News Posts matching #GPU

Return to Keyword Browsing

Microsoft Celebrates 10 Years of DirectX 12

DirectX 12 shipped in 2015 with a simple goal: give developers more control so games run faster, look better, and scale across Windows PC and console. Over the last decade, DirectX 12 delivered on that promise. We added features and made it easier for developers to focus on gameplay & graphics by providing more flexibility and achievable performance wins.

We built technologies that dramatically reduce load times and deliver consistently high frames for a polished gaming experience in the most demanding of titles. Our platform's continuous evolution brought advanced visuals previously unimaginable into the mainstream. Today, DirectX 12 sits at the heart of many AAA titles and modern game engines, powering some of the most visually stunning games ever made. Happy 10th birthday DirectX 12, here are some of the biggest milestones that made games better!

Rapidus Explores Glass Panel-Level Packaging to Rival TSMC

It seems that the Japanese foundry Rapidus is gearing up for strong competition with TSMC. The company has developed a panel-level packaging (PLP) prototype using a glass interposer, with plans to move towards mass production by 2028. At this week's SEMICON Japan 2025 in Tokyo, the prototype will be showcased. It replaces conventional round silicon wafers with large square glass panels, intended for use in high-performance AI packages that combine multi-chip GPUs and more than a dozen of HBM dies. This approach could reduce material waste and enable larger, denser multi-chip assemblies necessary for next-generation AI accelerators, without breaking the traditional laws of 300 mm wafer.

The company is developing 600x600 mm glass panels, which offer a significantly larger usable area than 300 mm silicon wafers and enable the production of more interposers from a single sheet. Glass substrates provide advantages in electrical performance and flatness compared to organic package materials, making them appealing for dense interconnects. However, glass also introduces new manufacturing challenges, such as fragility and potential warpage as panel sizes increase. Intel has notably discontinued its in-house development of glass substrates and has chosen to license the technology to others. The company can thus monetize its innovations, though large-scale Intel-based technology production is not anticipated until after 2030.

NVIDIA Plans to Reduce RTX 50 Production by Up to 40% in Early 2026

Reports from Asian supply chain sources suggest that NVIDIA might cut back on making GeForce RTX 50 series GPUs in the first six months of 2026, Videocardz reports. According to Board Channels, overall supply during H1 2026 could be down by roughly 30-40% compared to the same period in 2025, although the information remains unconfirmed. The report points out that the issue is not entirely related to GPU VRAM alone, but also to tight availability across GDDR6, GDDR7, and other memory components, including motherboard-related memory such as DDR5/DDR4. Board Channels also claims NVIDIA may adjust allocation strategies for add-in board (AIC) partners in Mainland China to better align supply with changing DIY market conditions.

Benchlife appears to support parts of the rumor, though without specifying a percentage reduction. Instead, it reports that two models (best value models in the Blackwell RTX 50 lineup), the GeForce RTX 5070 Ti and RTX 5060 Ti 16 GB are likely to see the earliest supply adjustments. While both Board Channels and Benchlife have long and good track records, the reports are based on regional sources and may not reflect global supply conditions. We will have to wait for NVIDIA or its partners to provide more information. However, it's worth noting that just a month ago, reports from Korean and Taiwanese outlets said AMD and NVIDIA were already considering GPU output cuts, while ASUS and other vendors were slowing motherboard plans due to memory shortages. In the end, this could end up being a broader, global production adjustment by NVIDIA rather than a region specific move.

NVIDIA Powers Nearly 72% of Cloud Accelerator Locations, AMD at 5.8%

The cloud accelerator market share study was conducted at UBS bank, one of the leading research hubs and close followers of market trends. According to UBS, NVIDIA commands 71.2% of the cloud accelerator location share, which includes GPUs and other types of ASICs. The study also included AMD, and other ASICs. For the AMD front, the situation is entirely different as the data shows AMD only captures 5.8% of the cloud accelerator location share. The remaining portion is split between ASICs designed by the likes of AWS, Meta, Broadcom, MediaTek, and many others, which command 22.3%.

