Phoronix reports:
AMD is building out what they are describing as an "elite" team of developers for driving Rust code "deep into the GPU stack" from firmware to drivers, shader compilers, and other GPU software in Rust. AMD Senior Fellow Harsh Meno posted this week to LinkedIn about this new effort...
And note this AMD job posting that's active for hiring a software development engineer to work on Rust code...
"We are building next-generation systems software for AMD GPUs with Rust as a core technical direction. The work spans compilers, runtimes, low-level GPU software, firmware, developer tooling, and methods for improving the safety and correctness of complex hardware-software systems.
"Rust is not incidental to this role. You will help establish how Rust is used in performance-sensitive and high-trust parts of the GPU stack, including the compiler and tooling support, system interfaces, engineering practices, and validation methods required for production deployment. You will have the opportunity to influence both near-term implementations and the longer-term architecture for using Rust across current and future AMD platforms."
"AMD also isn't alone," the article points out, "as NVIDIA has also been pursuing similar Rust-based initiatives for their GPU software too, including the development of the Nova Linux kernel driver written in Rust."
SteamOS is continuing its slow march to fuller support on a wider variety of non-Valve hardware. This weekend's release of the SteamOS 3.8.25 Beta includes "initial gamepad support" for a few recent gaming handhelds and improved support for a few others.
Those newly supported handhelds include MSI's Claw 8 EX AI+, the extremely expensive, extremely powerful handheld built around Intel's Arc G3 processor that launched this summer. While SteamOS has long included beta support for "other AMD powered handhelds," Valve has only recently been working on offering compatibility with Intel-based handhelds like the MSI Claw line. This weekend's update suggests the company hasn't abandoned those efforts and that Intel chips aren't being left out of the SteamOS conversation.
It feels like every week the component crisis worsens, driving the prices of hardware for even mid-level PC gaming to insane prices that only a few could afford without draining their savings.
Fortunately, price discounts are still a thing, giving us PC players brief respites from this RAMpocalypse. One such deal is this 26% discount for AMD Ryzen 7 7800XD, reducing the price of one of our favorite gaming CPUs from $449.99 to $333.99 on Amazon.
"Dominate any game at any level, the Ryzen 7 7800X3D offers ultimate performance with 8 cores, 16 threads, and up to 5.0GHz boost clock backed up with 96MB L3 cache." — Ben Wilson & Harish Jonnalagadda, Senior Editors
The AMD Ryzen 7 7800X3D in hand. (Image credit: Windows Central (Ben Wilson) | AMD)
The AMD Ryzen 7800X3D may be three years old, but it's still packing plenty of processing power to help your PC run games with extremely high performance that will make them feel so much smoother and more responsive to play with.
How high, you may ask? Well, my colleague, Ben Wilson, confirmed in his 4.5-star review of the AMD Ryzen 7 7800X3D that its 3D V-Cache makes an obvious difference to games, even if it doesn't thrill in productivity.
Shadow of the Tomb Raider benchmark.Ben Wilson | Windows CentralCyberpunk 2077 in-game at 200 FPS+.Ben Wilson | Windows CentralCyberpunk 2077 benchmark.Ben Wilson | Windows Central
You can boost the AMD Ryzen 7 7800X3D's performance further by overclocking it using the Precision Boost feature with the Ryzen Master app. This is all done while the AMD Ryzen 7 7800X3D's power management system keeps it stable and cool.
However, don't expect to use this CPU for grinding through productivity tasks in a speedy manner, as its performance rates in this regard are somewhat lacking compared to Intel's (often more expensive) offerings.
Additionally, the AMD Ryzen 7 7800X3D isn't compatible with DDR4 RAM, so folks running PC rigs with older RAM may need to upgrade (which is easier said than done thanks to this AI-induced RAM crisis sucking all our RAM dry).
