NVIDIA is working with companies worldwide to build out AI factories—speeding the training and deployment of next-generation AI applications that use the latest advancements in training and inference. The NVIDIA Blackwell architecture is built to meet the heightened performance requirements of these new applications. In the latest round of MLPerf Training—the 12th since the benchmark's introduction in 2018—the NVIDIA AI platform delivered the highest performance at scale on every benchmark and powered every result submitted on the benchmark's toughest large language model (LLM)-focused test: Llama 3.1 405B pretraining.

The NVIDIA platform was the only one that submitted results on every MLPerf Training v5.0 benchmark—underscoring its exceptional performance and versatility across a wide array of AI workloads, spanning LLMs, recommendation systems, multimodal LLMs, object detection and graph neural networks. The at-scale submissions used two AI supercomputers powered by the NVIDIA Blackwell platform: Tyche, built using NVIDIA GB200 NVL72 rack-scale systems, and Nyx, based on NVIDIA DGX B200 systems. In addition, NVIDIA collaborated with CoreWeave and IBM to submit GB200 NVL72 results using a total of 2,496 Blackwell GPUs and 1,248 NVIDIA Grace CPUs.

This year marks the 40th anniversary of the first commercially available field-programmable gate array (FPGA), introducing the idea of reprogrammable hardware. By creating "hardware as flexible as software," FPGA reprogrammable logic changed the face of semiconductor design. For the first time, developers could design a chip, and if specs or requirements changed mid-stream, or even after manufacturing, they could redefine its functionality to perform a different task. This flexibility enabled more rapid development of new chip designs, accelerating time to market for new products and providing an alternative to ASICs.

The impact on the market has been phenomenal. FPGAs launched a $10+ billion industry and over the past four decades we have shipped more than 3 billion FPGAs and adaptive SoCs (devices combining FPGA fabric with a system-on-chip and other processing engines) to more than 7,000 customers across diverse market segments. In fact, we've been the programmable logic market share leader for the past 25 consecutive years, and we believe we are well positioned for continued market leadership based on the strength of our product portfolio and roadmap.

Logitech G, a brand of Logitech and a leading innovator of gaming and sim racing technologies and gear, today announced the launch of the Logitech G522 Gaming Headset, a new addition to its award-winning G5 series, designed to elevate both gameplay and personal expression. "G522 was made for gamers who bring passion and personality to everything they do," said Ujesh Desai, vice president and general manager of Logitech G. "Whether you're climbing the ranks, streaming for your audience, or pulling all-night gaming sessions, the G522 ups your game with top-tier performance, features, and style."

G522 builds on the strong foundation of the widely popular Logitech G733 headset, carrying its legacy further while introducing an exciting evolution. At first glance, the family resemblance is clear, but as you look closer, the G522 reveals a fresh, elevated design. It's all about progression - sleeker curves, a more sophisticated silhouette, and thoughtful design choices that reflect a modern take on performance and style. The sharper edges of the G733 have been transitioned into smooth, flowing lines showcasing a refined aesthetic.

Marvell Technology, Inc., a leader in data infrastructure semiconductor solutions, today announced it is teaming with NVIDIA to offer NVLink Fusion technology to customers employing Marvell custom cloud platform silicon. NVLink Fusion is an innovative new offering from NVIDIA for integrating custom XPU silicon with NVIDIA NVLink connectivity, rack-scale hardware architecture, software and other technology, providing customers with greater flexibility and choice in developing next-generation AI infrastructure.

The Marvell custom platform strategy seeks to deliver breakthrough results through unique semiconductor designs and innovative approaches. By combining expertise in system and semiconductor design, advanced process manufacturing, and a comprehensive portfolio of semiconductor platform solutions and IP—including electrical and optical serializer/deserializers (SerDes), die-to-die interconnects for 2D and 3D devices, advanced packaging, silicon photonics, co-packaged copper, custom high-bandwidth memory (HBM), system-on-chip (SoC) fabrics, optical IO, and compute fabric interfaces such as PCIe Gen 7—Marvell is able to create platforms in collaboration with customers that transform infrastructure performance, efficiency and value.