For NVIDIA, the company has managed to penetrate its GPUs in 258 datacenter locations, serving as cloud GPU providers. This is a significant number considering that some of these customers are ordering tens and even hundreds of thousands of GPUs at once, especially as the demand expands. NVIDIA is still the backbone of the entire AI infrastructure, and even its older GPUs, which include A100 and H100, are capturing a significant part of the GPU cloud market share. As "Blackwell" SKUs, designed for a million GPU scale systems, are scaled, and we enter the "Rubin" era, NVIDIA's lead may only expand.

Enermax Launches REVOLUTION III S 1000 W Platinum ATX 3.1 PSU

ENERMAX, an industry-leading force dedicated designing high-performance computer hardware and cooling solutions, proudly announces the launch of the REVOLUTION III S 1000 W, a premium fully modular power supply designed to meet the latest standards in performance, reliability, and aesthetics. Available in both black and white, the REVOLUTION III S 1000 W offers enthusiasts and professionals the flexibility to build powerful systems without compromise. The REVOLUTION III S comes with a 13-year warranty that highlights ENERMAX's confidence in the product's reliability and long-term durability.

The newest addition to the REVOLUTION line is 80 Plus, Cybenetics, and PPLP Platinum certified, ensuring outstanding efficiency and reduced power consumption. Fully compliant with the Intel ATX 3.1 standard, the REVOLUTION III S guarantees stable and reliable power delivery for not only the latest generation of high-performance graphics cards and CPUs, but also the upcoming generations too. Equipped with one 12V-2x6 interface, the REVOLUTION III S 1000 W can deliver up to 600 W of dedicated power to the latest GPU, making it future-ready for enthusiast builds.

AMD's AIB Partner Wants More Design Freedom for Extreme OC GPUs

AMD's AIB partners are speaking up as they express more design freedom for their GPUs. According to Sapphire, AMD-exclusive GPU AIB, they are limited in their ability to modify the card and must adhere to chipmaker's official design rules. However, they now want more freedom to make GPUs focus on overclocking, extremely silent operation, and other traits that PC enthusiasts will appreciate. In a Hardware Unboxed interview with Ed Crisler, PR Manager at Sapphire Technology for North America, the AIB has expressed some frustration over their design rule limits.
Sometimes I really wish the chip makers would get out of the way and let us partners just make our cards. Give us the chip. Give us the RAM. Tell us what we have to provide to make it work with the board. And then let us make the cards. Let us have our fun. Let us go nuts. Let there be real differentiation. Sometimes it feels like this market becomes too too much the same.

SK Hynix Forecasts Tight Memory Supply Lasting Through 2028

SK Hynix held an internal company meeting, where the company reportedly presented some tough pills to swallow for many gamers. The company now forecasts the tight memory supply to last through 2028 for the commodity DRAM, which includes DDR5/DDR4, GDDR6/GDDR7, and LPDDR5x/LPDDR6. All the aforementioned DRAM variants are essential for PC and console components, making millions of gamers at risk of absorbing a massive price increase as a result. As memory supplier inventories deplete, production capacity will not increase to meet demand as it has in the past. This marks a departure from the usual response by memory manufacturers, who typically ramp up capacity in response to demand.

However, an interesting thing that SK Hynix has noted is that this situation will exclude the company's most advanced memory solutions—HBM and SOCAMM. These products are expected to get an additional capacity expansion, without tight supply impacting this part of the DRAM supply chain. These products are even in high demand as they are mostly consumed in higher volume than the regular commodity DRAM, due to their integration into AI products such as GPUs and servers.

Eurocom Launches 780 AC Adapter for Underpowered Dell Alienware 16/18 Area 51 Laptops

The newest generation of Alienware 16 and Alienware 18 Area-51 systems pushes high-end gaming and professional workloads to new levels with Intel Core Ultra (Arrow Lake-HX) processors, NVIDIA GeForce RTX 50-Series GPUs up to the RTX 5090, high-refresh QHD+ displays, advanced cooling, Wi-Fi 7, Thunderbolt 5, and optional mechanical CherryMX keyboards in a premium, reinforced chassis.

Despite these impressive hardware specifications, performance on these systems can still be held back by the standard 330 W / 360 W AC adapters. Under heavy or extended loads, the OEM power bricks often cannot supply enough sustained wattage, resulting in reduced turbo headroom, inconsistent frame rates, high adapter temperatures, throttling, or battery drain even while plugged in.

