#HBM3E

[electronics-dev]

🌟 HBM3e: Redefining the Future of High-Performance Memory 🌟

As we step into 2025, High Bandwidth Memory (HBM) is shaping the future of AI and High-Performance Computing (HPC). Among the latest innovations, HBM3e emerges as a game-changer, offering unprecedented speed, capacity, and efficiency.

💡 What Makes HBM3e Unique?

📊 16-Hi Stacks: Expanding capacity from 36 GB to 48 GB per stack.

🚀 Unmatched Speed: Achieving 1.2 TB/s bandwidth per stack.

🔥 Advanced Technology: MR-MUF and TSV ensure durability, heat management, and efficient data transfer.

🎯 NVIDIA’s Integration NVIDIA is setting benchmarks by incorporating HBM3e into its next-gen GPUs, enabling faster AI training, improved inference, and unparalleled performance for data centers and AI servers.

🌍 The Big Picture With the demand for AI and machine learning solutions soaring, HBM3e is driving a pivotal shift in memory technology. The market for high-performance memory is expected to double by 2025, and the development of HBM4 promises even greater advancements in the years ahead.

🔗 Ready to explore more? Discover how HBM3e is transforming the industry and shaping the future of computing!

[govindhtech]

AMD Instinct MI325X Accelerators Lead AI Performance

The AMD Pensando Pollara 400 NIC, AMD Pensando Salina DPU, and AMD Instinct MI325X accelerators are the newest networking and accelerator solutions that AMD unveiled today. These solutions will enable the next generation of AI infrastructure at scale. Performance benchmarks for Gen AI models and data centers are raised using AMD Instinct MI325X accelerators.

Based on the AMD CDNA 3 architecture, AMD Instinct MI325X accelerators are engineered to deliver outstanding performance and efficiency for challenging AI activities that include inferencing, fine-tuning, and foundation model training. With the help of these components, AMD partners and customers may develop AI systems that are highly efficient and optimized at the system, rack, and data center levels.

AMD Instinct MI325X Continues to Provide Superior AI Performance

Industry-leading memory capacity and bandwidth are provided by AMD Instinct MI325X accelerators; 256GB of HBM3E supporting 6.0TB/s offers 1.8X more capacity and 1.3X more bandwidth than the H200. In addition, the AMD Instinct MI325X has 1.3X higher peak theoretical compute performance for FP16 and FP8 than the H200.

Up to 1.3X, 1.2X, and 1.4X the inference performance on Mistral 7B at FP16, Llama 3.1 70B at FP8, and Mixtral 8x7B at FP16 of the H200, respectively, may be obtained with this leadership memory and compute.

Currently scheduled for production shipments in Q4 2024, AMD Instinct MI325X accelerators are anticipated to be widely available for use in systems from a variety of platform providers beginning in Q1 2025, including Dell Technologies, Eviden, Gigabyte, Hewlett Packard Enterprise, Lenovo, Supermicro, and others.

AMD showcased its upcoming AMD Instinct MI350 series accelerators, continuing its dedication to an annual roadmap cadence. In comparison to AMD CDNA 3-based accelerators, AMD Instinct MI350 series accelerators are built on the AMD CDNA 4 architecture and are intended to provide a 35x increase in inference performance.

The AMD Instinct MI350 series, which offers up to 288GB of HBM3E memory per accelerator, will continue to lead the market in memory capacity. The second part of 2025 is when the AMD Instinct MI350 series accelerators are expected to be ready.

AMD Next-Gen AI Networking

The most popular programmable DPU for hyperscalers is being used by AMD to support next-generation AI networking. AI networking, which is divided into two components the front-end, which provides data and information to an AI cluster, and the backend, which controls data transmission between accelerators and clusters is essential to making sure CPUs and accelerators are used effectively in AI infrastructure.

The AMD Pensando Pollara 400, the first AI NIC in the industry ready for the Ultra Ethernet Consortium (UEC), and the AMD Pensando Salina DPU were introduced by AMD to efficiently manage these two networks and promote high performance, scalability, and efficiency throughout the system.

The third iteration of the most powerful and programmable DPU in the world, the AMD Pensando Salina, offers up to two times the speed, bandwidth, and scalability of its predecessor. Optimizing performance, efficiency, security, and scalability for data-driven AI applications, the AMD Pensando Salina DPU is a crucial part of AI front-end network clusters, supporting 400G throughput for fast data transfer rates.

The AMD Pensando Salina DPU and AMD Pensando Pollara 400 are scheduled to be available in the first half of 2025, and they are now sampling with consumers in Q4 of 2024.

New Generative AI Capabilities Offered by AMD AI Software

In order to provide the AMD ROCm open software stack with powerful new features and capabilities, AMD keeps investing in expanding software capabilities and the open ecosystem.