According to the Financial Times, gathering data from satellite images and industry intelligence, Huawei has endeavored to develop a domestic AI supply chain to bypass foreign tech restriction influence. In Guanlan, China, Huawei started developing a complete facility for manufacturing semiconductors on 7 nm technology for its custom processors. Out of frustration with SMIC's low output capacity, Huawei has secured the entire silicon production, from sourcing materials, chemicals, and wafer fab equipment to chip-making equipment to actual chip design. According to Dylan Patel of SemiAnalysis, "Huawei has embarked on an unprecedented effort to develop every part of the AI supply chain domestically from wafer fabrication equipment to model building," adding, "We have never seen one company attempt to do everything before."

It is also reported that Huawei's rivals in silicon manufacturing, SMIC and SMEE, have deployed engineers to help Huawei develop its own manufacturing flow. A few companies, all backed by Huawei with funding and research, are the backbone of this operation. SiCarrier, which we reported on back in March, supplies optical and X-ray inspection tools, atomic force microscopes, and alignment systems for metrology; gas-based and atomic layer deposition tools for film coating; plasma etchers for patterning; rapid thermal processors for material tuning; and electrical testing platforms for reliability screening. SwaySure and Fujian Jinhua supply memory chips, Si'En and Pehgjin supply power chips, and PWX and PST deal with logic.

In the last few weeks, Intel rolled out two new performance‑enhancing presets for its Core Ultra 200S "Arrow Lake" processors: Intel Performance Optimizations (IPO) and the factory‑approved 200S Boost overclocking profile. Independent benchmarks shared by a BiliBili user reveal that IPO delivers a slightly stronger uplift in gaming performance compared with 200S Boost. Testing centered on a Core Ultra 7 265K paired with DDR5‑8000 memory and a GeForce RTX 5090D. Seven modern titles were measured at 4K with DLSS enabled where supported. Three system presets were compared: a baseline with XMP‑enabled DDR5‑8000, Intel's 200S Boost, and Intel IPO. Under 200S Boost, Intel ramps the chip's Die‑to‑Die (D2D) fabric from 2.1 GHz to 3.2 GHz and the Next‑Generation Uncore (NGU) fabric from 2.6 GHz to 3.2 GHz, while maintaining factory CPU clock speeds. By contrast, IPO takes a more holistic approach: P‑core and E‑core boost clocks rise to 5.4 GHz and 4.9 GHz (from 5.2 GHz/4.9 GHz), the ring bus ticks up to 4.0 GHz (from 3.8 GHz), and memory is overdriven to DDR5‑8400 with tightened timings.

IPO's D2D and NGU fabrics run at 3.1 GHz due to power-and-thermal headroom trade-offs. Across the board, IPO edged out 200S Boost by roughly 2% in average frame rates and 1% in the 1% lows. In Forza Horizon 5, IPO delivered 274 FPS versus 269 FPS on 200S Boost, with 1% lows of 198 FPS (+1 FPS). Cyberpunk 2077 saw a 3% jump to 297 FPS average and a 6% gain in 1% lows. Total War: Warhammer III's averages climbed 6%, while 1% lows rose 9% under IPO. The most pronounced advantage appeared in Counter‑Strike 2 (tested at 1080p), where IPO boosted averages by 16% and 1% lows by 20 %. Watch Dogs: Legion also benefited, with an 8% average and 9% low‑end uplift. These results suggest that IPO's balanced tuning of CPU cores, ring bus, interconnects, and memory yields consistent, measurable gains over the simpler fabric‑focused 200S Boost profile. Intel has yet to greenlight the IPO's broader rollout beyond its initial OEM launch in China, but these early figures hint at meaningful real‑world improvements for Arrow Lake gamers.

OKI Circuit Technology, the OKI Group printed circuit board (PCB) company, has successfully developed 124-layer PCB technology for wafer inspection equipment designed for next-generation high bandwidth memory, such as HBM mounted on AI semiconductors. This is a roughly 15% increase in the number of layers over conventional 108-layer designs. OTC is seeking to establish mass production technology by October 2025 at its Joetsu Plant in Joetsu City, Niigata Prefecture, which has a proven track record and advanced development and production capabilities in the field of high multilayer, high-precision, large-format PCBs for semiconductor inspection equipment.