NVIDIA Reaffirms Support for FP64, Next-Gen GPU to Bring HPC Improvements

NVIDIA's recent generations of GPUs, such as "Hopper" and "Blackwell," have been known for stalling their high-performance computing (HPC) oriented FP64 double-precision performance. However, the company is not inherently abandoning the HPC space and has confirmed for HPCWire that 64-bit floating point data is not going away. According to Dion Harris, the senior director of HPC and AI hyperscale infrastructure solutions at NVIDIA, the company is "definitely looking to bring some additional [FP64] capabilities in our future gen architectures. We are very serious about making sure that we can deliver the required performance to power those simulation workloads."

The acceleration of 64-bit floating-point data paths is crucial for the HPC community, particularly in the life sciences. When a workload demands sustained high-precision support, NVIDIA's recent generations have not met expectations. For comparison, NVIDIA's current most powerful B300 "Blackwell Ultra" accelerator achieves only 1.2 TeraFLOPS of FP64 performance. In contrast, the older H200 "Hopper" reaches an impressive 34 TeraFLOPS of FP64 compute at its peak. For FP8 low-precision, the B300 delivers 9 PetaFLOPS, while the H200 provides 3.958 PetaFLOPS.

AMD Unveils Radeon AI PRO R9700S and R9600D GPUs

AMD has officially expanded its Radeon AI PRO R9000 workstation lineup with two new Navi 48-based cards built on the RDNA 4 architecture. The two models are the Radeon AI PRO R9700S and the Radeon AI PRO R9600D. Both models feature 32 GB of GDDR6 memory on a 256-bit interface, delivering a peak memory bandwidth of 640 GB/s. These additions offer a consistent hardware platform for professional customers who require larger frame buffer sizes for compute and visualization tasks.

The Radeon AI PRO R9700S maintains the full Navi 48 configuration with 64 CUs and 4,096 SPs. AMD rates the card at up to 47.8 TeraFLOPS of FP32 performance when the boost clock reaches 2,920 MHz, with a total board power of 300 W. The Radeon AI PRO R9600D, on the other hand, reduces the Navi 48 configuration to 48 active CUs and 3,072 SPs, offering approximately 24.8 TeraFLOPS of FP32 throughput at a boost clock of 2,020 MHz, while targeting a lower 150 W board power.

NVIDIA's GPU-powered Accelerated Computing Platforms Have Replaced CPUs as the Engine of Invention

The NVIDIA accelerated computing platform is leading supercomputing benchmarks once dominated by CPUs, enabling AI, science, business and computing efficiency worldwide. Moore's Law has run its course, and parallel processing is the way forward. With this evolution, NVIDIA GPU platforms are now uniquely positioned to deliver on the three scaling laws—pre-training, post-training and test-time compute—for everything from next-generation recommender systems and large language models (LLMs) to AI agents and beyond.

The CPU-to-GPU Transition: A Historic Shift in Computing
At SC25, NVIDIA founder and CEO Jensen Huang highlighted the shifting landscape. Within the TOP100, a subset of the TOP500 list of supercomputers, over 85% of systems use GPUs. This flip represents a historic transition from the serial-processing paradigm of CPUs to massively parallel accelerated architectures. Before 2012, machine learning was based on programmed logic. Statistical models were used and ran efficiently on CPUs as a corpus of hard-coded rules. But this all changed when AlexNet running on gaming GPUs demonstrated image classification could be learned by examples. Its implications were enormous for the future of AI, with parallel processing on increasing sums of data on GPUs driving a new wave of computing.

DEEPX Launches DX-H1 V-NPU: The 30W Single-Card Solution That Challenges GPU Dominance

Ahead of CES 2026, DEEPX, a pioneer in ultra-low-power on-device AI semiconductors, today announced the official launch of the DX-H1 vision neural processing unit (V-NPU), a dedicated video intelligence chipset capable of processing hundreds of AI video channels at just 30 W.

