With MTIA v2 Chip, Meta Can Do AI Inference, But Not Training

If you control your code base and you have only a handful of applications that run at massive scale – what some have called hyperscale – then you, too, can win the Chip Jackpot like Meta Platforms and a few dozen companies and governments in the world have. …

With MTIA v2 Chip, Meta Can Do AI Inference, But Not Training was written by Timothy Prickett Morgan at The Next Platform.

Google Joins The Homegrown Arm Server CPU Club

If you are wondering why Intel chief executive officer Pat Gelsinger has been working so hard to get the company’s foundry business not only back on track but utterly transformed into a merchant foundry that, by 2030 or so can take away some business from archrival Taiwan Semiconductor Manufacturing Co, the reason is simple. …

Google Joins The Homegrown Arm Server CPU Club was written by Timothy Prickett Morgan at The Next Platform.

With Gaudi 3, Intel Can Sell AI Accelerators To The PyTorch Masses

We have said it before, and we will say it again right here: If you can make a matrix math engine that runs the PyTorch framework and the Llama large language model, both of which are open source and both of which come out of Meta Platforms and both of which will be widely adopted by enterprises, then you can sell that matrix math engine. …

With Gaudi 3, Intel Can Sell AI Accelerators To The PyTorch Masses was written by Timothy Prickett Morgan at The Next Platform.

Mixed Results For The Datacenter Thundering Thirteen In Q4

We have been tracking the financial results for the big players in the datacenter that are public companies for three and a half decades, but starting last year we started dicing and slicing the numbers for the largest IT suppliers for stuff that goes into datacenters so we can give you a better sense what is and what is not happening out there. …

Mixed Results For The Datacenter Thundering Thirteen In Q4 was written by Timothy Prickett Morgan at The Next Platform.

Intel’s Chips No Longer Pay More Than Their Fair Share Of Foundry Costs

The biggest benefit that is coming from the separation of the Intel chip design and marketing business from its foundry operations is that Intel’s chip product groups no longer have to shoulder the totality of the immense costs of its manufacturing operations. …

Intel’s Chips No Longer Pay More Than Their Fair Share Of Foundry Costs was written by Timothy Prickett Morgan at The Next Platform.

The Cloud Outgrows Linux, And Sparks A New Operating System

Ultimately, every problem in the constantly evolving IT software stack becomes a database problem, which is why there are 418 different databases and datastores in the DB Engines rankings and there are really only a handful of commercially viable operating systems. …

The Cloud Outgrows Linux, And Sparks A New Operating System was written by Timothy Prickett Morgan at The Next Platform.

HexaMesh: Chiplet Topologies Inspired By Nature

With the reticle limit for chip manufacturing pretty much set in stone (pun intended) at 26 millimeters by 33 millimeters down to 2 nanometer transistor sizes with extreme ultraviolet lithography techniques and being cut in half to 26 millimeters by 16.5 millimeters for the High-NA extreme ultraviolet lithography needed to push below 2 nanometer transistor sizes, chiplets are inevitable and monolithic dies are absolutely going to become a thing of the past. …

HexaMesh: Chiplet Topologies Inspired By Nature was written by Timothy Prickett Morgan at The Next Platform.

Pushing PCI-Express Switches And Retimers To Boost Server Bandwidth

Things would go a whole lot better for server designs if we had a two year or better still a four year moratorium on adding faster compute engines to machines. …

Pushing PCI-Express Switches And Retimers To Boost Server Bandwidth was written by Timothy Prickett Morgan at The Next Platform.

A Tale Of Two Nvidia Eos Supercomputers

Note: This story augments and corrects information that originally appeared in Half Eos’d: Even Nvidia Can’t Get Enough H100s For Is Supercomputers, which was published on February 15. …

A Tale Of Two Nvidia Eos Supercomputers was written by Timothy Prickett Morgan at The Next Platform.

There Is Still A Place For FPGAs In The Datacenter

By the time that the founders of Achronix, who were all techies from Cornell University, decided to found their own FPGA company twenty years ago, FPGAs had already been in the field for twenty years and the market was dominated by Xilinx (now part of AMD) and Altera (still part of Intel until it gets spun out sometime in the future). …

There Is Still A Place For FPGAs In The Datacenter was written by Timothy Prickett Morgan at The Next Platform.