AI processing requires the transmission of vast data volumes between graphics processing unit (GPU) semiconductors and memory. As semiconductor performance increases, the memory installed is also required to have high-speed, high-frequency, and high-density data transfer capabilities. HBM features a stacked DRAM structure, requiring technology capable of fabricating wafers even more thinly and precisely. This configuration also requires that the PCBs used in inspection equipment meet even higher levels of performance and quality.

BEHOLD THE ARGONATH, THE PILLARS OF THE KINGS! Standing guard over the northern borders of Gondor, the Argonath is a pair of statues without compare—even in a place like Middle- earth. Now, it's found a place on your desk. Follow the Fellowship South through storied statues. Depicted in stunning impressionist style by the talented artists at Dwarf Factory, the scene on this mat is pulled straight from The Fellowship of the Ring, following our nine heroes as they go south along the river Anduin.

GLIDE ACROSS A WATER-REPELLANT SURFACE WITH SPACE TO SPARE
Towering statues aren't the only breathtaking structures here. Nearly as impressive is our mat itself, measuring 36 inches wide, 16 inches tall, and 4 millimeters thick. It's also water-repellant-perfect for any accidental rivers across your desk. Work and play on mythical materials: made from high-density, low-friction Mercerized fabric and eco-friendly anti-slip rubber, this mat stays put and feels exceptional. It may just last as long as the Argonath itself.

Today, Intel introduced a new "200S Boost" factory-approved overclocking profile for the unlocked Intel Core Ultra 200S "Arrow Lake" processors. Namely, the Core Ultra 9 285K, 7 265K/265KF, and 5 245K/245KF SKUs are supported when installed on a compatible Intel Z890 motherboard with one‑DIMM‑per‑channel Intel XMP DDR5 memory. Activating the 200S Boost profile in the BIOS raises the System-on-Chip (fabric) clock from 2.6 GHz to up to 3.2 GHz and the die‑to‑die interconnect from 2.1 GHz to up to 3.2 GHz (within VccSA ≤ 1.20 V) while pushing DDR5 speeds from stock 6,400 MT/s to as much as 8,000 MT/s (VDD2/VDDQ ≤ 1.40 V). Intel has validated the profile on a selection of Z890 boards, including ASRock Z890 Taichi OCF, ASUS ROG Maximus Z890 Hero, Gigabyte Z890 Aorus Master, MSI MEG Z890 Ace, and others. For memory kits, Intel validated DDR5 memory kits from ADATA, Corsair, G.SKILL, Team Groupp, and V‑COLOR.

Enabling 200S Boost requires a BIOS update, selecting the "Intel 200S Boost" preset under the overclocking menu, and rebooting; stability should then be verified with benchmarks like Cinebench, and thermals/voltages monitored via Intel XTU or similar tools. Perhaps the most important fact is that using the 200S Boost profile does not void Intel's three‑year limited warranty on boxed Core Ultra 200S CPUs, provided they weren't manually overclocked before profile activation. Intel cautions that actual gains depend on motherboard design, cooling, and memory quality, that two‑DIMM‑per‑channel setups aren't officially supported, and that damage to non‑Intel components remains outside warranty coverage. This is more of a safe heaven for anyone wanting to do manual tuning, but not wanting to break any warranty and thus risk damaging their CPU without a backup plan.

Speed running scientists and challenge hunters, rejoice. After years of crafting, your invitation to Bionic Bay is now available on Steam and PlayStation 5. It began with a chance meeting on Reddit, and now the Bionic Bay story continues with you, our trusted Research Participants. In Bionic Bay you'll master the art of teleportation, control the very fabric of time itself, manipulate gravity and navigate a path filled with dangers. Our team will provide you with the tools you'll need to traverse this intricately crafted pixel art world. But don't let the stunning visuals fool you, this is no slow-moving narrative crawl. The best way to navigate Bionic Bay is with fast movements, precise controls, exceptional timing, puzzle solving intuition and just maybe a little bit of luck.

