Benchmarks

NVIDIA's Jetson TX2: Embedded Designs Gain a Deep Learning Upgrade

Jetson TX2 is NVIDIA's latest board-level product targeted at computer vision, deep learning, and other embedded AI tasks, particularly focused on "at the edge" inference (when a neural network analyzes new data it’s presented with, based on its previous training) (Figure 1). It acts as an upgrade to both the Tegra K1 SoC-based Jetson TK1, covered in InsideDSP in the spring of 2014, and the successor Tegra X1-based Jetson TX1, which BDTI evaluated for deep learning and other computer vision Read more...

Case Study: Careful Analysis Leads to Successful Products

Processor vendors and system designers share a common concern: how to make sure their products meet customer needs. For processor vendors, a key challenge is to design architectures with enough performance to meet the demands of current and anticipated applications while staying within acceptable power and cost constraints, and enabling good software developer productivity. Fundamentally, processor designers need to bring together the demands of algorithm workloads together with the Read more...

Case Study: Choosing the Right Benchmarks for the Job

As embedded processors and applications become increasingly complex, good benchmarks are more important than ever. System designers need good benchmarks to judge whether a processor will meet the needs of their applications, and to make accurate comparisons among processors. Processor developers need good benchmarks to assess how their processors stack up against the competition, and to prove their processors' capabilities to customers. But what exactly comprises a good benchmark? One obvious Read more...

Case Study: BDTI's Expert, Independent Analysis Enables Optimum Vision Processor Selection

A growing number of products are incorporating computer vision capabilities. This, in turn, has led to rapid growth in the number of processors being offered for vision applications.  Selecting the best processor (whether a chip for use in a system design, or an IP core for use in an SoC) is challenging, for several reasons. First, these processors use very diverse architecture approaches, which makes it tough to compare them. Second, because vision applications and algorithms are also quite Read more...

CEVA Software Framework Brings Deep Learning to Embedded Vision Systems

As Jeff Bier has mentioned in several of his recent columns, deep learning algorithms have gained prominence in computer vision and other fields where there's a need to extract insights from ambiguous data. Convolutional neural networks (CNNs) – massively parallel algorithms made up of layers of computation nodes – have shown particularly impressive results on challenging problems that thwart traditional feature-based techniques; when attempting to identify non-uniform objects, for example, or Read more...

Qualcomm's QDSP6 v6: Imaging and Vision Enhancements Via Vector Extensions

Qualcomm has been evolving its in-house DSP core for many years. Originally developed for use in Qualcomm’s cellular modems, more recently it has also found use as an application co-processor, offloading multimedia tasks from the CPU in smart phones and tablets. Earlier InsideDSP articles covered the v4 "Hexagon" DSP core in mid-2012 and early 2013, along with the v5 Hexagon architecture later that same year. Now, with its Hexagon v6 DSP core, which will see its first silicon implementation in Read more...

ARC Processor Core Enhancements Promise Performance, Energy Consumption Improvements

In May 2014, Synopsys expanded its ARC EM licensable processor core product line, which BDTI described as historically being "vanilla" Harvard architecture CPUs with no DSP-optimized features, via the addition of the digital signal processing pipeline-equipped EM5D and EM7D (“D” denoting DSP). This year's follow-on EM9D and EM11D make what at first glance seem to be minor upgrades, in the form of an optional incremental 2-64 KB of special-purpose embedded memory. But, according to company Read more...

Case Study: Squeezing Big Algorithms into Small Power Budgets Enables Mobile Audio Quality

Today's smartphones are technological marvels that deliver an extraordinary range of capabilities from GPS-based navigation to sophisticated photography. But sometimes we just want to make a phone call. And particularly when we're on the move, who hasn't struggled to hear the other party or to be heard on a mobile call? Realizing the importance of intelligible phone calls – not to mention a strong need to differentiate their products – smartphone manufacturers are incorporating increasingly Read more...

NVIDIA Toolsets Target GPU Acceleration of Deep Learning, Other Algorithms for High-Performance Computing

"General-purpose GPU" (or "GPGPU") refers to the use of graphics processors for a variety of non-graphics tasks, and is a frequently discussed topic here at InsideDSP. GPUs are massively parallel processors, originally designed to only handle vertex and pixel operations. However, with the emergence of programmable shader-based architectures beginning with NVIDIA's mainstream GeForce 3 line in 2001 and joined by ATI Technologies' (now AMD's) Radeon 9700 and derivatives unveiled the following Read more...

Xilinx Previews New Chips and Tools for Heterogeneous Processing

Back in early 2010, Xilinx first began discussing its "Extensible Processing Platform" concept, followed by a formal introduction of the Zynq-7000 product family one year later (with initial sampling another year after that). Zynq-7000 wasn't the first processor-plus-programmable logic combo chip; both Xilinx and competitors like Altera had previously developed such devices. But at the time it was unique in that it embedded a full-fledged processor subsystem, including a full peripheral set, Read more...