Jason Hiner (MacRumors):

In an exclusive conversation with The Deep View ahead of WWDC 2026 in June, Doug Brooks, senior product manager of Apple silicon, discussed how Apple’s long-term investments in neural processing, unified memory, power-efficient computing, and hardware-software integration have positioned its devices for the AI era.

He also explained why Macs have become a favorite platform among AI developers, why Mac minis and Mac Studios are emerging as the preferred AI agent machines, and how Apple sees the future of on-device AI across iPhone, iPad, and Mac.

[…]

“We build a chip knowing exactly what systems it’s going to go into. The chip can influence the system design, and the system design can influence the chip requirements. They’re married together very tightly. I’d actually go a step further and include software in that equation. As we enable new capabilities in the chip, we want to make sure those capabilities are exposed all the way through the software stack so developers can benefit and customers can benefit. Because if those transistors go unused, what’s the point? It’s not about adding something and hoping somebody finds a use for it. We want those capabilities available to solve real problems. I think that’s what allows us to build better systems and ultimately better products.”

It seems like a lot of it is just that developers like macOS.

Previously:

• OpenClaw (Formerly Moltbot)
• Mac Studio 2025
• Mac mini 2024
• Why Apple Believes It’s an AI Leader

Update (2026-07-10): See also: Hacker News.

Howard Oakley:

Hot Macs have their own paradox: open Activity Monitor, select the CPU view, and at the top of the CPU % list will be kernel_task hogging the CPU cores with 100% or more, rather than its usual 4% or so. Kerb any temptation to kill it and hope it goes away, as it has taken control of your Mac to let it cool.

[…]

There are rare occasions when fans blow full on and kernel_task goes wild without any thermal problem. In Intel Macs, resetting the SMC is usually curative, but this could instead be the result of a fault in a thermal sensor, or in the SMC. Hardware diagnostics should tell you more. By far the most common cause of persistent problems is dust and debris in the air ducts, and in the case of some Intel Mac notebooks overheating of the left USB-C ports. For more details, see Apple’s note about kernel_task, which also confirms what I have written here.

[…]

If your Mac is showing early signs of thermal strain and still running, encourage heat loss by immediately[…]

Paul Haddad:

It boggles my mind that the biggest software company in the world, with total control of their hardware/software stack is totally incapable of shipping an OS that doesn’t include one or more daemons who just endlessly waste CPU cycles.

Bonus irony points because dasd’s purpose is supposedly to schedule low priority background processes.

Howard Oakley:

Continuing with the results from file compression, it’s straightforward to calculate the total energy required to compress the test file using the P and E cores:

• When compressing using the ten P cores, power used was 51 W for a total of about 8.4 seconds, thus 428 J, or 43 J per core.
• When compressing using the four E cores, power used was 255 mW for a total of about 150 seconds, thus 38 J, or 10 J per core.

[…]

In omitting GPU power from Energy Impact, many power- and energy-intensive tasks will appear far less demanding than they really are. This is becoming even more important now Apple silicon chips are being used increasingly to run AI computation locally.

Neither powermetrics nor Activity Monitor provide any estimates of power or energy used by other parts of a Mac, such as memory or the internal SSD, although those can also be important determinants of battery endurance and heat generation.

Previously:

• “If You Can’t Stand By a Feature, You Shouldn’t Launch It.”
• The LG UltraFine 5K, kernel_task, and Me
• Throttling Due to Thunderbolt Left Proximity Sensor

Update (2026-07-02): Howard Oakley:

This article explains how you can improve your Mac’s cooling to cope better with extreme summer temperatures.

A groundbreaking advancement in photonic technology promises to redefine the landscape of medical artificial intelligence (AI) diagnostics, as a research consortium led by Professor Han Zhang at Shenzhen University has unveiled a novel all-fiber photonic AI platform utilizing black phosphorus (BP)-based tunable modulators. This innovative approach, which hinges on photon-based computing, surmounts the intrinsic constraints posed by traditional electronic processors, delivering unprecedented energy efficiency and ultrafast performance crucial for real-time clinical decision-making.

The prevailing paradigm in medical AI, deeply reliant on electronic architectures such as GPUs, faces severe limitations related to energy consumption, heat dissipation, and processing latency. These factors collectively impede the integration of AI into time-sensitive diagnostic workflows and exacerbate systemic carbon footprints. Photonic computing, with its reliance on photons rather than electrons, offers dramatically accelerated data transmission speeds and parallelism through wavelength multiplexing, alongside near-zero heat generation. Until now, however, the practical implementation of photonic AI had been hampered by the bulkiness, inefficiency, and fabrication challenges synonymous with conventional optical modulators.

