The Most Obvious Hyperscaler To Do Custom Chips Was Always Facebook

Et tu, Meta?

Of all of the world’s hyperscalers and large cloud builders, only Meta Platforms, dominated by its Facebook and related social network businesses like Instagram and WhatsApp, is a pure-play hyperscaler. …

The Most Obvious Hyperscaler To Do Custom Chips Was Always Facebook was written by Timothy Prickett Morgan at The Next Platform.

Google Needs Another Database To Attack Oracle, DB2, And SQL Server Directly

Why does Google need another database, and why in particular does it need to introduce a version of PostgreSQL highly tuned for Google’s datacenter-scale disaggregated compute and storage? …

Google Needs Another Database To Attack Oracle, DB2, And SQL Server Directly was written by Timothy Prickett Morgan at The Next Platform.

Google Stands Up Exascale TPUv4 Pods On The Cloud

It is Google I/O 2022 this week, among many other things, and we were hoping for an architectural deep dive on the TPUv4 matrix math engines that Google hinted about at last year’s I/O event. …

Google Stands Up Exascale TPUv4 Pods On The Cloud was written by Timothy Prickett Morgan at The Next Platform.

Intel Unrolls DPU Roadmap, With A Two Year Cadence

There is a fundamental disconnect between the cadence that chip makers want for their devices and what the hyperscalers and cloud builders would prefer. …

Intel Unrolls DPU Roadmap, With A Two Year Cadence was written by Timothy Prickett Morgan at The Next Platform.

Intel Pits New Gaudi2 AI Training Engine Against Nvidia GPUs

Nvidia is not the only company that has created specialized compute units that are good at the matrix math and tensor processing that underpins AI training and that can be repurposed to run AI inference. …

Intel Pits New Gaudi2 AI Training Engine Against Nvidia GPUs was written by Timothy Prickett Morgan at The Next Platform.

How Much Of A Premium Will Nvidia Charge For Hopper GPUs?

There is increasing competition coming at Nvidia in the AI training and inference market, and at the same time, researchers at Google, Cerebras, and SambaNova are showing off the benefits of porting sections of traditional HPC simulation and modeling code to their matrix math engines. …

How Much Of A Premium Will Nvidia Charge For Hopper GPUs? was written by Timothy Prickett Morgan at The Next Platform.

Data Analytics Can Be The Next HPC For IBM Power

In the next few months, Big Blue will launch its entry and midrange Power10 servers, and to be blunt, we are not sure what the HPC and AI angle is going to be for these systems. …

Data Analytics Can Be The Next HPC For IBM Power was written by Timothy Prickett Morgan at The Next Platform.

Lawrence Livermore Kicks In Funds Foster Omni-Path Networking

Decades before there were hyperscalers and cloud builders started creating their own variants of compute, storage, and networking for their massive distributed systems, the major HPC centers of the world fostered innovative technologies that may have otherwise died on the vine and never been propagated in the market at large. …

Lawrence Livermore Kicks In Funds Foster Omni-Path Networking was written by Timothy Prickett Morgan at The Next Platform.

Can AMD Keep Doubling Its Datacenter Business?

Imagine, if you will, that AMD could make as many Epyc CPUs and Instinct GPU accelerators as it wanted at a reasonable yield and cost. …

Can AMD Keep Doubling Its Datacenter Business? was written by Timothy Prickett Morgan at The Next Platform.

Playing The Long Game With The Hyperscalers And Clouds

To build a very large IT business often takes a channel approach, where manufacturers sell a certain amount of their own product – say to the key 10 percent or 20 percent of their accounts – and then offload the rest of the sales job to a third party. …

Playing The Long Game With The Hyperscalers And Clouds was written by Timothy Prickett Morgan at The Next Platform.

Testing Out HPC On Google’s TPU Matrix Engines

In an ideal platform cloud, you would not know or care what the underlying hardware was and how it was composed to run your HPC – and now AI – applications. …

Testing Out HPC On Google’s TPU Matrix Engines was written by Timothy Prickett Morgan at The Next Platform.

Intel Adjusts, However Slowly, To New Realities In The Datacenter

While chip designer and maker Intel has a new strategy and a new executive team to implement it, it is going to take a long time for changes made last year and this year to be felt and for product and process roadmap changes to put the company into a better competitive situation. …

Intel Adjusts, However Slowly, To New Realities In The Datacenter was written by Timothy Prickett Morgan at The Next Platform.

The Hyperscalers Point The Way To Integrated AI Stacks

Sponsored Feature. Enterprises know they want to do machine learning, but they also know they can’t afford to think too long or too hard about it. …

The Hyperscalers Point The Way To Integrated AI Stacks was written by Timothy Prickett Morgan at The Next Platform.

With $130 Million In Hand, Ayar Labs Looks To Commercialize Optical I/O

When it comes to co-packaged optical interconnects, there are two camps: Lasers in and lasers out. …

With $130 Million In Hand, Ayar Labs Looks To Commercialize Optical I/O was written by Timothy Prickett Morgan at The Next Platform.

The Financial Longevity That Red Hat Gives IBM

It is hard to imagine, but someday, IBM may not care much about its proprietary System z and Power Systems platforms. …

The Financial Longevity That Red Hat Gives IBM was written by Timothy Prickett Morgan at The Next Platform.

Wanted: A Complete – And Heavily Customizable – HPC Software Stack

Sponsored Feature. There are a lot of things that the HPC centers and hyperscalers of the world have in common, and one of them is their attitudes about software. …

Wanted: A Complete – And Heavily Customizable – HPC Software Stack was written by Timothy Prickett Morgan at The Next Platform.

PSC Upgrades Neocortex AI Supercomputer With New Cerebras Engines

If you were going to build an electronic brain in 2022, it might look something like the Neocortex supercomputer at the Pittsburgh Supercomputing Center at Carnegie Mellon University. …

PSC Upgrades Neocortex AI Supercomputer With New Cerebras Engines was written by Timothy Prickett Morgan at The Next Platform.

With Aquila, Google Abandons Ethernet To Outdo InfiniBand

Frustrated by the limitations of Ethernet, Google has taken the best ideas from InfiniBand and Cray’s “Aries” interconnect and created a new distributed switching architecture called Aquila and a new GNet protocol stack that delivers the kind of consistent and low latency that the search engine giant has been seeking for decades. …

With Aquila, Google Abandons Ethernet To Outdo InfiniBand was written by Timothy Prickett Morgan at The Next Platform.

Expanding DevOps With Infrastructure As Code

The hyperscalers have taught us many lessons in the past two decades, and one of them is that everything that can be defined in software should be so that it can be controlled automatically and programmatically – and that goes double for hardware, which has required so much human babysitting over the decades. …

Expanding DevOps With Infrastructure As Code was written by Timothy Prickett Morgan at The Next Platform.

Stacking Up L2 Cache, RIKEN Shows 10X Speedup For A64FX By 2028

Let the era of 3D V-Cache in HPC begin.

Inspired by the idea of AMD’s “Milan-X” Epyc 7003 processors with their 3D V-Cache stacked L3 cache memory and then propelled by actual benchmark tests pitting regular Milan CPUs against Milan-X processors using real-world and synthetic HPC applications, researchers at RIKEN Lab in Japan, where the “Fugaku” supercomputer based on Fujitsu’s impressive A64FX vectorized Arm server chip, have fired up a simulation of a hypothetical A64FX follow-on that could, in theory, be built in 2028 and provide nearly an order of magnitude more performance than the current A64FX. …

Stacking Up L2 Cache, RIKEN Shows 10X Speedup For A64FX By 2028 was written by Timothy Prickett Morgan at The Next Platform.