AMD News Archives

AMD has been teasing the Ryzen 9000X3D Zen 5 CPUs with 3D V-Cache and today they formally announced the specs of the Ryzen 7 9800X3D processor that will begin shipping 7 November.

Going back to earlier in the year AMD Linux engineers have been prepping the kernel for PCI Express TLP Processing Hints (TPH) support that allows for hints that can be injected to improve latency and lowering traffic congestion when there are several possible cache locations on the server with the TPH noting the optimal location of a Transaction Layer Packet (TLP). This PCIe TPH support is set to be merged upstream with the forthcoming Linux 6.13 cycle.

It turns out the latest AMD Ryzen desktop processors offer support for AMD Smart Trace Buffer (STB) that previously was only limited to mobile platforms.

Linus Torvalds took to some coding himself today to fix a user-address masking non-canonical speculation issue. The Linux kernel needed an adaptation for this "Meltdown Lite" issue due to different behavior with the latest AMD Zen 5 processors.

For the recently launched AMD EPYC 9005 series "Turin" processors there is good support out-of-the-box running on the likes of Linux 6.8 as found with Ubuntu 24.04 LTS. The one exception is if wanting to engage CPU power monitoring you need a RAPL/PowerCap patch that was just upstreamed in v6.12. But what about using a newer kernel for greater performance in light of all the upstream optimizations to the kernel in general? Here are some Linux 6.8 vs. 6.11 vs. 6.12 kernel benchmarks on a dual AMD EPYC 9755 server.

It was just earlier this week that AMD posted Linux patches to switch EPYC over to using the AMD P-State driver rather than the long-used generic ACPI CPUFreq driver. This should lead to better power efficiency out-of-the-box and is a change being made just for EPYC 9005 "Turin" CPUs and future server processors. Already it's looking like this change will be introduced for the upcoming Linux 6.13 merge window.

Making for an exciting Monday morning, AMD Linux engineers have kicked off the new week with a patch series introducing an exciting and long-awaited change: using the AMD P-State CPU frequency scaling driver by default for EPYC server platforms moving forward rather than the ACPI CPUFreq driver.

Merged today to Linux 6.12 Git were bug fixes to AMD's Indirect Branch Predictor Barrier (IBPB) handling that can be optionally used as part of the Retbleed and Speculative Return Stack Overflow (SRSO) mitigations on older AMD processors.

Following last week's release of the LLVM/Clang-downstream AOCC 5.0 for optimized compiler support extended to Zen 5 CPUs, the GPU side of the house at AMD this week released AOMP 20.0-0 as their LLVM/Clang downstream focused on GPU device offloading.

Back in January AMD published an open-source XDNA Linux kernel driver for supporting their Ryzen AI NPUs. But it wasn't until July that the formal review process for the AMD XDNA driver began as the necessary prerequisite for getting picked up into the mainline Linux kernel. On Friday the fourth iteration of those patches for review were published as it hopefully is closing in on landing within the mainline kernel.

A Friday evening job posting has confirmed and reinforced details around their future AI GPU compute stack, presumably what's been referred to as the Unified AI Software Stack.

As another interesting AMD announcement this week following their Advancing AI event yesterday where they launched the EPYC 9005 series and other new hardware, they've continued with a few more soft announcements in the lead-up to the OCP Global Summit happening next week. The latest interesting tid-bit is their plans to incorporate Project Caliptra into their products beginning in 2026.

With 5th Gen AMD EPYC "Turin" processors now launched, AMD provided a same-day release of their updated AMD Optimizing C/C++ Compiler "AOCC". This is AMD's downstream version of LLVM/Clang/Flang where they provide optimized AMD processor support with code that hasn't yet worked its way up into LLVM proper.

Back in August I wrote about AMD beginning work on a new Linux driver to help with heterogeneous core CPUs. On Thursday a second iteration of the AMD HFI Linux driver patches were posted with this driver continuing to work its way toward the mainline kernel.

After the AMD Advancing AI Event yesterday where they launched AMD 5th Gen EPYC processors, Instinct product updates, and new high-end networking gear, they also put out a blog post to affirm their "commitment to open security technologies in the data center."

In addition to announcing the EPYC 9005 "Turin" processors and the latest on the AMD Instinct front, Lisa Su at the AMD Advancing AI event in San Francisco also announced the AMD Pensando Salina 400 DPU and AMD Pensando Pollara 400 Ultra Ethernet AI NIC.