Among the most popular AI frameworks, libraries, and models, such as PyTorch, Triton, Hugging Face, and many more, AMD is promoting support for AMD compute engines within the open software community. For well-known generative AI models like Stable Diffusion 3, Meta Llama 3, 3.1, and 3.2, as well as the more than one million models at Hugging Face, this work translates to unconventional performance and support with AMD Instinct accelerators.

With the addition of the newest features to support cutting-edge training and inference on generative AI workloads, AMD is further developing its ROCm open software stack outside of the community. Flash Attention 3, Kernel Fusion, FP8 datatype, and other important AI capabilities are now supported by ROCm 6.2. For a range of LLMs, ROCm 6.2 offers up to a 2.4X performance boost on inference and 1.8X on training when compared to ROCm 6.0.

Read more on govindhtech.com

[pulsaris]

NVIDIA HGX H200 - HBM3e

A NVIDIA anunciou a disponibilização de uma nova plataforma de computação com a designação de HGX H200. Esta nova plataforma é baseada na arquitectura NVIDIA Hopper e utiliza memória HBM3e (um modelo avançado de memória com mais largura de banda e maior capacidade de memória utilizável).

Este é o primeiro modelo a utilizar memória HBM3e, introduzindo outras novidades tecnológicas no processador e na GPU que vão ao encontro das exigências dos mais recentes “modelos de linguagem” complexos e dos projectos de “deep learning”.

A plataforma está optimizada para utilização em Datacenters (centros de dados) e estará comercialmente disponível no segundo semestre de 2024.

Saiba tudo em detalhe na página oficial da NVIDIA localizada em: https://nvidianews.nvidia.com/news/nvidia-supercharges-hopper-the-worlds-leading-ai-computing-platform

______ Direitos de imagem: © NVIDIA Corporation (via NVIDIA Newsroom).

[smbomcom]

Samsung HBM3E

At the heart of the Samsung HBM3E memory lies a manufacturing process that's nothing short of revolutionary. The memory is constructed using an Extreme Ultraviolet (EUV) lithography process, which is the fourth generation of 10nm technology. To be precise, it operates at an effective 14nm level.

Get more details: Samsung HBM3E memory

[uniprismnew]

Next Generation Memory Market Resilience and Risk Factors Impacting Growth to 2033

The Next-Generation Memory (NGM) market is undergoing a transformative phase, driven by the escalating demands of artificial intelligence (AI), Internet of Things (IoT), and data-intensive applications. These advanced memory technologies, including Magnetoresistive RAM (MRAM), Resistive RAM (ReRAM), Phase-Change RAM (PCRAM), and High Bandwidth Memory (HBM), are poised to redefine data storage and processing paradigms.

Key Technologies and Applications

High Bandwidth Memory (HBM): Advanced memory solutions like HBM3E and HBM4 are being developed to meet the high-speed data processing requirements of AI and machine learning applications. For instance, Nvidia has requested SK Hynix to expedite the supply of HBM4 chips to support its AI technology development.

MRAM and ReRAM: These non-volatile memory technologies offer high endurance and low power consumption, making them ideal for IoT devices and embedded systems. Their integration into these devices ensures reliable data storage and fast access speeds.

PCRAM: Utilized in applications requiring high-speed data access and low latency, such as in-memory computing and real-time data processing systems.

Download a Free Sample Report: - https://tinyurl.com/yshzhzxy

Market Drivers

AI and Machine Learning: The surge in AI applications necessitates memory solutions that can handle vast amounts of data with minimal latency. Next-generation memory technologies are tailored to meet these demands, offering high-speed data access and processing capabilities.

IoT Expansion: The proliferation of IoT devices generates enormous data volumes, requiring efficient and reliable memory solutions. NGM technologies like MRAM and ReRAM provide the necessary speed and endurance to support the growing IoT ecosystem.

Data Center Evolution: Modern data centers require memory solutions that balance speed, capacity, and energy efficiency. NGM technologies are being integrated into data centers to enhance performance and reduce energy consumption.

Challenges and Considerations

High Manufacturing Costs: The production of next-generation memory technologies involves complex processes and specialized materials, leading to higher manufacturing costs compared to traditional memory solutions.

Integration Complexities: Integrating NGM technologies into existing systems and infrastructure can be challenging, requiring significant investment in research and development and potential redesigns of hardware components.

Future Outlook

The next-generation memory market is poised for substantial growth, driven by technological advancements and the increasing demands of AI, IoT, and data-intensive applications. As companies continue to innovate and develop more efficient and cost-effective memory solutions, the adoption of NGM technologies is expected to accelerate across various industries.

Recent Industry Developments

Nvidia's Request to SK Hynix: Nvidia has asked SK Hynix to expedite the delivery of HBM4 chips by six months to support its AI technology development, highlighting the growing need for advanced memory solutions in AI applications.

