What’s in store for PCs & laptops

In the dynamic landscape of the technology industry, the year 2023 witnessed remarkable hardware innovations, particularly in the PC sector. The advancements were intriguing and innovative, from AI-powered gaming characters to solid-state cooling.

In the ever-evolving landscape of technology, 2023 proved to be a pivotal year marked by unprecedented hardware innovations and paradigm-shifting advancements in the PC industry. We have witnessed the emergence of groundbreaking trends and transformative technologies. From the integration of AI-powered gaming characters to the unveiling of cutting-edge processors by industry giants like Intel and AMD, the pace of innovation has been nothing short of remarkable. Today we will delve into the intricate details of the hardware landscape in 2023, exploring the nuances of semiconductor breakthroughs, revolutionary PC designs, and the identification of key technology trends. As we navigate through the intricate web of technological progress, the anticipation of future developments in CPU technology adds an extra layer of excitement to the ever-expanding realm of possibilities.

Computing Hardware Innovations

The processor

Intel has unveiled its highly anticipated 4nm Meteor Lake and Arrow Lake processors, promising unprecedented levels of performance and efficiency. The recent announcement at the keynote event not only introduced these groundbreaking processors but also shed light on Intel's ambitious roadmap, showcasing a commitment to pushing the boundaries of CPU capabilities. In this article, we will delve into the key features of the Meteor Lake and Arrow Lake processors, the advancements in AI-focused Core Ultra 'Meteor Lake' CPU, and Intel's future roadmap, including the next-gen Lunar Lake Chip and beyond.

AI-Focused Core Ultra 'Meteor Lake' CPU: A Leap into Power Efficiency (December 2023)

Intel's journey into innovation begins with the release of the Meteor Lake CPUs, set to debut on December 14 under the new Intel Core Ultra brand. These processors mark a significant step forward with the integration of a neural processing unit (NPU), designed to facilitate power-efficient AI acceleration and local inference on PCs. What sets these CPUs apart is their ability to strike a delicate balance between performance and power, achieved through the utilization of Intel Arc GPU technology for discrete-level graphics performance. As the industry witnesses an increasing demand for AI-driven tasks, the Core Ultra processors leverage CPU, GPU, and NPU functionalities, providing a comprehensive solution for diverse computing needs.

Next-Gen Lunar Lake Chip and Beyond (2024-2025)

In a surprising turn of events, Intel's CEO Pat Gelsinger unveiled the Lunar Lake system, a next-generation client CPU anticipated to arrive in 2024. This revelation serves as a testament to Intel's continuous commitment to technological advancement. Following Meteor Lake, the roadmap unfolds with the Arrow Lake processor succeeding in 2024, paving the way for Lunar Lake's arrival in 2024 or 2025. Looking even further into the future, Panther Lake is set to make its debut in 2025, highlighting Intel's strategic vision and dedication to maintaining a leadership position in performance per watt.

source Intel

5th-Gen Intel Xeon Scalable Processors

As Intel propels into the future, it doesn't overlook the server domain. The fifth-generation Xeon Scalable server processors are scheduled to launch on December 14, representing a significant milestone in server technology. Looking ahead, a future generation, slated in the next year, will introduce a 288-core CPU based on the Emerald Rapids architecture. Additionally, Sierra Forest processors, equipped with efficiency cores, are expected to enhance rack density and performance per watt in the first half of 2024. Following closely is Granite Rapids, a successor to Sierra Forest, anticipated in 2024, promising improved AI performance and solidifying Intel's position in the server market.

Source: Intel

AMD has unveiled its Ryzen 7000 series processors, showcasing the cutting-edge Zen 4 architecture for unparalleled computing power. In addition to delivering significant improvements in both single-threaded and multi-threaded performance, AMD's annual Corporate Responsibility Report reinforces the company's commitment to environmental sustainability, diversity, and inclusion.