Redefining Video AI Infrastructure
The DX-H1 V-NPU is an all-in-one video intelligence solution combining a multi-channel engine capable of decoding, encoding and transcoding with dedicated NPU architecture. Unlike traditional setups that require parallel configurations of GPU servers and hardware codecs, the DX-H1 V-NPU handles the entire pipeline on a single chip -- from stream input to preprocessing, AI inference, and re-encoding.

Moore Threads CEO Teases Next-gen GPU Architecture Ahead of MUSA Developer Conference

Moore Threads, a growing force within China's domestic graphics card development industry, has scheduled its very first MUSA Developer Conference (MDC 2025). The Beijing-based event will kick of on December 19, and wrap up a day later. Earlier today, Zhang Jianzhong—the company's founder, chairman and CEO—issued a pre-game primer. Primarily, his upcoming keynote presentation will focus on full-stack development strategy—utilizing Moore Threads Unified System Architecture (MUSA). The former NVIDIA China executive also mentioned an imminent unveiling of his firm's next-generation GPU design. This teased future technology is set to succeed previous attempts, mostly launched three years ago—TechPowerUp has covered previous-gen "Chunxiao" MTT S80 16 GB (mainstream desktop) and MTT S4000 48 GB (AI workstation-grade) card models in the recent past.

Moore Threads leadership is likely keen ride a large wave of momentum—mostly generated by a $1.1 billion listing, following the company's early December debut IPO. Last week, this valuation buzz generated lofty headlines, with a couple of titles dubbing Moore Threads as the "Chinese NVIDIA." This month's MUSA Developer Conference seems to be aimed at the enterprise market segment, so desktop gaming products are not expected to feature prominently (if at all) during a run through of the Moore Threads hardware roadmap. According to local news reports, Chinese GPU specialists are expected to rapidly respond to NVIDIA's newly-awarded export license—clearing the way for a major injection of H200 data center accelerators into the region.

ASUS Denies GeForce RTX 5090 ROG MATRIX Recall

ASUS has officially denied a recall of its limited edition GeForce RTX 5090 ROG MATRIX graphics card, which was reportedly facing some quality issues. However, ASUS was quick to dispute these claims. In the past, an ASUS reportedly representative informed Swedish e-tailer "Inet" that they had identified a quality issue with the RTX 5090 32 GB ROG Matrix Platinum 30th Anniversary Edition, preventing planned deliveries while they worked on an undated replacement. However, in ComputerBase update from Christian Wefers, part of ASUS Public Relations, the situation is now clearer:
Christian Wefers, ASUS PRThere is no product recall for the ROG Matrix GeForce RTX 5090. We proactively adjusted the original timeline to allow additional time for product improvements and to ensure an optimal customer experience. The products will arrive at our partners within the next few days. Availability may vary by region.

Imec 3D HBM-on-GPU STCO Study Shows Major Thermal Improvements

This week, at the 2025 IEEE International Electron Devices Meeting (IEDM), imec, world-leading research center in advanced semiconductor technologies, presents the first thermal system-technology co-optimization (STCO) study of 3D HBM-on-GPU (high-bandwidth memory on graphical processing unit), a promising compute architecture for next-gen AI applications. By combining technology and system-level mitigation strategies, peak GPU temperatures could be reduced from 140.7°C to 70.8°C under realistic AI training workloads - on par with current 2.5D integration options. The result demonstrates the strength of combining cross-layer optimization (i.e., co-optimizing the knobs at all the different abstraction layers) with broad technological expertise, a combination that is unique to imec.

Integrating high bandwidth memory (HBM) stacks directly on top of graphical processing units (GPUs) offers an attractive approach for building next-gen compute architectures for data-intensive AI workloads. This 3D HBM-on-GPU promises a huge leap forward in compute density (with four GPUs per package), memory per GPU, and GPU-memory bandwidth compared to current 2.5D integration options where HBM stacks are placed around (one or two) GPUs on a silicon interposer. However, the aggressive 3D integration approach is prone to thermal issues because of higher local power density and vertical thermal resistance.