AMD today quietly posted a new open-source Linux kernel driver for review... the AMD 3D V-Cache Performance Optimizer Driver. This AMD 3D V-Cache Performance Optimizer Driver for Linux is intended to help optimize performance on systems sporting 3D V-Cache such as the AMD Ryzen "X3D" parts and the EPYC "X" processors.

A new set of patches from AMD Linux engineers today aim to boost the performance for heterogeneous CPU designs such as the recent Ryzen AI 300 "Strix Point" SoCs that have multiple core types.

Building off last year's release of the EPYC 8004 "Siena" processors featuring up to sixty-four Zen 4C cores, AMD today announced the EPYC Embedded 8004 series.

AMD today announced "AMD-135M" as their first small language model they are publicly releasing. AMD-135M is open-source with the training code, dataset, and weights all being open-source to help in the development of other SLMs and LLMs.

With the Linux 6.12 merge window wrapping up this weekend and the bulk of the new feature merges now in the tree, I've begun running some Linux 6.12 benchmarks. Here is an initial look at Linux 6.10 vs. 6.11 vs. 6.12 Git on an AMD Ryzen 9 9950X desktop.

Cloudflare's always-interesting technical blog laid out their details today concerning their next-gen "12th Generation" in-house servers that will be powering their vast web infrastructure. With these next-gen Cloudflare servers they are going with AMD EPYC 9684X Genoa-X processors.

Last week the initial AMD Zen 5 "znver5" enablement for LLVM/Clang was posted by an AMD compiler engineer. That code has since undergone review and merged for LLVM 20 Git and yesterday then back-ported for LLVM 19.

The Error Detection And Correction (EDAC) driver updates were among the early pull requests submitted for the Linux 6.12 kernel cycle in advance of this week's Linux Kernel Maintainer Summit in Austria. Among the EDAC work this cycle is preparing memory address translation support for future AMD platforms.

AMD engineers today posted the first "request for comments" patches in enabling support for Secure AVIC guest handling as a new hardware feature with upcoming processors.

AMD engineers continue work toward upstreaming their XDNA kernel driver for Linux in enabling the Ryzen AI NPU on open-source. The "v3" patches were posted on Wednesday but given the timing it looks like it will be missing out still on merging for the upcoming Linux 6.12 LTS cycle.

Early in the year we enjoyed seeing AMD Zen 5 "znver5" support upstreamed for the GCC 14 compiler in making it into that annual GNU Compiler Collection feature release. It was great seeing AMD Zen 5 support make it into this open-source compiler well ahead of any Zen 5 products being announced. Since then the GCC support for the new Znver5 target has continued to be improve upon meanwhile we've been waiting to see similar treatment for the LLVM/Clang compiler stack. Finally this week that AMD Zen 5 (znver5) support has been submitted for review in upstreaming it for LLVM.

Last year to much excitement in our community was the new AMD project announcement of openSIL as an open-source CPU silicon initialization project that is an advancement for open-source firmware and to eventually replace AMD's AGESA across both client and server processors. This week an exciting new update on AMD OpenSIL was shared and that they are still on-track for having it production-ready next year.

Following yesterday's initial tuning of the "znver5" target for the AMD Zen 5 CPUs with the GCC 15 compiler, several more rounds of compiler tuning/optimizations were merged for benefiting the Ryzen AI 300 series, Ryzen 9000 series desktops, and upcoming EPYC Turin processors.

Merged today for the GCC 15 compiler in development and potentially for back-porting to the next GCC 14 point release is a second round of AMD Zen 5 "znver5" tuning.

This week's batch of power management fixes for the Linux 6.11 kernel are just a set of three patches for AMD processors.

The AMD PMC driver used for the SoC power management controller already supports the initial AMD Zen 5 SoCs but new patches coming in as "fixes" for the Linux 6.11 kernel extend the coverage for some upcoming AMD platforms.

AMD for months has already been working on heterogeneous core topology optimizations for Linux within the AMD P-State CPUFreq driver and other heterogeneous CPU topology improvements for dealing with Ryzen systems sporting a mix of "classic" (full) cores with the denser "C" cores. Today though they've announced a brand new "Heterogeneous Core Driver" for further enhancing Linux support for AMD platforms sporting a combination of core types.