Ryzen 7000 Series Processors

AMD's Ryzen 7000 series processors, driven by the Zen 4 architecture, have set a new standard for embedded computing. The processors, announced at Smart Production Solutions 2023, bring together the prowess of "Zen 4" architecture and integrated Radeon graphics, offering unmatched performance and functionality tailored for industrial markets. The utilization of next-generation 5nm technology underscores AMD's dedication, with a commendable 7-year manufacturing availability commitment, marking a milestone in the embedded processor landscape.

Source: AMD

Key Features:

5nm Technology: The Ryzen Embedded 7000 Series is the first to leverage next-generation 5nm technology, ensuring efficiency and performance for a wide range of industrial applications.

Integrated Radeon RDNA 2 Graphics: Eliminating the need for a discrete GPU, the processors integrate AMD Radeon RDNA 2 graphics, enhancing visual processing for industrial applications such as robotics, video surveillance, and more.

Operating System Support: Recognizing the diverse needs of embedded applications, AMD ensures compatibility by supporting Windows Server, Linux Ubuntu, Windows 10, and Windows 11, providing flexibility for system designers.

Scalable Design: With up to 12 high-performance "Zen 4" CPU cores, the processors offer a scalable design, catering to the increasing complexity of industrial applications.

Prioritizing energy efficiency, AMD aims for a 30x improvement in processors and accelerators for AI-training and HPC by 2025. With a 13.5x improvement from the 2020 base year, AMD is on track to meet this ambitious goal.

AMD is actively reducing greenhouse gas emissions, achieving a 19% reduction in Scope 1 and 2 emissions in 2022 compared to 2020. Recognized as a CDP Supplier Engagement Leader, AMD collaborates with suppliers and customers to address climate change.

AMD's commitment to human rights is evident in its Human Rights Saliency Assessment and focus on diversity and inclusion. Retention of female engineers is strong, and 92% of global employees attest to AMD creating an environment for diverse backgrounds to succeed.

Fujitsu has unveiled groundbreaking technologies to tackle GPU shortages and elevate High-Performance Computing (HPC). These innovations focus on real-time optimization of CPU and GPU usage and dynamic switching between multiple programs in HPC systems.

Figure 1. Image of CPU and GPU allocation switching

Figure 2. Differences in communication methods used to switch program execution

Source: Fujitsu

Optimizing CPU and GPU Usage:

Fujitsu introduces the world's first real-time optimization technology for CPUs and GPUs. This breakthrough addresses the GPU shortage by prioritizing processes with high execution efficiency. The system dynamically allocates resources, distinguishing between CPU and GPU requirements even during program processing.

Key Features:

• Program Prioritization: Identifies programs suitable for CPU/GPU processing.
• Real-Time Allocation Switching: Dynamically allocates GPUs based on performance metrics.
• Applications: Enables quick model training for graph AI data, benefiting AI development and image recognition.

Real-Time Switching in HPC Systems:

Fujitsu achieves real-time switching between programs in HPC systems, eliminating the need to wait for the completion of running programs. This breakthrough caters to applications demanding real-time performance and extensive computational resources.

Key Features:

• Broadcast Communication: Enables real-time batch switching in HPC systems.
• Optimal Communication Selection: Balances performance improvement and potential degradation due to packet loss.

Future Applications:

Fujitsu plans to integrate these technologies into a future computer workload broker, empowering AI to autonomously select resources based on user requirements. The CPU/GPU optimization technology will be applied to Fujitsu Kozuchi AI Platform, while the HPC optimization technology will benefit quantum computing and Computing as a Service HPC offerings.

Fujitsu's demonstrations at SC23 underscore its commitment to driving innovation, solving societal challenges, and contributing to a sustainable future through cutting-edge computing solutions.

The Storage

The adoption of 3D NAND technology has ushered in a new era of data storage, allowing for increased memory densities, heightened reliability, and extended endurance. This has been made possible by the utilization of Multi-Level Cell (MLC) and Triple-Level Cell (TLC) technologies, enabling higher-capacity SSDs while maintaining data integrity and performance.

