data Recovery examined failure rates and lifespans of 2,007 hard drives from six manufacturers.
The Alveo-based card offloads video processing functions from the CPU to reduce latency and maximize channel density.
HPE adds hardware and software to support file and block storage as part of its Green Lake on-prem leasing.
Ahrefs estimates it would cost $400 million over three years if it were to switch from its colocation environment to AWS.
Sierra Forest will be the first to use efficient cores for lower processing power.
Supermicro's new server uses a self-contained liquid cooling system, and vendors outside of the IT infrastructure market are showing interest in the technology.
Scaleway explains how it plans to use servers for a decade and why it's getting rid of RAID controllers.
New Nvidia hardware and software are tailored to supporting AI.
In addition to sinking billions into OpenAI, Microsoft spent hundreds of millions on hardware.
Dell teams with colocation services provider Cyxtera to provide on-demand access to PowerEdge infrastructure, while a partnership with ZutaCore gives Dell access to liquid cooling technology.
Intel boosts bandwidth per channel in latest FPGA, Marvell unveils low-latency Teralynx 10 switch chip, and Ranovus previews co-packaged optics platform with AMD.
Storage vendor Vast Data adds built-in metadata index, called the Vast Catalog, which details every bit of data stored on its arrays.
Competitors AMD and Nvidia are already working on high-performance computing processors that combine CPU and GPU cores.
Nvidia says it will make it's brand of generative AI available through Oracle Cloud, Microsoft Azure, Google Cloud Platform, and other cloud providers.
AWS claims its new Amazon EC2 M7g and R7g instances provide 25% better performance vs. the past generation of instances.
After a dip due to surplus inventory and economic concerns, server CPU market expected to rebound.
Snapdragon X35 5G Modem-RF caters to edge devices in low-power environments.
Hardware durability metrics are drawn from the cloud storage provider’s portfolio of more than 230,000 hard drives under management.