Anthropic Fires Off Performance And Price Salvos In AI War

It is a strange time in the generative AI revolution, with things changing on so many vectors so quickly it is hard to figure out what all of this hardware and software and people-hours costs and what it might be worth when it comes to transforming, well, just about everything. …

Anthropic Fires Off Performance And Price Salvos In AI War was written by Timothy Prickett Morgan at The Next Platform.

The AI Wave Finally Starts Lifting Dell And HPE

It is beginning to look like the Dell Technologies and Hewlett Packard Enterprose, the world’s two biggest original equipment manufacturers, are finally going to start benefitting from the generative AI wave, mainly because they are finally getting enough allocations of GPUs from Nvidia and AMD that they can start addressing the needs of customers who don’t happen to be among the hyperscalers and largest cloud builders. …

The AI Wave Finally Starts Lifting Dell And HPE was written by Timothy Prickett Morgan at The Next Platform.

The Once And Future FPGA Maker Altera

Back in 2015, when we were launching The Next Platform, a lot of stuff was going on all at the same time, which is part of the zeitgeist that we were tapping into and that we wanted to chronical upon and participate within. …

The Once And Future FPGA Maker Altera was written by Timothy Prickett Morgan at The Next Platform.

SambaNova Pits LLM Collective Against Monolithic AI Models

There is more than one way to get to a large language model with over 1 trillion parameters that can do lots of different things and enterprises can use to create AI training and inference infrastructure to extend and enrich their thousands of applications. …

SambaNova Pits LLM Collective Against Monolithic AI Models was written by Timothy Prickett Morgan at The Next Platform.

He Who Can Pay Top Dollar For HBM Memory Controls AI Training

What is the most important factor that will drive the Nvidia datacenter GPU accelerator juggernaut in 2024? …

He Who Can Pay Top Dollar For HBM Memory Controls AI Training was written by Timothy Prickett Morgan at The Next Platform.

Intel: I Was Lostry, But Now I Am Foundry

Pat Gelsinger, current chief executive officer at Intel and formerly the head of its Data Center Group as well as its chief technology officer, famously invented the tick-tock method of chip launches to bring some order and reason to the way the world’s largest chip maker – as it was in the mid-2000s – mitigated risk and spurred innovation in its products. …

Intel: I Was Lostry, But Now I Am Foundry was written by Timothy Prickett Morgan at The Next Platform.

Nvidia Will Be The Next IT Giant To Break $100 Billion In Sales

Here is a history question for you: How many IT suppliers who do a reasonable portion of their business in the commercial IT sector – and a lot of that in the datacenter – have ever broken through the $100 billion barrier? …

Nvidia Will Be The Next IT Giant To Break $100 Billion In Sales was written by Timothy Prickett Morgan at The Next Platform.

Arm Neoverse Roadmap Brings CPU Designs, But No Big Fat GPU

Spoiler alert!

A lot of neat things have just been added to the Arm Neoverse datacenter compute roadmap, but one of them is not a datacenter-class, discrete GPU accelerator. …

Arm Neoverse Roadmap Brings CPU Designs, But No Big Fat GPU was written by Timothy Prickett Morgan at The Next Platform.

OSC Blends Intel HBM CPUs And Nvidia HBM GPUs For “Cardinal” Supercomputer

For a lot of state universities in the United States, and their equivalent political organizations of regions or provinces in other nations across the globe, it is a lot easier to find extremely interested undergraduate and graduate students who want to contribute to the font of knowledge in high performance computing than it is to find the budget to build a top-notch supercomputer of reasonable scale. …

OSC Blends Intel HBM CPUs And Nvidia HBM GPUs For “Cardinal” Supercomputer was written by Timothy Prickett Morgan at The Next Platform.

Half Eos’d: Even Nvidia Can’t Get Enough H100s For Its Supercomputer

UPDATED: Getting your hands on an Nvidia “Hopper” H100 GPU is probably the most difficult thing in the world right now. …

Half Eos’d: Even Nvidia Can’t Get Enough H100s For Its Supercomputer was written by Timothy Prickett Morgan at The Next Platform.