The integration of Non-Volatile Memory Express (NVMe) has been a game-changer, overcoming performance limitations associated with older interfaces. The PCIe Gen4 interface further accelerates data transfer speeds, providing robust reliability for enterprise workloads and ensuring responsiveness in demanding applications.

Impact of Artificial Intelligence on SSD Technology

AI has left an indelible mark on SSD technology, with AI-powered storage solutions leveraging machine learning algorithms. These algorithms analyze workload patterns in real-time, optimizing performance and reducing latency issues through predictive analytics. The result is an SSD ecosystem that adapts dynamically to application requirements.

Emergence and Progression of QLC Technology

Quad-Level Cell (QLC) technology represents a significant stride in achieving larger SSD capacities at economical prices. By storing four bits per memory cell, QLC-based SSDs cater to the demands of modern applications without compromising on performance or endurance, making them a viable solution for diverse computing needs.

Reliable Enterprise-Grade Storage Solutions with TLC

Triple-Level Cell (TLC) technology brings cost-efficiency to enterprise-grade SSDs, offering benefits comparable to MLC counterparts but at lower costs. These SSDs, with a focus on server-centric applications, align with the requirements of data-driven enterprises, providing a balance between performance and budget considerations.

SSD Backup Solutions: Protecting Data Integrity

The reliability of SSD technology is a key factor in reducing failure rates compared to traditional HDD backup systems. The absence of mechanical parts enhances data redundancy methods, safeguarding critical information and minimizing the risk of loss or corruption, thereby ensuring the integrity of stored data.

Future Innovations: More Than Fast Loading Times

Looking ahead, the future of SSDs extends beyond speed improvements, with quantum computing presenting exciting possibilities for significant advancements in storage capacities and data processing speeds. Developments like NRAM (Non-Volatile Random-Access Memory) and OptiNAND from Western Digital showcase the industry's commitment to higher-performing and more efficient storage solutions.

Advanced HDD Innovations

In addition to SSDs, the hard disk drive (HDD) landscape has witnessed remarkable innovations aimed at enhancing capacities, performance, and reliability.

OptiNAND (Western Digital)

• Combines HDDs with embedded flash drives for faster, smarter hard drives with higher densities.
• Reduces adjacent track interference, minimizing latency and improving performance.
• Stores essential metadata in the flash drive for quick indexing and efficient data management.
• Utilizes write-cache function to prevent data loss in case of power loss.

UltraSMR (Western Digital)

Source: WD

• Extends the capacity advantage of shingled magnetic recording (SMR) over conventional magnetic recording.
• Integrates hardware, software, and firmware technologies for higher-capacity HDDs.
• Increases areal density and storage capacity, enhancing error correction capabilities and reducing seek-induced vibration.

Triple-Stage Actuator (Western Digital)

• Employs triple-stage actuators for precise head positioning, improved performance, and higher data density.
• Includes Voice Coil Motor (VCM) actuator, milli-actuator, and micro-actuator for high-precision positioning.
• Reduces read/write head access time and seek-induced vibration.

Helio-Seal (Hitachi)

• Involves filling hard drives with helium to reduce resistance to spinning disks and heat generation.
• Increases drive capacity by moving platters closer, improving reliability, longevity, reducing noise, and lowering power consumption.

HAMR (Seagate)

Source: Seagate

• Heat-assisted magnetic recording (HAMR) increases HDD storage capacity using a nanoscopic laser diode.
• Overcomes limitations of perpendicular magnetic recording (PMR) by creating stable bits at room temperature.
• Allows data densities ranging between 2 and 5 terabits per square inch (Tbpsi), significantly higher than conventional drives.

MACH.2 (Seagate)

Source: Seagate

• Seagate’s multi-actuator technology that improves drive read/write performance.
• Two independent actuators access different parts of the drive simultaneously, doubling IOPS performance.
• Uses 40% less power than two single-actuator drives and achieves a sustained data throughput of 480 MB/sec.