AMD executives took to the stage at its Advancing AI 2026 event in San Francisco today to detail the company’s next generation of AI infrastructure solutions, from Instinct MI455X AI accelerator GPUs and 6th Gen EPYC “Venice” CPUs, to Pensando networking, ROCm.AI software and its Helios rack-scale platform that ties it all together.
AMD has been working towards rack-scale AI system solutions for years. Its ZT Systems acquisition last year added valuable engineering talent and intellectual property that is now finally bearing the real fruits. Its Helios AI platform is a major platform evolution for AMD, with shipments scheduled to begin in the second half of this year (which is here and now).
The announcements at Advancing AI show how the company has engineered its AI platform solutions for large reasoning models, sustained inference and agentic workflows. These workloads pressure memory capacity, data movement, networking and CPU orchestration. AMD’s approach is to keep as much data close to the compute engines as possible and move it more efficiently throughout the system, but there’s deeper nuance here that’s obvious versus AMD’s chief rival, NVIDIA.
AMD’s MI455X targets the AI memory wall
The Instinct MI455X GPU is the compute engine that fuels the Helios rack, and the first GPU based on AMD’s new CDNA 5 architecture. Built with a modular mix of 2nm and 3nm chiplets, it carries 432GB of HBM4 and 23.3TB/s of peak memory bandwidth.
Compared to AMD’s current MI355X, the MI455X offers 1.5 times the memory capacity, up to 2.9 times the peak memory bandwidth and up to four times the peak matrix performance with MXFP4 and MXFP8 data types, which are lower-precision numerical formats designed to accelerate AI processing while reducing memory demands. With MXFP6 (6-bit floating point), performance is rated at up to twice that of MI355X.
AMD also shared some actual, measured internal results using production silicon. The company claims MI455X delivers 3.8 times higher FP8 decode performance, 3.5 times more measured FP4 compute performance and between 2.5 and 3.5 times more networking bandwidth than MI355X, depending on the transfer path tested. Those figures provide more context than just numerical specifications, though they remain AMD-provided comparisons that will need independent validation.
AMD
The architectural choices behind the numbers are important. Reasoning models and long context windows require sizeable KV caches for maintaining AI attention states, while mixture-of-experts models frequently move large amounts of data across accelerators. MI455X should let more model data, activation states and cache remain local. New dedicated IP in hardware can transfer data while the GPU continues processing, and expanded cache and multicast capabilities are designed to reduce redundant data movement to further improve efficiency.
The aforementioned lower-precision formats can also raise throughput and reduce memory use, but model developers still have to determine where they can be applied without unacceptable accuracy loss.
AMD’s Helios rack takes aim at Vera Rubin
Dave Altavilla
Helios is AMD’s primary rack-scale competitor to NVIDIA’s Vera Rubin platform. Each liquid-cooled rack combines 72 MI455X GPUs, 18 single-socket Venice host CPUs and Pensando networking technologies.
In its most complete, premium configuration, AMD rates Helios for 2.9 exaflops of low-precision AI compute, with 31TB of aggregate HBM4 capacity, 1.7PB/s of memory bandwidth, 260TB/s of bidirectional scale-up bandwidth and 43TB/s of scale-out bandwidth.
These are formidable figures, but they are technical specifications rather than actual application benchmarks. The more consequential development is AMD’s move from collections of eight-GPU servers to a 72-GPU shared-memory domain. Models too large for one node can operate across the rack without treating every exchange as a scale-out networking transaction, which benefits large-model inference as well as training.
AMD uses UALink over Ethernet, or UALoE, for an open standard scale-up fabric. Each MI455X provides 3.6TB/s of bidirectional scale-up bandwidth, while the complete rack delivers all-to-all connectivity through a single switch layer. AMD also claims six times more scale-out bandwidth per GPU than MI355X when MI455X is configured with three Pensando Vulcano 800 AI NICs.
While open standards give cloud providers more control over suppliers and system design, AMD and its partners now have to prove those components can deliver the predictable performance, reliability and deployment experience customers expect from a tightly controlled, more vertically integrated platform.