Professor Zhang’s team has ingeniously engineered an atomically thin van der Waals heterostructure by stacking black phosphorus and molybdenum disulfide (MoS2), leveraging the extraordinary optoelectronic properties of these 2D materials. This heterostructure is meticulously integrated onto a microfiber knot resonator (MKR), an ultra-fine looped optical fiber with a diameter narrower than a human hair. This configuration significantly intensifies the interaction between light and material, enabling modulation of the refractive index with minimal applied voltage, which shifts the light’s wavelength to encode computational information with remarkable precision and energy economy.

The fabrication process commences with mechanical exfoliation to obtain pristine BP/MoS2 bilayers, followed by a dry transfer technique to position these heterostructures onto the MKR. The MKR’s torsional geometry enhances evanescent field coupling, ensuring that the optical mode overlaps maximally with the active layered materials. To broaden the modulator’s linear response and suppress nonlinear distortions, the researchers incorporated a Ring-Assisted Mach–Zehnder Interferometer (RAMZI) design. This integration extends the operational voltage range, ensuring fidelity and robustness essential for medical-grade AI computations.

The team transcended single-device innovation by architecting a full all-fiber photonic neural network (PNN). By deploying two RAMZI modulators in tandem with a highly sensitive photoreceiver within a sophisticated time-division multiplexing schema, they realized a closed-loop, coherent computing unit capable of executing neural network operations entirely in the optical domain. This milestone substantiates the transition of photonic neural networks from theoretical constructs to viable, deployable platforms for clinical diagnostics.

Clinical performance validation was conducted in collaboration with esteemed hospitals. The PNN demonstrated expert-level diagnostic accuracy in two stringent medical tasks: detecting retinal detachment from B-scan ultrasound imagery and diagnosing hepatocellular carcinoma (HCC) using multiphase liver computed tomography (CT) scans. Diagnostic accuracy and specificity metrics paralleled those of seasoned radiologists, underscoring the system’s potential for integration into frontline healthcare.

Beyond accuracy, the photonic system showcased transformative efficiency advantages. While a conventional NVIDIA A100 GPU requires approximately 85 milliseconds to process a liver CT study, the photonic platform accomplished the same in a mere 0.8 milliseconds. Additionally, the energy expenditure per operation plummeted to an astonishing 0.608 femtojoules, a staggering 246-fold decrease compared to electronic counterparts consuming 150 femtojoules. This synergy of speed and efficiency paves the way for deploying AI diagnostics in resource-constrained or mobile medical environments like rural clinics and ambulances, dramatically expanding access to critical health services.

The system’s environmental implications cannot be overstated. Conventional AI’s sizable energy footprint contributes appreciably to healthcare’s carbon emissions. The BP/MoS2-based photonic platform fundamentally disrupts this model by combining ultrafast processing with near-negligible thermal dissipation, forging a sustainable pathway for green AI implementations in medicine. This breakthrough aligns with expanding global commitments to environmentally responsible technologies without sacrificing computational power.

Acknowledging the infancy of current implementations, the authors outline avenues for scaling. Presently limited to a single processing layer realized through two modulators, the system’s complexity and image resolution capacity necessitate expansion. A promising strategy involves wavelength-division multiplexing (WDM), exploiting the device’s ∼30 nm bandwidth resonances to channel up to 40 distinct wavelengths. This parallelism would amplify computational density approximately 40-fold while maintaining the existing clock rate, facilitating thousands of multiplications per network layer—an enhancement critical to tackling high-dimensional medical image data.

Device longevity and stability pose additional challenges, given the sensitivity of BP to environmental degradation. While the MoS2 layer affords interim encapsulation, the team plans industrial collaboration to pioneer large-scale protective coatings via atomic layer deposition of Al2O3 and scalable chemical vapor deposition processes. These advancements will bolster device durability, uniformity, and manufacturability, accelerating clinical adoption and commercial viability.

The study also signifies the maturation of fiber-based photonic neural networks into practical diagnostic platforms rather than experimental novelties. Optical fibers’ inherently low transmission loss (<0.2 dB/km), synergized with the modulator’s efficient 0.25 V·cm electro-optic response and expanded linear range, addresses critical bottlenecks impeding prior photonic computing endeavors. Consequently, the technology promises to empower future applications in drug discovery, genomics, and real-time clinical imaging with unmatched energy footprints.