Valve Trying to Unblock Open-Source HDMI 2.1 Support on Steam Machine

Valve is reportedly attempting to persuade the HDMI Forum—the consortium responsible for HDMI development—to allow open-source support for HDMI 2.1 on Linux, enabling the standard to function without restrictions on SteamOS. When Valve launched its Steam Machine, it was listed with HDMI 2.0 support, which seemed an odd choice given that HDMI 2.1 had become more prevalent. However, the Steam Machine does indeed feature an HDMI 2.1 port, but due to the HDMI Forum's stance against open implementations, Valve officially listed it as supporting the older HDMI 2.0.

According to an interview with Ars Technica, Valve is now trying to clarify this situation. The company has been "working on trying to unblock things there" to gain official support and approval for HDMI 2.1 from the HDMI Forum. In early 2024, AMD proposed incorporating key HDMI 2.1 features, such as 4K at 120 Hz and 5K at 240 Hz, into their open-source Linux graphics driver, AMDGPU. However, the HDMI Forum quickly criticized the proposal, and AMD stated: "The HDMI Forum has rejected our proposal, unfortunately. At this time, an open-source HDMI 2.1 implementation is not possible without violating HDMI Forum requirements."

NVIDIA Announces CUDA Tile with CUDA 13.1

With its largest advancement since the NVIDIA CUDA platform was invented in 2006, CUDA 13.1 is launching NVIDIA CUDA Tile. This exciting innovation introduces a virtual instruction set for tile-based parallel programming, focusing on the ability to write algorithms at a higher level and abstract away the details of specialized hardware, such as tensor cores. CUDA exposes a single-instruction, multiple-thread (SIMT) hardware and programming model for developers. This requires (and enables) you to exhibit fine-grained control over how your code is executed with maximum flexibility and specificity. However, it can also require considerable effort to write code that performs well, especially across multiple GPU architectures.

There are many libraries to help developers extract performance, such as NVIDIA CUDA-X and NVIDIA CUTLASS. CUDA Tile introduces a new way to program GPUs at a higher level than SIMT. With the evolution of computational workloads, especially in AI, tensors have become a fundamental data type. NVIDIA has developed specialized hardware to operate on tensors, such as NVIDIA Tensor Cores (TC) and NVIDIA Tensor Memory Accelerators (TMA), which are now integral to every new GPU architecture. With more complex hardware, more software is needed to help harness these capabilities. CUDA Tile abstracts away tensor cores and their programming models so that code using CUDA Tile is compatible with current and future tensor core architectures.

AMD Radeon Price Hike Confirmed at $10 per 8GB of VRAM, More Increases Expected in 2026

AMD's rumored price hike at the end of 2025 is becoming a reality, as confirmed by anonymous GPU distributors to Tom's Hardware. The increase is a more modest $10 per 8 GB of VRAM. This means a $10 increase for 8 GB models like the Radeon RX 9060 XT 8 GB, and a $20 increase for 16 GB models like the Radeon RX 9060 XT 16 GB and RX 9070 / 9070 XT. This adjustment is just half of what was anticipated based on an earlier report from Chinese Board Channels, which suggested AMD was considering a $20 increase for 8 GB models and a $40 increase for 16 GB models.

However, this is not the end of price hikes. AMD is expected to implement another increase in January 2026, though the specifics are currently unknown. Earlier information from Board Channels suggested that gamers might see a price increase of 300 RMB ($42) for 8 GB models and 600 RMB ($85) for 16 GB SKUs. As we approach the final weeks of 2025, these details remain unconfirmed. AMD's AIB partners will pass these increases to customer, who will absorb any additional cost of manufacturing.

Lisuan Technology's 7G106 Card Could be First dGPU to Offer "Windows on Arm" Support

Chinese GPU firms are bravely attempting to catch up with Western designs. Lisuan Tech is a major contender; having introduced the nation's first-ever 6 nm gaming GPU back in June. The startup's 7G106 graphics card model is based on in-house "TrueGPU" architecture. According to a fresh ITHome article, video footage of working hardware emerged online late last month. Surprisingly, the alleged Windows-based demonstration rig utilized an Arm-based P1 CP8180 12-core CPU. Elaborating further, the publication states that the 6 nm AI PC processor is a domestically designed product.