Finally, AMD designed Helios with automatic rerouting around failed links, virtual rack partitions, tray-level serviceability and rack-wide power, cooling and health monitoring. Major hyperscalers and potentially large-scale enterprise customers will likely key in on these capabilities, which can affect the availability, total cost and consistency of the AI services they consume.
Kind of like cowbell, AMD Venice gives agentic AI more CPU
AMD
AMD’s agentic CPU messaging regarding its upcoming Venice-based EPYC processors is mostly marketing speak, but the underlying requirement is very real. An AI agent can invoke retrieval, databases, security checks, code execution and other tools before a GPU generates a response. Running many agents concurrently increases the amount of conventional compute requirements surrounding the accelerators.
Venice scales to 256 Zen 6 cores with support for 512 threads, 16 memory channels, up to 1GB of L3 cache per socket, along with PCIe 6.0 and CXL 3.1 connectivity. AMD is also offering several Venice configurations for other applications, including general-purpose servers, high-frequency workloads, GPU hosts and high-density CPU sandbox systems used to execute agent tools.
Treating the CPU solely as a GPU host understates its role. Gateways, tokenization, vector search, databases and short-lived code execution stress different mixes of per-core performance, thread count, memory bandwidth and I/O. Specifically, AMD’s internal testing shows Venice significantly outperforming its current EPYC 9965 Turin CPU across five parts of the agentic AI pipeline, including gateway processing, context assembly, vector search, enterprise applications and short-lived tool execution. Individual gains vary by workload, but AMD details the overall generational improvement at up to a 1.7 times lift. As with the MI455X figures though, these comparisons come from AMD and will require independent validation.
Pensando networking and ROCm software advance
Keeping GPUs fed with data and coordinating traffic across racks directly affects utilization and operating costs. In fact, GPU utilization is a pretty sad state of affairs currently for some of the major frontier model providers.
As such, Pensando networking has become central to AMD’s roadmap. Helios can connect each MI455X to as many as three 800Gbps Vulcano AI NICs, while Salina DPUs handle front-end networking and infrastructure services.
On the software side, which is an equally critical component, AMD also introduced ROCm.AI, an AI-assisted development layer due to arrive in August. It includes reusable skills for coding agents, simplified management and Hyperloom, which can profile workloads, tune serving configurations, modify kernels and validate results.
These tools address two persistent AMD challenges: developer efficiency and ease of use, and software tuning. Automated optimization still has to produce repeatable gains without creating hard-to-maintain code, however. And while ROCm has progressed significantly over the last few years, NVIDIA’s CUDA retains an advantage in maturity, tooling and developer familiarity.
AMD now has commitments that give its MI450 generation and Helios considerably more weight. Meta and OpenAI have announced multi-generation agreements composed of up to 6GW of AMD compute capacity, with initial 1GW deployments planned for the second half of 2026.
Oracle plans a 50,000-GPU public cloud cluster beginning in the third quarter, while Microsoft will deploy Helios for Azure AI inference. Finally, just before the AMD event, Anthropic announced a strategic partnership for up to 2 Gigawatts of AMD-fueled AI compute, with its first gigawatt expected online in the first half of 2027.
Commitments of this scale reflect confidence in more than just MI455X performance. These customers are evaluating the complete architecture, including Venice CPUs, Pensando networking, ROCm software, rack integration, serviceability and AMD’s ability to deliver and execute across multiple product generations.
There is some financial alignment behind the agreements as well. AMD issued OpenAI performance-based warrants and committed to investing up to $5 billion in Anthropic. That context matters when evaluating these deals as market validation, but these planned deployments are substantial nonetheless and put Helios on a much stronger foundation as it begins shipping.
AMD expands its robotics and embedded foundation
AMD also expanded its physical AI portfolio, building on credible traction from its Xilinx-derived Kria adaptive system-on-modules and embedded technologies that are already powering robotics, machine vision and industrial automation applications.