Micron's Expansion Plans: Micron Technology is expanding its market share in high-bandwidth memory chips, with new R&D facilities in the U.S. and potential production in Malaysia, aiming to meet the increasing demand for advanced memory solutions.

Read Full Report: - https://www.uniprismmarketresearch.com/verticals/semiconductor-electronics/next-generation-memory

[jcmarchi]

Samsung AI strategy delivers record revenue despite semiconductor headwinds

New Post has been published on https://thedigitalinsider.com/samsung-ai-strategy-delivers-record-revenue-despite-semiconductor-headwinds/

Samsung AI strategy delivers record revenue despite semiconductor headwinds

Samsung Electronics’ strategic focus on AI has delivered high revenue in the first quarter of 2025, as the South Korean tech giant navigates semiconductor market challenges and growing global trade uncertainties.

The company posted an all-time quarterly high revenue of KRW 79.14 trillion ($55.4 billion), marking a 10% increase year-over-year, according to its financial results released on Wednesday. Operating profit climbed to KRW 6.7 trillion ($4.68 billion), representing a modest 1.5% increase compared to the same period last year.

The results exceeded Samsung’s earlier forecast of KRW 79 trillion and analysts’ expectations.

Smartphone success counters chip challenges

The Mobile Experience (MX) Business emerged as the best performer, contributing KRW 37 trillion in consolidated revenue and KRW 4.3 trillion in operating profit – its highest level in four years. The success was driven primarily by strong sales of the flagship Galaxy S25 series, which features AI abilities via Galaxy AI.

“Enhanced cost competency and price declines for some components also contributed to solid double-digit profitability,” the company’s earnings report said.

In contrast, Samsung’s Device Solutions (DS) Division, which includes its semiconductor operations, posted KRW 25.1 trillion in revenue and KRW 1.1 trillion in operating profit – a 42% decline from the previous year. The performance reflects ongoing challenges in the semiconductor market, particularly in high-bandwidth memory (HBM) sales.

“Overall earnings were impacted by the erosion of average selling price (ASP), as well as a decrease in HBM sales due to export controls on AI chips and deferred demand in anticipation of upcoming enhanced HBM3E products,” Samsung said.

Trade tensions cloud future outlook

Despite the record revenue, Samsung has expressed caution about the second quarter, dropping its usual business outlook due to growing macroeconomic uncertainties stemming from global trade tensions and slowing economic growth.

“Due to the rapid changes in policies and geopolitical tensions among major countries, it’s difficult to accurately predict the business impact of tariffs and established countermeasures,” a Samsung executive stated during Wednesday’s earnings call.

Of particular concern are US President Donald Trump’s “reciprocal” tariffs, most of which have been suspended until July but threaten to impact dozens of countries including Vietnam and South Korea, where Samsung produces smartphones and displays.

While Samsung noted that its flagship products like semiconductors, smartphones, and tablets are currently exempt from these tariffs, the company revealed that Washington is conducting a product-specific tariff probe into these categories.

“There are a lot of uncertainties ahead of us […] we are communicating with related countries to minimise negative effects,” Samsung said during the call.

In response to its challenges, the company disclosed it is considering relocating production of TVs and home appliances.

AI investment and future strategy

Despite these headwinds, Samsung remains committed to its artificial intelligence strategy, allocating its highest-ever annual R&D expenditure for 2024. In the first quarter of 2025, the company increased R&D spending by 16% compared to the same period last year, amounting to KRW 9 trillion.

For the remainder of 2025, Samsung plans to expand its AI smartphone lineup through the introduction of “Awesome Intelligence” to the Galaxy A series and the launch of the Galaxy S25 Edge in Q2. Later in the year, the company will strengthen its foldable lineup with enhanced AI user experiences.

In the semiconductor space, Samsung aims to strengthen its position in the high-value-added market through its server-centric portfolio and the ramp-up of enhanced HBM3E 12H products to meet initial demand. The company expects AI-related demand to remain high in the second half of 2025, coinciding with the launch of new GPUs.

“In the mobile and PC markets, on-device AI is expected to proliferate, so the Memory Business will proactively respond to this shift in the business environment with its industry-leading 10.7Gbps LPDDR5x products,” Samsung stated.

The company’s foundry business remains focused on its 2nm Gate-All-Around (GAA) process development, which remains on schedule despite current challenges.

Market reaction and competitive landscape

Samsung shares were trading down approximately 0.6% following the announcement, reflecting investor concerns about the uncertain outlook.

The results highlight Samsung’s complex position in the AI market – succeeding in consumer-facing applications while working to catch up with competitors in AI-specific semiconductor components.