A nearby placard, captured on camera, makes a number of claims. ITHome has interpreted the demo system's major credentials, crucially that it provides: "a complete Windows on Arm (WoA) PC platform solution. Currently, graphics card manufacturers such as NVIDIA, AMD, and Intel have not officially supported WoA. Is it possible for Lisuan to achieve a leapfrog development?" A 3DMark Steel Nomad session (under DirectX 12) points to Lisuan Tech's development of a working ARM64 Windows driver. The young company is working towards a full launch of its 7G106 12 GB gaming card, and a professional 7G105 24 GB model.

Intel 4-Core "Panther Lake" Xe3 iGPU Outperforms AMD Radeon 840M by 26%

Intel recently added another "Panther Lake" engineering sample to public benchmarking records. Geekbench reported a Core Ultra 7 366H processor featuring a four-core Xe3 integrated GPU built on the Intel 3 node. The system achieved a Vulkan API score of 22,813, placing this compact Xe3 configuration within the entry-level discrete GPU category. This score is roughly 26% higher than the 4-core AMD Radeon 840M and slightly surpasses the mobile GTX 1050 Ti, though it remains below the 8-core Radeon 860M. This indicates that Intel is now reaching performance levels similar to AMD's "Krackan Point" APUs with RDNA 3.5, making it suitable for gaming on the go.

The processor details in the listing match previous leaks. The Core Ultra 7 366H has a 16-core design, which includes 4 performance cores, 8 efficient cores, and 4 low-power efficient cores. It operates at a base clock speed of 2.00 GHz and can boost up to 4.8 GHz. The processor also includes approximately 18 MB of L3 cache. The Xe3 GPU increases its L2 cache to 4 MB in the 4-core configuration and 16 MB in the 12-core configuration, equating to 1 MB per core on the smaller tile and 1.5 MB per core on the larger tile. Additionally, Xe3 features updated ray tracing units with faster bounding-box traversal and a larger BVH cache, enhancing performance consistency in hybrid-rendered games. Nonetheless, this is an interesting preview of the upcoming "Panther Lake" Xe3 iGPU, and we expect to see more at CES 2026.

Samsung HBM4 Development Complete, Mass Production Pending NVIDIA Approval

Samsung has reportedly finished development of its first sixth-generation HBM (HBM4) chips and is now in the final pre-mass-production stage. According to new reporting from AjuNews and Korean industry sources, Samsung's Device Solutions division has cleared Production Readiness Approval (PRA), the last internal checkpoint before full-volume manufacturing. HBM4 is expected to deliver roughly 60% higher bandwidth than the current HBM3E chips. Samsung has already shipped HBM4 samples to NVIDIA, which is evaluating them for use in its upcoming Rubin platform. Initial samples have already exceeded NVIDIA's next-gen GPU requirement of 11 Gbps per pin. The company is using improved 1c-class DRAM paired with a 4 nm logic base die, a combination that helps close the gap with rivals while keeping thermals and power in check at these higher speeds. Once NVIDIA signs off, Samsung can move straight into volume production, with manufacturing lines already prepared.

Samsung had already signaled strong progress during its Q3 2025 earnings call on October 30, noting that HBM4 samples were in the hands of global customers and that mass production is planned for 2026. "HBM3E is currently in mass production and being sold to all related customers, while HBM4 samples are simultaneously being shipped to key clients," the company said at the time. Samsung also confirmed that its foundry arm will prioritize stable 2 nm GAA output and HBM4 base-die production in 2026, alongside the ramp of its new Taylor, Texas fab. The company is also said to be working on a faster HBM4 variant targeting another 40% performance uplift, with an announcement possible as early as mid-February 2026.

AMD Reportedly Raising 8GB, 16GB GPU SKUs by $20-$40

AMD is finalizing its GPU price increase strategy for end-of-the-year 2025. According to a new report from Chinese Board Channels, AMD is looking to increase its 8 GB SKUs pricing by $20, while the 16 GB models will absorb a $40 price increase. Overall, gamers are expected to see a price increase of 300 RMB ($42) for 8 GB models and 600 RMB for 16 GB SKUs, which is nearly $85. The price increase for memory alone will prompt AIBs to push these aforementioned price increases, confirming what we have been hearing within the industry over the past few weeks, with insights from PowerColor and others.
Board Channels ChinaAMD has officially announced a price increase for its graphics cards, with a significant increase expected this year. It is reported that several AMD graphics card brands have officially notified their channels to prepare inventory. The first wave of price increases will be $20 for 8 GB and $40 for 16 GB, with an estimated overall increase of around 300 RMB for 8 GB and 600 RMB for 16 GB by the end of the year. No new products will be launched throughout 2027.

