Author's Latest Posts


Spiking Neural Networks Place Data In Time


Artificial neural networks have found a variety of commercial applications, from facial recognition to recommendation engines. Compute-in-memory accelerators seek to improve the computational efficiency of these networks by helping to overcome the von Neumann bottleneck. But the success of artificial neural networks also highlights their inadequacies. They replicate only a small subset of th... » read more

Challenges For Compute-In-Memory Accelerators


A compute-in-memory (CIM) accelerator does not simply replace conventional logic. It's a lot more complicated than that. Regardless of the memory technology, the accelerator redefines the latency and energy consumption characteristics of the system as a whole. When the accelerator is built from noisy, low-precision computational elements, the situation becomes even more complex. Tzu-Hsian... » read more

Compute-In Memory Accelerators Up-End Network Design Tradeoffs


An explosion in the amount of data, coupled with the negative impact on performance and power for moving that data, is rekindling interest around in-memory processing as an alternative to moving data back and forth between the memory and the processor. Compute-in-memory (CIM) arrays based on either conventional memory elements like DRAM and NAND flash, as well as emerging non-volatile memori... » read more

Scaling Up Compute-In-Memory Accelerators


Researchers are zeroing in on new architectures to boost performance by limiting the movement of data in a device, but this is proving to be much harder than it appears. The argument for memory-based computation is familiar by now. Many important computational workloads involve repetitive operations on large datasets. Moving data from memory to the processing unit and back — the so-called ... » read more

Moving To GAA FETs


How do you measure the size of a transistor? Is it the gate length, or the distance between the source and drain contacts? For planar transistors, the two values are approximately the same. The gate, plus a dielectric spacer, fits between the source and drain contacts. The contact pitch, limited by the smallest features that the lithography process can print, determines how many transistors ... » read more

New Packaging Roadmap


Historically, the electronics industry has drawn sharp distinctions between the integrated circuit chip, the package that protects it from the environment, and the board that connects it to other devices in a complete system. The circuit and systems worlds have been largely isolated from each other, using different tools, different processes, and different metrics for success. While integrated ... » read more

New Trends In Wafer Bonding


Unable to scale horizontally, due to a combination of lithography delays and power constraints, manufacturers are stacking devices vertically. This has become essential as the proliferation of mobile devices drives demand for smaller circuit footprints, but the transition isn't always straightforward. Three-dimensional integration schemes take many forms, depending on the required interconne... » read more

Building An MRAM Array


MRAM is gaining traction in a variety of designs as a middle-level type of memory, but there are reasons why it took so long to bring this memory to market. A typical magnetoresistive RAM architecture is based on CoFeB magnetic layers, with an MgO tunneling barrier. The reference layer should have zero net magnetization to make sure that it doesn’t influence the orientation of the free lay... » read more

Magnetic Memories Reach For Center Stage


Wearable heart rate sensors. Networked smoke detectors. Smart lighting. Smart doorbells. While desktop computers and even smartphones are powerful standalone tools, Internet of Things devices share a need to collect data from the environment, store it, and transmit it to some other device for action or further analysis. In many systems, data storage and working memory account for the majorit... » read more

Creating 2D Compounds


A 2D material, by definition, has no surface dangling bonds. A bulk material with plate-like structure, such as graphite, is composed of thin layers with a weakly bonded cleavage plane between. What this means is a monolayer of graphite will seek to satisfy its exposed dangling bonds by absorbing other materials. A monolayer of graphene, in contrast, is energetically complete without a secon... » read more

← Older posts