The pioneering work led by Professor Han Zhang not only foregrounds interdisciplinary collaboration across physics, materials science, and biomedicine but also sets a precedent for industry-academia partnerships catalyzing the commercialization of ultrafast photonic devices poised to transform healthcare AI. As research progresses towards integrated multi-layer architectures and industrial encapsulation, the healthcare sector stands on the cusp of embracing a new generation of diagnostic tools that are faster, greener, and broadly accessible.

This heralds a paradigm shift where medical AI transcends the limitations inherent to electronic systems, embracing photon-driven computing as a sustainable, high-performance alternative that aligns with the twin imperatives of clinical excellence and environmental stewardship.

Subject of Research: People

Article Title: Tunable phosphorene modulators – accelerating medical diagnosis with ultra-efficient photonic platforms

News Publication Date: 28-May-2026

References:
Zhang, H., Liu, Y., Wang, H., Zhu, H., et al. (2026). Tunable phosphorene modulators – accelerating medical diagnosis with ultra-efficient photonic platforms. Opto-Electronic Advances. DOI: 10.29026/oea.2026.250332

Image Credits: Professor Han Zhang of Shenzhen University

Keywords: photonic computing, black phosphorus, molybdenum disulfide, van der Waals heterostructure, microfiber knot resonator, electro-optic modulator, photonic neural network, medical imaging AI, energy efficiency, wavelength-division multiplexing, ultrafast diagnostics, sustainable healthcare

In the rapidly evolving realm of artificial intelligence and deep learning, conventional electronic computing architectures are increasingly strained by the demands of large-scale data processing and intricate computational tasks. Optical computing has risen as a compelling alternative, leveraging the inherent physical phenomena of light—such as interference and diffraction—to execute computations fundamentally differently from traditional electronics. By harnessing these optical effects, optical computing systems (OCS) promise unparalleled advantages: notably superior processing speeds, heightened energy efficiency, and a natural aptitude for massive parallelism. These attributes make optical computing especially attractive for intensive applications like image processing, machine learning, and big data analytics, holding transformative potential across these domains.

Yet, despite these promising characteristics, contemporary optical computing platforms grapple with a significant bottleneck rooted in their hardware-centric development workflows. Conventionally, researchers must directly engage with physical OCS hardware to develop, tune, and validate computational tasks. This process entails sequential access to a limited number of costly and often heavily utilized devices. Users must physically occupy the system to load parameters, fine-tune configurations, and iteratively calibrate the hardware—an intrinsically tedious and time-consuming cycle. Each new user or task demands resetting the system state, leading to protracted queues, repetitive adjustments, and redundant calibrations. The resulting workflows impose long hours of equipment occupation and elevated trial-and-error expenses, stifling the possibility for concurrent task development and drastically curtailing research throughput and system agility.

To surmount these entrenched challenges, cutting-edge research has introduced the concept of the Digital Twin Optical Computing System (DT-OCS). This innovative approach establishes a high-fidelity digital replica—a “digital twin”—of the physical OCS that accurately models its input-output behavior under varying configuration parameters. Crucially, the digital twin allows researchers to perform extensive offline simulations, training, and optimization without the constraints of physical hardware availability. Analogous to a digital simulator software that mirrors a complex “real machine,” DT-OCS enables extensive task development and parameter tuning within a virtual environment, dramatically decoupling software development from hardware dependency. Consequently, optimized parameters validated digitally can be seamlessly transferred back to the physical system for deployment, bypassing numerous iterative hardware trials.

The implications of DT-OCS extend well beyond mere time savings. By effectively creating a “digital development toolkit” for optical computing, this framework revolutionizes the research ecosystem—transforming optical computing platforms from isolated, hardware-bound experiments to versatile, shareable research infrastructures. Researchers can now conduct task training, performance benchmarking, and method comparisons entirely within a unified digital realm, fostering collaborative innovation. Distinguished from legacy OCS paradigms, DT-OCS supports parallel task development workflows, allowing multiple computational problems to be advanced simultaneously without hardware conflicts. This capability markedly enhances flexibility, reduces development bottlenecks, and accelerates the maturation and practical adoption of optical computing technologies.

Envisioning the trajectory of optical computing, the integration of digital twin models with physical hardware appears indispensable. Contemporary transportation systems, for instance, benefit from the synergy between tangible road networks and dynamic digital mapping platforms. Similarly, next-generation optical computing platforms should embody dual representations: a robust physical computing infrastructure supplemented by perpetually updated digital twins. This paradigm champions accessibility, reproducibility, and scalability—cornerstones for fostering broad-scale collaboration. It paves the way for unified validation standards and equitable performance comparisons, catalyzing optical computing’s evolution into a general-purpose, widely deployable resource rather than a niche experimental setup.