What you've seen in the Bionic Bay demo is just the beginning. Bionic Bay v1.0 includes 23 painstakingly created levels, with an increasing challenge to simply stay alive through your journey. Importantly, however, the path can be as helpful as it is dangerous, and Research Participants will need to judge the way ahead accordingly. In addition to the main game, we've also developed the Bionic Bay Online mode, with a dedicated build for speed runners and those hungry for a challenge. In this demanding arena, you'll have the opportunity to customize your own scientist, and claim your place on our transient and global leaderboards.

E-WIN is thrilled to announce the launch of its latest innovation, the E-WIN Breathable SoftWeave Fabric Flash XL Rev Edition (FLJ-XL-REV), poised to be the best gaming chair of 2025. This groundbreaking gaming chair redefines comfort, style, and performance for gamers and professionals alike. Inspired by the high-octane world of Formula 1, the FLJ-XL-REV combines advanced technology with a striking design, ensuring users experience both luxury and durability in their gaming setups.

The FLJ-XL-REV introduces cutting-edge features to elevate your seating experience. Equipped with magnetic headrests, magnetic side covers, and magnetic armrests, this chair allows for seamless customization, letting users adjust components effortlessly to suit their preferences. The 4D built-in lumbar support adapts to your spine, providing ergonomic support that promotes healthy posture during long gaming or work sessions. Crafted with breathable SoftWeave fabric, the chair ensures optimal airflow, keeping you cool and comfortable even during the most intense moments.

A great surface makes a whole lot of difference for PC gaming and work. It's part aesthetics, part utility. Here's what you need to know when shopping around. A mousepad is not just a simple accessory. SteelSeries QcK mousepads have proven, over and over, that they are best in gaming. They're washable and durable, feature a non-slip base, and micro-woven cloth optimized for low and high CPI mouse tracking. With more than 15 million sold, our iconic esports mousepads have enabled gamers around the globe to deliver top tier performance. And now, you can join them. Let's dive into a few reasons why.

Pick the exact style you want
The great thing about our new QcK Performance mousepad series—our latest—is that you can pick the kind of surface you want to adapt to your gameplay style. Do you prefer your mouse to glide super smoothly, or do you want more controlled precision with a little friction? Improved glide, more control, or something in between. This gaming mousepad is truly reflective of your style and preference.

Going back a year, we started hearing about an OpenAI proprietary AI chip project—this (allegedly) highly ambitious endeavor included grand plans for a dedicated fabrication network. TSMC was reportedly in the equation, but indirectly laughed at the AI research organization's ardent requests. Fast-forward to the present day; OpenAI appears to be actively pursuing a proprietary GPU design through traditional means. A Reuters exclusive report points to 2025 being an important year for the company's aforementioned "in-house" AI chip—the publication believes that OpenAI's debut silicon design has reached the finalization stage. Insiders have divulged that the project is only months away from being submitted to TSMC for "taping out." The foundry's advanced 3-nanometer process technology is reported to be on the cards. A Reuters source reckons that the unnamed chip features: "a commonly used systolic array architecture with high-bandwidth memory (HBM)...and extensive networking capabilities."

Broadcom is reportedly assisting with the development of OpenAI's in-house design—we heard about rumored negotiations taking place last summer. Jim Keller's tempting offer—of creating an AI chip for less than $1 trillion—was ignored early last year; OpenAI has instead assembled its own internal team of industry veterans. The October 2024 news cycle posited that former Google TPU engineers were drafted in as team leaders, with a targeted mass production window scheduled for 2026. The latest Reuters news article reiterates this projected timeframe, albeit dependent on the initial tape going "smoothly." OpenAI's chip department has grown to around forty individuals with recent months, according to industry moles—a small number relative to the headcounts at "Google or Amazon's AI chip program."