The new Ryzen AI Embedded X100 combines up to 16 Zen 5 CPU cores, integrated Radeon graphics, a second-generation NPU and as much as 128GB of unified LPDDR5X memory shared across its compute engines. To me this looks a lot like a repackaging and optimization of the company’s Strix Halo platform, but with specific optimizations for the embedded space. Regardless, AMD is pairing X100 with the Kria AI Robotics Developer Platform, which includes a System Module or SOM, and a new Robotics Partner Network spanning hardware, software and platform providers.
Samples began shipping in June, with full production expected in the fourth quarter. This broader objective is to give developers a path across AMD x86 CPUs, GPUs, NPUs and FPGAs for real-time autonomous systems, rather than requiring them to assemble those hardware engines and software components independently.
Execution for AMD is now the test
AMD has assembled a credible platform for the burgeoning agentic AI market that’s blowing up currently with no signs of stopping. MI455X addresses memory and data movement, Venice handles dense agentic CPU workloads, Pensando networking connects global system resources, and ROCm.AI addresses software complexity. Finally, Helios assembles these components into a true competitive threat for NVIDIA’s latest Vera Rubin platform.
AMD’s open architecture may appeal to customers seeking supplier choice, but openness must also translate into reliable deployments, competitive total cost and software that does not require a significant rip-up. NVIDIA enters this cycle with a stronger ecosystem and far more rack-scale deployment experience. The true test will be how easily and reliably customers can integrate, operate and maintain these AMD solutions at scale.
As it stands, AMD now has major customers and a clearly defined architecture with systems engineering expertise behind it. Delivering Helios on schedule and showing that its performance claims translate into a real production workload throughput advantage and total cost of ownership gains will determine how much the competitive gap narrows. And of course, this is in a market that is clamoring for ever-more compute resources with a seemingly insatiable demand for AI services and capacity. That’s an environment for big iron success. Now AMD just has to deliver optimized, turnkey AI platforms. This is far easier said than done, but time will soon tell as deployments take shape this year.
This article is published as part of the Foundry Expert Contributor Network. Want to join?
Consumer AMD CPUs will once again offer encryption protections against physical attacks after facing user backlash for silently removing the feature.
As Ars reported last week, AMD stripped the protection, known as TSME, from consumer Ryzen processors. Short for Transparent Secure Memory Encryption, TSME encrypts the entire contents stored in memory, making the data useless to adversaries performing cold boot attacks and similar intrusions requiring physical access.
Now you see it, now you don't, soon you'll see it again
About a decade ago, AMD added TSME to its high-end CPUs. Over the next few years, AMD added the protection to lower-end processors, including the consumer version of its Ryzen chips, a CPU that costs less than the Pro version. Over the years, users of these lower-end chips have gotten used to the added security, although some security experts (and plenty of novices, too) note that consumer chips are far less likely to be targeted by physical attacks. Recently and without warning or notice, the lower-end line of AMD chips suddenly dropped the protection, and it did so in a way that was impossible to detect on Windows machines and required a fair amount of technical work when using Linux. AMD last week declined to explain or acknowledge the change.
Consumer AMD CPUs will once again offer encryption protections against physical attacks after facing user backlash for silently removing the feature.
As Ars reported last week, AMD stripped the protection, known as TSME, from consumer Ryzen processors. Short for Transparent Secure Memory Encryption, TSME encrypts the entire contents stored in memory, making the data useless to adversaries performing cold boot attacks and similar intrusions requiring physical access.
Now you see it, now you don't, soon you'll see it again
About a decade ago, AMD added TSME to its high-end CPUs. Over the next few years, AMD added the protection to lower-end processors, including the consumer version of its Ryzen chips, a CPU that costs less than the Pro version. Over the years, users of these lower-end chips have gotten used to the added security, although some security experts (and plenty of novices, too) note that consumer chips are far less likely to be targeted by physical attacks. Recently and without warning or notice, the lower-end line of AMD chips suddenly dropped the protection, and it did so in a way that was impossible to detect on Windows machines and required a fair amount of technical work when using Linux. AMD last week declined to explain or acknowledge the change.