MAS-MAMR (Toshiba)

• Microwave-assisted magnetic recording (MAS-MAMR) increases HDD storage capacity without sacrificing performance or reliability.
• Relies on short bursts of highly targeted microwave energy to change the magnetic orientation of bits on the drive.
• Aims to build HDDs exceeding 30 terabytes with stable performance.

NVMe HDDs

• Introduction of NVMe HDDs using the non-volatile memory express (NVMe) protocol instead of SATA or SAS.
• Provides higher bandwidth and data throughput, maximizing performance potential.
• Supports multi-actuator technologies for drastic performance improvement, leading to cost reductions and energy savings.

Innovations in DRAM Technology

As the demand for higher-capacity and faster memory solutions intensifies, DRAM technology continues to evolve, pushing the boundaries of what's possible.

Samsung's pursuit of new 3D structures aims to enable larger single-chip capacities exceeding 100 gigabits (Gb). The development of the next-generation 11nm-class DRAM promises the industry's highest density. Breakthroughs in NAND flash aim for 1,000-layer vertical NAND (V-NAND), with the ninth-generation V-NAND utilizing a double-stack structure.

Samsung introduces the next-generation HBM3E DRAM, named Shinebolt, designed for AI applications. Shinebolt offers a speed of 9.8 gigabits per second per pin, resulting in transfer rates exceeding 1.2 terabytes-per-second. Non-conductive film (NCF) technology is optimized to enable higher-layer stacks and improve thermal characteristics. Mass production of Samsung’s 8H and 12H HBM3 products is ongoing, with samples for Shinebolt already shipping.

Samsung introduces memory solutions supporting high-performance, high-capacity, low-power, and small form factors for edge devices. The 7.5Gbps LPDDR5X CAMM21 is highlighted as a game-changer for next-generation PC and laptop DRAM markets. Other solutions include 9.6Gbps LPDDR5X DRAM, LLW2 DRAM for on-device AI, next-gen Universal Flash Storage (UFS), and high-capacity Quad-Level Cell (QLC) SSD BM9C1 for PCs.

The GPU Technology

The landscape of GPU technology is undergoing a dynamic transformation, marked by significant advancements and promising trends that are reshaping the capabilities of graphics cards. In this comprehensive exploration, we delve into the recent developments in GPU memory, the releases of 2023, and the exciting prospects that await in 2024 and beyond.

GPU Memory Advancements

One of the key highlights in GPU technology is the imminent launch of Micron's GDDR7 memory in early 2024. This next-generation memory is set to revolutionize graphics cards, providing the ability to store and display detailed images and textures at resolutions beyond the current standards of 1440p and 4K. The introduction of GDDR7 is poised to elevate the performance of upcoming graphics cards, offering a glimpse into the future of immersive visual experiences.

High-Bandwidth Memory (HBM) Technology Gain

The year 2023 witnessed a surge in the adoption of High-Bandwidth Memory (HBM) technology in GPUs. HBM, with its significantly higher memory bandwidth compared to traditional GDDR memory, became a catalyst for handling even the most demanding workloads. This transition marked a pivotal moment in GPU memory technology, setting the stage for further innovations.

2024 and Beyond

As we look ahead to 2024 and beyond, several trends and advancements are expected to shape the future of GPU technology.

RT and AI Technologies Development: Continued evolution of Ray Tracing (RT) and AI technologies is anticipated, with more advanced algorithms and robust hardware support.

Widespread Adoption of HBM Technology: HBM technology is projected to become more widespread, potentially replacing traditional GDDR memory in high-end GPU models.

Emergence of New Memory Technologies - CXL Memory: New memory technologies, such as Compute Express Link (CXL) memory, might emerge, offering even higher bandwidth and lower latency than HBM.

Introduction of New GPU Architectures: Both Nvidia and AMD are expected to unveil new GPU architectures, promising enhanced performance and efficiency improvements.