UK based premium office and gaming chair specialist, Boulies, today announces the arrival of its newest chair, the Master Rex. Following on from the best-selling Master Chair, the new Master Rex features an ergonomic design and is built for use for work, leisure and gaming. Featuring upgrades such as a built-in leg rest for added comfort and an Enhanced Stable Tilt Mechanism for outstanding stability and durability, it ensures a secure and comfortable experience.

The Boulies Master Rex offers unparalleled sitting comfort, whether working hard, gaming or relaxing. Designed with adjustable features and a sleek, modern aesthetic, it provides ergonomic support whilst blending with a home or office seamlessly. With an integrated leg rest, the Master Rex ensures pressure-free reclining for added relaxation. The built-in lumbar support also helps to promote better posture and boost productivity thanks to it being customizable with an up and down adjustment that allows users to fine-tune the lumbar support to precisely align with the spine's natural curvature. Whether requiring a higher or lower level of support, this chair allows users to customize to exact specifications to ensure it is adapted to the individual.

Geometric Future launched its Model 0 Flamingo "folding box" chassis series, half a year after it was showcased at the Computex 2024 Taipei Computer Show. The Model 0 Flamingo is a Mini-ITX chassis with a unique feature: it wraps around your system like a cardboard box. The chassis has a compact format measuring 81 × 206 × 208 mm with a volume of 3.47L. It is built from 1.2 mm thick steel plates coated with anti-static, fire-proof, and mildew-proof fabric in black color with touches of gold-like elements. For cooling, it relies on a giant mesh top circle and numerous ventilation holes. There is no room for additional fans, except it can host a 60 mm height CPU cooler. The chassis comes equipped with a compact custom Guitarra 200 W power supply manufactured by Quanhan that measures 70 × 35 × 170 mm.

As we said, the Geometric Future Model 0 unique feature is the ability to be folded without the need for any tools, with magnets included in the material of the case itself keeping it together. Additionally, detachable magnetic feet allow users to orient the case in both vertical and horizontal modes. However, as interesting as it looks, the Model 0 has some drawbacks: it is only compatible with ITX motherboards and does not support PCIe expansion (no graphics card can be added). Currently, there is no information about pricing. However, it should be around $100 if Geometric Future would like to attract SFF/HTPC users.

ENDORFY is unveiling the Crystal Series Mousepads as a new lineup of high-quality, visually stunning mousepads designed to elevate the gaming and workspace setups. These new mousepads not only mark ENDORFY's venture beyond the flagship Cordura series but also bring fresh designs and impeccable quality to the table, setting a new benchmark for affordable style and performance.

Eye-catching designs for every taste
The Crystal Series features an array of vibrant, printed graphic patterns, including the sleek Crystal Black, the elegant Crystal Onyx White, the vibrant Crystal Spectrum, and the soothing Crystal Blue, available in two versatile sizes (L and XL). Whether the user is into minimalist aesthetics or bold visuals, the Crystal Series adds a splash of personality to the desk while ensuring functionality remains paramount.

TSMC is reportedly making decent progress with its advanced 2 nm (N2) node—industry news pieces from earlier this month pointed to the initiation of production lines across three fabrication sites. Taiwan's Economic Daily News has kept close tabs on these trial runs—insiders have indicated that TSMC's Kaohsiung plant is capable of matching the Baoshan location's targeted manufacturing output (5000 wafers per month, 60 percent yield). Reports suggest that the Kaohsiung 2 nm trial production will start up later this month—much earlier than anticipated.

The Taiwanese chip foundry giant is taking on the challenge of meeting "greater than expected" demand for its new generation 2 nm product—TSMC chairman C.C. Wei has previously stated that its latest and greatest is more popular (pre-launch) with customers than older 3 nm lines. Apple is rumored to be first in line—not a big surprise since TSMC has (supposedly) rolled out the VVIP red carpet for them in recent times. The Economic Daily News article also mentions Qualcomm and MediaTek being next in the queue for N2. TSMC's best foundries are expected to initiate mass production by the end of 2025.

ASUS Republic of Gamers (ROG) today announced the ROG Slash Sling Bag 4.0, ROG Slash Sleeve 4.0, and ROG Backpack 4.0—all of which are new additions to the ROG urban gear range. Boasting bold aesthetics and functional designs, these bags are the ideal lifestyle travel gear for the urban jungle.