AMD and HPE Expand Collaboration to Advance Open Rack-Scale AI Infrastructure

Today, AMD announced an expanded collaboration with HPE to accelerate the next generation of open, scalable AI infrastructure built on AMD leadership compute technologies. HPE will become one of the first system providers to adopt the AMD "Helios" rack-scale AI architecture, which will integrate a purpose-built HPE Juniper Networking scale-up switch - in collaboration with Broadcom - and software for seamless, high-bandwidth connectivity over Ethernet. "Helios" combines AMD EPYC CPUs, AMD Instinct GPUs, AMD Pensando advanced networking and the AMD ROCm open software stack to deliver a cohesive platform optimized for performance, efficiency, and scalability. The system is engineered to simplify deployment of large-scale AI clusters, enabling faster time to solution and greater infrastructure flexibility across research, cloud, and enterprise environments.

"HPE has been an exceptional long-term partner to AMD, working with us to redefine what is possible in high-performance computing," said Dr. Lisa Su, Chair and CEO, AMD. "With 'Helios', we're taking that collaboration further, bringing together the full stack of AMD compute technologies and HPE's system innovation to deliver an open, rack-scale AI platform that drives new levels of efficiency, scalability, and breakthrough performance for our customers in the AI era."

NVIDIA GeForce RTX 5070 Overcomes RTX 4070 in Steam Survey Rankings

NVIDIA's newest GeForce RTX 5070 "Blackwell" GPU has surpassed the RTX 4070 "Ada" in the latest Steam Hardware Survey, achieving a 2.23% share for November 2025, compared to the RTX 4070's 2.16%. With +0.35% month-on-month (MoM) gain, this makes the RTX 5070 the most widely used RTX 50 series card on the Steam platform so far and represents the largest MoM share increase among NVIDIA's new generation offerings during that period. This result is significant as it provides one of the final insights into PC hardware trends for 2025, offering a near-year-end view of how recent GPU launches have been embraced by the gaming community.

The overall Steam data continues to show NVIDIA's dominance in discrete GPU usage, with a 73.83% market share among surveyed Steam users, while AMD and Intel hold 18.05% and 7.74%, respectively. The top positions on the platform are still occupied by a mix of laptop and desktop models, led by the RTX 4060 Laptop at 4.44%. AMD's latest RDNA 4-based Radeon RX 9000 series have yet to make a significant impact in the survey, indicating their share is minimal for now. NVIDIA's progression from 40-series generation to 50-series generation seems to be more pronounced within the gaming community, as the change of share seems to happen faster than with AMD.

AMD Radeon RX 9000 Series Still Absent from Steam Hardware Survey

As we enter December, Valve has completed processing data and officially released the November Steam Hardware and Software Survey, highlighting trends in the gaming industry. Interestingly, AMD's latest RDNA 4 designs from the Radeon RX 9000 series have yet to make a significant impact on the market, indicating that many Steam survey participants are not using AMD Radeon RX 9000 series GPUs. The top spot in the GPU usage list is held by NVIDIA's GeForce RTX 4060 (notebook), with the first AMD product appearing nine spots later. Valve lists this first AMD Radeon GPU as an unnamed entry, likely representing integrated GPUs or other systems-on-chip products with unconventional configurations compared to standard discrete GPUs.

The first AMD discrete GPU on the list is the older RDNA 2-based Radeon RX 6600, followed by the RDNA 3-based RX 7800 XT. However, nine months after the RDNA 4 series began selling, there are no RDNA 4 GPUs present in the general survey. This could suggest that Steam gamers prefer other options or that the RDNA 4 series isn't selling well. While we don't have a definitive conclusion, the data points to a concerning trend.
Return to Keyword Browsing
Nov 22nd, 2026 17:16 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts