Thread: Anandtech News

For Intel’s 2021 Architecture Day presentations, the yin to the CPU company’s traditional yang is GPUs. Intel has spent the last few years preparing to enter the market for discrete GPUs, with aims of competing in everything from consumer video cards up to HPC-class accelerators. And now, as the company already begins preparing for 2022, those efforts are finally coming to a head. Intel’s big breakout effort in the discrete GPU space starts in earnest next year with Xe-HPG and Xe-HPC, so for their 2021 architecture day, Intel is opening up a bit more about the GPU architectures that will be driving their first wave of high-performance products next year.


More...

Alongside a sneak peek at their forthcoming Xe-HPG architecture, the other big reveal today from Intel’s consumer graphics group comes from the software side of the business. Along with preparing Intel’s software stack for the 2022 launch of the first Arc products, the group has also been hard at work at their own take on modern, neural net-driven image upscaling techniques. The product of that research is Xe Super Sampling, or XeSS, which Intel is pitching as the best solution yet for high image quality and low processing cost image upscaling.

More...

In our coverage of the semiconductor space, we typically think of two main vectors of hardware – the CPU and the GPU. Beyond that, we look at FPGAs, microcontrollers, and this decade is bringing the advent of the dedicated AI processor. What ties all of these products together is actually the FPGA - a field programmable gate array that allows a skilled technician to essentially build a custom circuit out of configurable gates. This means an FPGA can be used to design and simulate a full CPU or GPU, but also an FPGA offers a reconfigurable way to offer optimized compute power that adapts to the needs of its users without the cost of millions or tens of millions to design dedicated silicon. One of the first FPGA companies on the market was Lattice Semiconductor, which now focuses on small power efficient FPGA designs that end up in everything from consumer devices to servers.
Over the last three years at Lattice, Jim has initiated a cultural shift that is playing out in the company roadmaps – new products, a more agile approach, and a need to focus on enabling machine learning at every part of its product stack. The recent financial disclosures at Lattice show an increasing demand for its hardware, as well as the company making strides to double its addressable market over the next five years. I thought this would be a good time to reconnect with Jim to find out exactly what he’s doing at Lattice to earmark the next generation of growth at this foundational FPGA company.

More...

This week as part of the annual Hot Chips semiconductor conference, Samsung’s memory division has presented a poster/slides on a project it is currently working on with impressive end-point results. The company details a 512 GB module of DDR5 memory, running at DDR5-7200, designed for server and enterprise use. This is a step up from the 256 GB modules at the top end of the market today, but to get there, Samsung has been introducing some new features and functionality.

More...

Over the last couple of years, a great deal of concern has developed around the future of semiconductor manufacturing, both with respect to total capacity and where the next generation of fabs will be hosted. The current chip crunch has underscored that current fab capacity is too small for a world where there’s a silicon chip in practically everything, and meanwhile geopolitical matters have made nations increasingly worried about where today’s cutting-edge fabs are located – mostly in Taiwan and South Korea. Consequently, we’ve seen governments kick-start initiatives to woo fab companies or otherwise incentivize the domestic construction of next-generation fabs, including the United States Department of Defense, whom today is awarding Intel an agreement to provide commercial foundry services for the DoD.
As announced by Intel this morning, Intel’s Foundry Services group has secured an agreement with the United States Department of Defense to provide fab services under the Rapid Assured Microelectronics Prototypes - Commercial (RAMP-C) program. RAMP-C is one of several US government programs to encourage domestic chip production, with this program focused on chip production for defense needs. In short, the DoD wants to ensure it will be able to have its chips (and other necessary commercial hardware) fabbed within the United States on a leading-edge commercial manufacturing node, and it is tapping a consortium of companies lead by Intel to develop the necessary foundry ecosystem.
Along with Intel, the consortium also includes IBM, Cadence, Synopsys and other companies, all of whom will be providing their relevant expertise and technologies to the project. These companies will be working together on what’s a fairly forward-looking service agreement, as the DoD is looking at fab needs several years down the line. Ultimately, the group is being tasked with establishing a semiconductor IP ecosystem around Intel’s forthcoming 18A process – the most advanced process on their development roadmap – which isn’t due to start ramping until 2025.

At this point Intel and the DoD are not announcing the value of the services agreement. There is no doubt some hedging going on, and there are multiple milestones Intel & co will need to hit between now and 2025 as part of their participating in the RAMP-C program.
But in the meantime, even being able to claim the DoD as a major customer for Intel Foundry Services is a big win for the group, which is still in the early stages of lining up customers and proving that it has learned from past mistakes, both with regards to offering contract foundry services, and in operating a leading-edge fab ecosystem. As a reminder, Intel has previously announced that it will be spending around $20 billion to build a pair of new fabs in Arizona, so success for IFS hinges on finding big customers like the DoD to fill those fabs with orders.

More...

Welcome to Hot Chips! This is the annual conference all about the latest, greatest, and upcoming big silicon that gets us all excited. Stay tuned during Monday and Tuesday for our regular AnandTech Live Blogs. Today we start at 8:45am PT, so set your watches and notifications to return back here! The first set of talks is all about CPUs: Intel Alder Lake, AMD Zen 3, IBM Z, and Intel Sapphire Rapids.

More...

Welcome to Hot Chips! This is the annual conference all about the latest, greatest, and upcoming big silicon that gets us all excited. Stay tuned during Monday and Tuesday for our regular AnandTech Live Blogs.

More...

Welcome to Hot Chips! This is the annual conference all about the latest, greatest, and upcoming big silicon that gets us all excited. Stay tuned during Monday and Tuesday for our regular AnandTech Live Blogs.


More...

Welcome to Hot Chips! This is the annual conference all about the latest, greatest, and upcoming big silicon that gets us all excited. Stay tuned during Monday and Tuesday for our regular AnandTech Live Blogs.

More...

Welcome to Hot Chips! This is the annual conference all about the latest, greatest, and upcoming big silicon that gets us all excited. Stay tuned during Monday and Tuesday for our regular AnandTech Live Blogs.

More...