Local rival SK Hynix, which reported a 158% jump in operating profit last week to KRW 7.4 trillion, has overtaken Samsung in overall DRAM market revenue for the first time, capturing 36% global market share compared to Samsung’s 34%, according to Counterpoint Research.

SK Hynix’s success has been particularly pronounced in the high-bandwidth memory segment, which is crucial for AI server applications.

“Samsung has assumed that the uncertainties are diminished, it expects its performance to improve in the second half of the year,” the company noted, striking a cautiously optimistic tone despite the challenges ahead.

Samsung’s record revenue masks a pivotal crossroads for the tech giant: while its AI-enhanced smartphones flourish, its once-dominant semiconductor business risks falling behind in the AI revolution.

The coming quarters will reveal whether Samsung’s massive R&D investments can reclaim lost ground in HBM chips, or if we’re witnessing a fundamental power shift in Asian tech manufacturing that could alter the global AI supply chain for years to come.

For a company that rebuilt itself numerous times over its 56-year history, the AI semiconductor race may prove to be its most consequential transformation yet.

(Image credit: Anthropic)

See also: Baidu ERNIE X1 and 4.5 Turbo boast high performance at low cost

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with other leading events including Intelligent Automation Conference, BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.

Explore other upcoming enterprise technology events and webinars powered by TechForge here.

[groovy-computers]

🚀 Breaking News: Samsung struggles to meet NVIDIA's HBM3E standards, paving way for Micron's ascension! 📈 Samsung's bid to dominate the HBM3E sector hits turbulence as it fails NVIDIA's qualification tests, triggering a shift in key clients, including Google, towards competitors like Micron. Samsung's late entry into the HBM market, where giants like SK Hynix and Micron already thrive, puts its market position at risk. Despite hopes for a foothold with NVIDIA, export restrictions further complicate Samsung's plans. This setback underscores the fierce competition within the tech supply chain industry. As companies vie for partnerships with industry leaders, meeting rigorous standards becomes critical for survival. Will Samsung bounce back with innovative solutions, or will Micron's momentum overshadow its attempts? Your thoughts? 🤔 #Samsung #HBM3E #Micron #NVIDIA #TechNews #Innovation #Semiconductors #AI #TechTrends

[qubixo1]

South Korea announced more than $ 23 billion of chip like Trump's trees on semiconductoringtor Inters

Visitors look at the Sk Hynix Inc. 12-Lunch Hbm3e Transnition SEMONIL SELECTION (SEEX Korea, Wednesday, Oct. 23, 2024. Bloomberg | Bloomberg | The Getty photos South Korea announced on Tuesday packages 33 trillion won ($ 23.25 billion) due to its semiconDuc, due to the uncertainty of home companies. This comes after the US President Donald Trump he says he said Becomes advertising the amount of…

[newtras]

South Korea announces over 23 billion dollars for the chip sector, because Trump's tariffs for semiconductor imports are circulating

Visitors look at the SK Hynix Inc. show. 12-layer HBM3E memory chips at a semiconductor exhibition (Sedex) in Seoul, South Korea on Wednesday, October 23, 2024. Bloomberg Bloomberg Getty images On Tuesday, South Korea announced a support package of 33 trillion wins ($ 23.25 billion) for its important semiconductor industry, because increased uncertainty about American tariffs threatens domestic…

[satrthere]

South Korea announces over 23 billion dollars for the chip sector, because Trump's tariffs for semiconductor imports are circulating

Visitors look at the SK Hynix Inc. show. 12-layer HBM3E memory chips at a semiconductor exhibition (Sedex) in Seoul, South Korea on Wednesday, October 23, 2024. Bloomberg Bloomberg Getty images On Tuesday, South Korea announced a support package of 33 trillion wins ($ 23.25 billion) for its important semiconductor industry, because increased uncertainty about American tariffs threatens domestic…

[10bmnews]

South Korea announces over $23 billion for chip sector as Trump tariffs on semiconductor imports loom

Visitors look at the display of SK Hynix Inc. 12-layer HBM3E memory chips at the Semiconductor Exhibition (SEDEX) in Seoul, South Korea, on Wednesday, Oct. 23, 2024. Bloomberg | Bloomberg | Getty Images South Korea announced Tuesday a support package of 33 trillion won ($23.25 billion) for its vital semiconductor industry, as heightened uncertainty over U.S. tariffs threatens domestic…

[rawstorys]

South Korea announces $ 23 billion for the chip region as a rate of Trump on importing semiconductor

On Wednesday, October 23, 2024, SK Hinics Inc. at the semiconductor display (Sedex) in South Korea’s Soul. Visitors to the exhibition of 12-layer HBM3E memory chips. Bloomberg | Bloomberg | Getty South Korea on Tuesday announced a support package of TR 33 trillion von ($ 1.2.5 billion) for its important semiconductor industry, as domestic companies are in danger due to uncertainty at US…