An anonymous reader quotes a report from Ars Technica: A decade ago, AMD added a protection to its high-end CPUs to protect them against cold boot attacks and other types of physical exploits that siphon sensitive data out of the connected memory chips. Short for Transparent Secure Memory Encryption, TSME encrypts the entire contents stored in memory, making the data useless to physical attackers. Over time, AMD added TSME to lower-end processors, including the consumer version of its Ryzen chips, a CPU that costs less than the Pro version. Over the years, users of these lower-end chips have gotten used to the added security. Recently and without warning or notice, this lower-end line of AMD chips suddenly dropped the protection, and did so in a way that was impossible to detect on Windows machines and required a fair amount of technical work when using Linux.
AMD has yet to say why TSME worked on these CPUs, or even to confirm the change. AMD declined to answer questions sent by email other than to say TSME "is a security feature only applied to PRO CPUs as part of AMD PRO Technologies." The statement is the first known time the chipmaker has explicitly made this restriction public. [...] There's no indication that AMD ever advertised or marketed TSME as being available in consumer CPUs. AMD has long said that a related memory protection, Secure Memory Encryption (SME), is available only in the Pro and Epyc CPU tiers. SME is OS-managed. It uses a single key and allows the OS to selectively encrypt individual memory pages. TSME is firmware-managed. It encrypts all RAM with no OS involvement. When active, it provides protection against physical attacks, including cold boot exploits, DRAM interface snooping, and memory module removal. It activates silently when enabled in the BIOS, making it the more practically useful of the two protections. Ben Kilpatrick, a self-described "privacy-conscious Linux hobbyist," discovered that TSME had stopped working on his consumer Ryzen processor despite remaining enabled in the BIOS. He spent months investigating, persuaded MSI engineers to test multiple CPUs, motherboards, and firmware versions, and filed a public AMD bug report that traced the change to newer AGESA firmware apparently disabling TSME on consumer chips while retaining it on Pro and EPYC models.
"AMD engineers' comments, such as those mentioned above, and the years of TSME working just fine in the lower-cost tier processors, have understandably conditioned Kilpatrick and other users to reasonably regard it as an expected part of the chip package," reports Ars Technica. "AMD quietly removing it and providing no acknowledgment or explanation strikes these users as something of a betrayal."
Joe Fitzgerald, an expert in silicon-level security, said in an interview: "They could have not realized they did it leading to their cagey responses, or they could have done it intentionally and tried to get away with it, leading to the same cagey responses. But I really feel like an explanation should be in order, even if it was 'TSME was never supposed to be supported. We did ship some firmwares that erroneously enabled it, but you shouldn't use them since we can't guarantee it'll work properly.'"
A decade ago, AMD added a protection to its high-end CPUs to protect them against cold boot attacks and other types of physical exploits that siphon sensitive data out of the connected memory chips. Short for Transparent Secure Memory Encryption, TSME encrypts the entire contents stored in memory, making the data useless to physical attackers.
Over time, AMD added TSME to lower-end processors, including the consumer version of its Ryzen chips, a CPU that costs less than the Pro version. Over the years, users of these lower-end chips have gotten used to the added security. Recently and without warning or notice, this lower-end line of AMD chips suddenly dropped the protection, and did so in a way that was impossible to detect on Windows machines and required a fair amount of technical work when using Linux.
Now you see it, now you don't
AMD has yet to say why TSME worked on these CPUs, or even to confirm the change. AMD declined to answer questions sent by email other than to say TSME "is a security feature only applied to PRO CPUs as part of AMD PRO Technologies." The statement is the first known time the chipmaker has explicitly made this restriction public.