Nvidia is expected to release its next-generation GPU architecture, codenamed "Blackwell," in late 2024 or early 2025. Blackwell is expected to be based on a new manufacturing process and offer significant performance and efficiency improvements over the current Ada Lovelace architecture.

Here are some of the key features that are expected to be included in Blackwell:

New CUDA cores: Blackwell is expected to feature a new generation of CUDA cores, which are the basic processing units of Nvidia GPUs. These new cores are expected to be more powerful and efficient than the current CUDA cores.

New memory architecture: Blackwell is expected to feature a new memory architecture that is designed to improve bandwidth and reduce latency. This could include the use of HBM3 memory or a new type of memory altogether.

Improved AI acceleration: Blackwell is expected to feature improved AI acceleration capabilities, enabling more powerful and efficient AI workloads. This could include the use of new AI accelerators or improvements to existing AI hardware.

AMD's Next-Generation GPU Architecture

AMD is expected to release its next-generation GPU architecture, codenamed "RDNA4," in late 2025 or early 2026. RDNA4 is expected to be based on a new manufacturing process and offer significant performance and efficiency improvements over the current RDNA3 architecture.

Here are some of the key features that are expected to be included in RDNA4:

New compute units: RDNA4 is expected to feature a new generation of compute units, which are the basic processing units of AMD GPUs. These new units are expected to be more powerful and efficient than the current compute units.

New memory architecture: RDNA4 is expected to feature a new memory architecture that is designed to improve bandwidth and reduce latency. This could include the use of HBM3 memory or a new type of memory altogether.

Improved ray tracing: RDNA4 is expected to feature improved ray tracing capabilities, enabling more realistic and efficient ray tracing. This could include the use of new hardware accelerators or improvements to existing ray tracing hardware.

These are just some of the expected features of Nvidia's and AMD's next-generation GPU architectures. But Intel has also entered the arena with its latest GPUs.

Intel has been actively developing its GPU technology in recent years, with a focus on delivering high-performance graphics for gaming, content creation, and machine learning. In 2023, Intel released its Arc Alchemist GPU series, which marked the company's first foray into the discrete GPU market.

Key Innovations in Intel GPU Technology

Xe Graphics Architecture: Intel's Xe graphics architecture is the foundation for its Arc GPUs. This architecture is designed to deliver high performance, efficiency, and scalability for a wide range of applications.

Xe HPG Microarchitecture: The Xe HPG microarchitecture is a specialized version of the Xe architecture that is tailored for high-performance gaming and professional applications. It features enhanced shader units, improved memory bandwidth, and support for advanced graphics features like ray tracing and variable rate shading.

XeSS (Xe Super Sampling):XeSS is Intel's proprietary technology for upscaling images and resolution. It utilizes AI to enhance the quality of lower-resolution images, making them comparable to higher-resolution ones.

Source: Intel

OneAPI Programming Model: Intel's OneAPI programming model provides a unified development environment for creating applications that can run on both Intel CPUs and GPUs. This simplifies the development process and allows developers to take advantage of the full potential of Intel's hardware.

Future Directions for Intel GPU Technology

Intel is continuously investing in GPU technology and has set ambitious goals for future advancements. Some of the key areas of focus include:

• Enhanced Performance: Intel aims to deliver even higher performance with its future GPU architectures, targeting enthusiast-level gaming and professional workloads.
• Improved Efficiency: Intel is committed to improving the energy efficiency of its GPUs to reduce power consumption and environmental impact.
• Expanded Functionality: Intel is exploring new applications for its GPUs beyond gaming and graphics, such as Machine Learning, AI, and data analytics.
• Enhanced Software Ecosystem: Intel is working to expand its software ecosystem and provide developers with the tools and support they need to create innovative applications for Intel GPUs.

Intel's GPU technology innovations have the potential to significantly impact the PC gaming and graphics landscape. As the company continues to develop and refine its GPU offerings, we can expect to see even more powerful, efficient, and versatile GPUs emerge in the years to come.