Compact commuter
The ROG Slash Sling Bag 4.0 offers a sleek and elegant carrying case for the ROG Ally or other gaming handhelds, complete with soft inner lining and custom cutouts to safely secure the device. Built from tear- and water-resistant materials, the Sling Bag 4.0 is designed to protect gear from inclement weather as well as the bumps and knocks of daily life. A magnetic Fidlock quick-release buckle enables easy access to the main compartment, while 360° snap hooks and adjustable shoulder straps ensure the bag is always perfectly suited to the user's body and preferred carrying position. Custom ROG monograms inside the bag and ROG logo on the exterior fabric accentuate the bag's bold style.

Last week, we got a preview of Microsoft's Azure production-ready NVIDIA "Blackwell" GB200 system, showing that only a third of the rack that goes in the data center is actually holding the compute elements, with the other two-thirds holding the cooling compartment to cool down the immense heat output from tens of GB200 GPUs. Today, Google is showing off a part of its own infrastructure ahead of the Google Cloud App Dev & Infrastructure Summit, taking place on October 30, digitally as an event. Shown below are two racks standing side by side, connecting NVIDIA "Blackwell" GB200 NVL cards with the rest of the Google infrastructure. Unlike Microsoft Azure, Google Cloud uses a different data center design in its facilities.

There is one rack with power distribution units, networking switches, and cooling distribution units, all connected to the compute rack, which houses power supplies, GPUs, and CPU servers. Networking equipment is present, and it connects to Google's "global" data center network, which is Google's own data center fabric. We are not sure what is the fabric connection of choice between these racks; as for optimal performance, NVIDIA recommends InfiniBand (Mellanox acquisition). However, given that Google's infrastructure is set up differently, there may be Ethernet switches present. Interestingly, Google's design of GB200 racks differs from Azure's, as it uses additional rack space to distribute the coolant to its local heat exchangers, i.e., coolers. We are curious to see if Google releases more information on infrastructure, as it has been known as the infrastructure king because of its ability to scale and keep everything organized.

Astera Labs, Inc, a global leader in semiconductor-based connectivity solutions for AI and cloud infrastructure, today announced a new portfolio of fabric switches, including the industry's first PCIe 6 switch, built from the ground up for demanding AI workloads in accelerated computing platforms deployed at cloud-scale. The Scorpio Smart Fabric Switch portfolio is optimized for AI dataflows to deliver maximum predictable performance per watt, high reliability, easy cloud-scale deployment, reduced time-to-market, and lower total cost of ownership.

The Scorpio Smart Fabric Switch portfolio features two application-specific product lines with a multi-generational roadmap:

• Scorpio P-Series for GPU-to-CPU/NIC/SSD PCIe 6 connectivity- architected to support mixed traffic head-node connectivity across a diverse ecosystem of PCIe hosts and endpoints.
• Scorpio X-Series for back-end GPU clustering-architected to deliver the highest back-end GPU-to-GPU bandwidth with platform-specific customization.

Researchers at Fraunhofer IAF have made a significant advance in semiconductor materials by successfully fabricating aluminum yttrium nitride (AlYN) using metal-organic chemical vapor deposition (MOCVD). AlYN, known for its outstanding properties and compatibility with gallium nitride (GaN), shows great potential for energy-efficient, high-frequency electronics. Previously, AlYN could only be deposited via magnetron sputtering, but this new method opens the door to diverse applications. Dr. Stefano Leone from Fraunhofer IAF highlights AlYN's ability to enhance performance while reducing energy consumption, making it vital for future electronics.

In 2023, the team achieved a 600 nm thick AlYN layer with a record 30% yttrium concentration. They have since developed AlYN/GaN heterostructures with high structural quality and promising electrical properties, particularly for high-frequency applications. These structures demonstrate optimal two-dimensional electron gas (2DEG) properties and are highly suitable for high electron mobility transistors (HEMTs).