The DT-OCS framework’s hallmark is its strategic decoupling of task development from hardware constraints. In traditional optical computing, task training and parameter fine-tuning necessitate iterative hardware engagements involving adjustments, diagnostics, and recalibrations. This approach encumbers the process with inefficiencies and precludes simultaneous multi-task development. DT-OCS obviates these impediments by faithfully modeling the physical system’s responses across parameter spectra, enabling exhaustive offline optimization. This digital-first workflow increases research agility, minimizes hardware occupancy, and substantially compresses the development timeline. Consequently, DT-OCS amplifies the throughput and adaptability of optical computing research endeavors.

The real-world efficacy of the DT-OCS framework has been experimentally validated using a cutting-edge high-speed optical computing system incorporating silicon photonic feature-computing chips. These empirical studies demonstrated DT-OCS applications in diverse computational domains, including image classification and sequential decision-making. Notably, configurations optimized within the digital twin translated flawlessly to the physical hardware, with observed task performance closely mirroring digital forecasts. This underscores not only the high fidelity of the DT-OCS model but also its robust transferability across practical scenarios. Moreover, by enabling parallel task development within the digital domain, DT-OCS expedites the overall research workflow, markedly elevating efficiency.

An illustrative application of the DT-OCS system involved offline training and deployment for the Fashion-MNIST image classification task—a benchmark dataset widely used in machine learning. This practical demonstration affirms DT-OCS’s capacity to bridge digital model development with real-world optical computing implementations. Beyond efficiency gains, DT-OCS signifies a conceptual shift toward disentangling computational task design from the idiosyncrasies of physical system constraints. Historically, optical computing research has been hamstrung by its dependence on specific hardware platforms and accessibility limitations, hindering reproducibility and cross-task comparisons. The introduction of an open-source digital twin framework surmounts these barriers by democratizing access and standardizing evaluation protocols.

By openly releasing the DT-OCS framework alongside associated task datasets, the research fosters a community-driven ecosystem where software-based task design, training, and validation can proceed independently of physical system constraints. This openness engenders scalability, reproducibility, and collaborative potential that traditional optical computing architectures cannot easily achieve. Researchers worldwide gain the freedom to explore broader task horizons, perform comprehensive benchmarking, and validate innovative approaches within a consistent computational substrate. Such a paradigm shift propels optical computing from isolated experimental configurations into accessible and extensible platforms, expediting technological advancements and practical deployments.

This pioneering work advocates a transformative application paradigm for optical computing platforms: the symbiotic coexistence of physical hardware and computationally equivalent open-source digital twin models. Only by embracing such dual-system architectures can the field transcend limitations imposed by experimental variability and hardware scarcity. The resultant platforms will be shareable, replicable, and adaptable to a diverse array of research challenges. As optical computing moves towards becoming a mainstream computational resource, the principles embodied by DT-OCS may well establish the foundational infrastructure that underpins scalable, collaborative innovation in photonic information processing.

Subject of Research: Not applicable

Article Title: Digital twin optical computing system

News Publication Date: 21-Apr-2026

Web References:

• DOI: 10.29026/oea.2026.250254

References:
DOI: 10.29026/oea.2026.250254

Image Credits: Dr. Tingzhao Fu from the National University of Defense Technology, China, and Dr. Hongwei Chen from Tsinghua University, China

Keywords

Artificial intelligence, Computer science, Photonics, Engineering, Machine learning, Computer modeling, Computer processing, Semiconductors, Optics, Applied optics

A decade ago, AMD added a protection to its high-end CPUs to protect them against cold boot attacks and other types of physical exploits that siphon sensitive data out of the connected memory chips. Short for Transparent Secure Memory Encryption, TSME encrypts the entire contents stored in memory, making the data useless to physical attackers.

Over time, AMD added TSME to lower-end processors, including the consumer version of its Ryzen chips, a CPU that costs less than the Pro version. Over the years, users of these lower-end chips have gotten used to the added security. Recently and without warning or notice, this lower-end line of AMD chips suddenly dropped the protection, and did so in a way that was impossible to detect on Windows machines and required a fair amount of technical work when using Linux.

Now you see it, now you don't

AMD has yet to say why TSME worked on these CPUs, or even to confirm the change. AMD declined to answer questions sent by email other than to say TSME "is a security feature only applied to PRO CPUs as part of AMD PRO Technologies." The statement is the first known time the chipmaker has explicitly made this restriction public.

Read full article

Comments