As part of catering to Google requirements around Google Chromebooks, AMD has been supporting Sound Open Firmware as the Intel-initiated open-source project across their APUs/SoCs. Sent out today was the patch for enabling ACP 7.0 SOF support for their newest SoCs like Strix Point to have Sound Open Firmware support.

If you are picking up one of the new AMD Ryzen 9000 series desktop processors soon for Linux use, you may be wondering whether it's worthwhile or even necessary moving to the latest Linux kernel code compared to the likes of Ubuntu 24.04 LTS that are shipping a Linux 6.8 derived kernel. Here are some quick benchmarks in looking at that question.

AMD just announced that they have entered into a definitive agreement to acquire ZT Systems, a hyperscale solutions provider focused on AI.

AMD Linux engineers are preparing the kernel for Smart Data Cache Injection (SDCI) as a feature for AMD EPYC server processors. Smart Data Cache Injection is a nifty new feature that allows for direct insertion of data from I/O devices into the CPU's L2/L3 cache.

Next month AMD will be providing an update on their long-term strategy for open-source firmware. Central to their open-source firmware plans is their OpenSIL effort that remains in development for eventually replacing AGESA on future generations of Ryzen and EPYC platforms.

Going back four years ago Intel engineers worked out bus lock detection for the Linux kernel to benefit their processors able to detect bus locks and then notify the kernel, given the negative performance implications associated with bus locks. That Intel support was merged in Linux 5.13 back in 2021 while now AMD has their equivalent ready for mainlining in the Linux kernel.

For those still running an AMD Ryzen 3000 series "Zen 2" desktop it really ought to be time to upgrade soon for better performance and power efficiency given the Zen 5 performance benchmarks thus far, but for those still planning to use the Ryzen 3000 series for some time, a quirk/workaround is on the way for enabling more of those older platforms to work with the AMD P-State Linux driver.

At the start of the year AMD posted an open-source XDNA Linux driver to GitHub for supporting the Ryzen API NPU IP found within their newest Ryzen mobile SoCs. It wasn't until last month though in mid-July that AMD began the process of submitting the driver for review so that it can work its way toward the mainline Linux kernel within the "accel" accelerator subsystem. Today brings a second revision to that driver.

Ahead of the Linux 6.11-rc2 kernel due for release later today there is the weekly "x86/urgent" material to merge.

For those that have experienced a buggy AMD HDMI audio experience when using recent versions of the Linux kernel, a fix has been submitted today for Linux 6.11 and in turn for back-porting to stable series in addressing "another long-time regression fix for AMD HDMI."

Now that I am through with my testing of the initial Ryzen AI 9 HX 370 and Ryzen AI 9 365 Linux performance benchmarking and support exploration, I've begun diving in to other areas of the Linux support/performance for these Zen 5 "Strix Point" SoCs. The area for a quick look today is with the yet-to-be-merged AMD Heterogeneous Core Topology patches.

Earlier this month AMD talked more about their Unified AI Software Stack plans for debuting in the coming months to provide a unified software view where AI work can be seamlessly offloaded to Ryzen processors, AMD graphics, or AMD Ryzen AI NPU hardware. Another possible and exciting prospect came to mind when going through the LLVM/Clang 19 changes this week.

While we have been super eager for the AMD Ryzen 9000 series "Zen 5" desktop processor launch that's been set for 31 July, AMD has issued a last minute delay. Instead the processors will launch in two stages in August.

One of the newest patch series out from AMD this week is on providing I3C HCI driver support for their MIPI I3C IP block found within their latest processors.

Back in January AMD quietly posted an XDNA Linux kernel driver for enabling the Ryzen AI NPUs. The driver has been maintained within that GitHub repository since but without any clear effort for getting this accelerator driver reviewed and merged into the upstream Linux kernel. Today that first step is finally being taken with the Ryzen AI XDNA Linux kernel driver patches posted to the Linux kernel mailing list and dri-devel to begin facilitating the upstream review process for getting this AI accelerator driver in the mainline kernel.

The AMD Secure Encrypted Virtualization (SEV) changes have been submitted for the recently opened Linux 6.11 merge window. Notable this cycle is getting support in the mainline kernel for SEV-SNP guest support over a Secure VM Service Module (SVSM).

Alongside all of the exciting Ryzen 9000 and Ryzen AI 300 series details shared last week at the AMD Tech Day in Los Angeles, what I also found to be very interesting was AMD sharing a bit more about a "Unified AI Software Stack" they are working to release in the coming quarters.