A CPU-Z screenshot of an alleged Intel Core Ultra 9 285K "Arrow Lake-S" desktop processor engineering sample is doing rounds on social media, thanks to wxnod. CPU-Z identifies the chip with an Intel Core Ultra case badge with the deep shade of blue associated with the Core Ultra 9 brand extension, which hints at this being the top Core Ultra 9 285K processor model, we know it's the "K" or "KF" SKU looking at its processor base power reading of 125 W. The chip is built in the upcoming Intel Socket LGA1851. CPU-Z displays the process node as 7 nm, which corresponds with the Intel 4 foundry node.

Intel is using the same Intel 4 foundry node for "Arrow Lake-S" as the compute tile of its "Meteor Lake" processor. Intel 4 offers power efficiency and performance comparable to 4 nm nodes from TSMC, although it is physically a 7 nm node. Likewise, the Intel 3 node is physically 5 nm. If you recall, the main logic tile of "Lunar Lake" is being built on the TSMC N3P (3 nm) node. This means that Intel is really gunning for performance/Watt with "Lunar Lake," to get as close to the Apple M3 Pro as possible.

South Korean startup Panmnesia has unveiled an interesting solution to address the memory limitations of modern GPUs. The company has developed a low-latency Compute Express Link (CXL) IP that could help expand GPU memory with external add-in card. Current GPU-accelerated applications in AI and HPC are constrained by the set amount of memory built into GPUs. With data sizes growing by 3x yearly, GPU networks must keep getting larger just to fit the application in the local memory, benefiting latency and token generation. Panmnesia's proposed approach to fix this leverages the CXL protocol to expand GPU memory capacity using PCIe-connected DRAM or even SSDs. The company has overcome significant technical hurdles, including the absence of CXL logic fabric in GPUs and the limitations of existing unified virtual memory (UVM) systems.

At the heart of Panmnesia's solution is a CXL 3.1-compliant root complex with multiple root ports and a host bridge featuring a host-managed device memory (HDM) decoder. This sophisticated system effectively tricks the GPU's memory subsystem into treating PCIe-connected memory as native system memory. Extensive testing has demonstrated impressive results. Panmnesia's CXL solution, CXL-Opt, achieved two-digit nanosecond round-trip latency, significantly outperforming both UVM and earlier CXL prototypes. In GPU kernel execution tests, CXL-Opt showed execution times up to 3.22 times faster than UVM. Older CXL memory extenders recorded around 250 nanoseconds round trip latency, with CXL-Opt potentially achieving less than 80 nanoseconds. As with CXL, the problem is usually that the memory pools add up latency and performance degrades, while these CXL extenders tend to add to the cost model as well. However, the Panmnesia CXL-Opt could find a use case, and we are waiting to see if anyone adopts this in their infrastructure.Below are some benchmarks by Panmnesia, as well as the architecture of the CXL-Opt.

Presenting the release of Fractal Mood, a PC case designed to streamline the gaming station. It is a low-volume case in a space-saving pillar design format, which minimizes the footprint of the PC on the desk. Its outer body features breathable fabric in a clean shape and color palette to harmonize with the home.

For access to the interior, users can pop out the back panel and simply slide off the fabric-covered enclosure for full access to their build. Inside, Mood can house a graphics card up to 325 mm in length, which is easily connected through the included PCIe 4.0 riser cable.

PCI-SIG, the organization responsible for the widely adopted PCI Express (PCIe) standard, today announced the release of the CopprLink Internal and External Cable specifications. The CopprLink Cable specifications provide signaling at 32.0 and 64.0 GT/s and leverage well-established industry standard connector form factors maintained by SNIA.

"The CopprLink Cable specifications integrate PCIe cabling seamlessly with the PCIe electrical base specification, providing longer channel reach and topological flexibility," said Al Yanes, PCI-SIG President and Chairperson. "The CopprLink Cables are intended to evolve with the same connector form factors, scale for future PCIe technology generations and meet the demands of emerging applications. The Electrical Work Group has already begun pathfinding work on CopprLink Cables for PCIe 7.0 technology at 128.0 GT/s, showcasing PCI-SIG's commitment to the CopprLink Cable specifications."