Phoronix reports:
The AMD R600 Gallium3D driver saw 59 commits [last] Sunday to Mesa 26.2. Making this code restructuring and code cleaning all the more notable is that the improvements to this old AMD Radeon graphics driver was done in part by GitHub Copilot.
Gert Wollny has been among the few open-source developers left working on the AMD R600g driver that covers from the Radeon HD 2000 series through Radeon HD 6000 series graphics cards... [T]he old open-source GPU driver support is being assisted by AI long after the upstream vendor has stopped working on this driver — the Radeon HD 2000 "R600" series launched in 2007.
At some point during the fog of 2021 or 2022, I noticed that my son's preferred brand of fruit snacks had switched from including 0.9 ounces per pouch to 0.8 ounces per pouch. Most shrinkflation is meant to fly under the radar, but in this case, I just happened to notice it. It felt bad! It's tangible evidence that your money is not going as far as it did in the very recent past.
A little over a year ago, AMD launched the Radeon RX 9070 for a suggested retail price of $549. This month, it's launching the similarly named Radeon RX 9070 GRE for a suggested retail price of $549. This new card (actually the US launch of a GPU that's been available in China for a year or so) has 85 percent as many GPU cores, 75 percent as much memory, and 66 percent as much memory bandwidth as the regular RX 9070.
We'll evaluate the RX 9070 GRE in the context of the current GPU market, where prices have been edging upward due to the same AI-driven RAM shortages and price hikes that have made PC building and buying such a miserable experience for the last few months. But it's hard not to be a little upset about such a clear example of GPU shrinkflation—the same money for a markedly inferior product.
One of the benefits of building an AMD PC is that the company has historically supported its processor sockets for longer than Intel does, allowing the same motherboard (and RAM kit, if you want) to power your PC through multiple CPU upgrades. Today at Computex, AMD announced chips for the current AM5 socket and the improbably-still-around AM4 socket that will help extend their lives a little further, a nod to just how expensive it has become to build a new PC or perform a major upgrade these days.
The first of these announcements is something we knew about already: the relaunch of 2022's Ryzen 7 5800X3D, the first of AMD's commercially available 3D V-Cache processors. Dubbed a "10th Anniversary Edition" in reference to how long Socket AM4 has been around, the re-released chip is slower than regular 8-core Ryzen 5000-series CPUs in general productivity tasks but comes with 64MB of extra L3 cache that disproportionately benefits games. If you're trying to use a high-end GPU with an AM4 motherboard, it could help keep your CPU from being a performance bottleneck. The 5800X3D (re-)releases on June 25 for a suggested retail price of $349, which is less than it currently costs to buy secondhand.
As for the current AM5 socket, AMD officially announced that it was extending its support to at least 2029—it was originally planned to last until 2025, then until "2027+," so that means between two and four years of additional support, depending on how you're counting.
It's not an ideal time to be buying a new PC or doing a major upgrade. Price crunches for RAM and storage chips are making all kinds of components more expensive, and the shift to DDR5 in modern Intel and AMD CPUs means that a lot of people would need to pay money to replace their current DDR4 kits if they wanted to step up to a significantly newer, faster CPU and motherboard.
AMD may have something on the horizon for people who are looking to stretch their current PC (and its DDR4 RAM kit) just a little further. Leaks spotted by Tom's Hardware point to the existence of an "AMD Ryzen 7 5800X3D 10th Anniversary Edition," a re-release of a 4-year-old out-of-circulation CPU that might nevertheless be an upgrade for people with older Ryzen CPUs in Socket AM4 motherboards.
The "X3D" in the chip's name signifies that it comes with 64MB of extra L3 cache stacked on top of the main CPU die, bringing the total amount of L3 cache to 96MB. Workloads that benefit from extra cache—including most games—will perform much better on the 5800X3D than they do on the vanilla Ryzen 7 5800X.
Login
