Estimating Motion in Images Image based motion estimation is a class of techniques which analyze sequential image frames and determine how each region of an image moves from one frame to the next. Being able to track object movement between image frames has a broad range of applications and use cases, including SpaceWarp-style frame extrapolation, used in gaming and spatial computing devices to improve the smoothness of the digital experience. Our Adreno Motion Engine support, powered by the Qualcomm Adreno GPU, has been previously covered here . Aperture issue One common challenge motion esti ..read more

Co-written with Apoorva Gokhale Figure 1: Generation on Qualcomm Cloud AI 100 Ultra (1) Baseline with FP16 weights (2) Acceleration with MX6 (3) Acceleration with MX6 and SpD Speculative Sampling (SpS), also known as Speculative Decoding (SpD), and weight compression through MXFP6 microscaling format, are two advanced techniques that significantly enhance large language model (LLM) decoding speeds for LLM inference AI workloads. Both techniques are available for LLM acceleration on Qualcomm Technologies' data center AI accelerators. To achieve a significant inference performance speedup, star ..read more

From health care to education and manufacturing, AI use cases grow exponentially. Developers choose popular AI Frameworks like PyTorch and TensorFlow to analyze predictions, train models, leverage data, and improve future outcomes. In this article, we explain how to improve the performance of framework trained models. Read on to learn more about a powerful platform complemented by high performance AI software development kits. Introduction to Windows on Snapdragon For the past few years, Qualcomm Technologies and Microsoft have been partnering to create a powerful platform called Windows on Sn ..read more

With smartphones, portable speakers, earbuds and in-vehicle audio, it keeps getting easier to listen to your music anywhere you want to. But what about making the sound in harmony with real listener preference? Alango Technologies Ltd. has more than 20 years of experience in digital signal processing (DSP) software for speech, hearing and audio applications. The Israel-based company’s software is used in devices like hands-free car kits, conference systems, speakerphones, Bluetooth headsets, True Wireless Stereo (TWS) earbuds, hearing enhancement products and voice-controlled devices. SonicVib ..read more

One of the main uses of a GPU is image (texture) sampling and processing. GPUs have specialized, built-in hardware to perform nearest-neighbor, bilinear, and bicubic filtering (with the VK_EXT_filter_cubic extension). However, some use cases require sampling with even larger kernels or with customized kernel weights. These cases can be manually implemented in a fragment or compute shader using the existing sampling instructions. However, that requires many round trips between the texture and shader units, which is not ideal from a power or performance perspective. In this post, we’ll describe ..read more

Our Qualcomm Developer Network YouTube channel continues growing as a valuable resource for development videos. Recently, we added several videos to the channel’s Windows on Snapdragon (WoS) playlist for Windows app developers: Building on Windows on Snapdragon: 5 Reasons to Develop an App on WoS Building on Windows on Snapdragon: Power of Artificial Intelligence on WoS Building on Windows on Snapdragon: Introduction to Snapdragon Profiler Building on Windows on Snapdragon: Snapdragon LLVM for WoS Let’s take a quick look at what each video has to offer. 1. Building on Windows on Snapdragon: 5 ..read more

Motion-based security systems are known for their high rate of false positives from events such as passing animals and the wind rustling nearby trees. It’s a use case ripe for AI, which can use computer vision and inference to weed out both false positives and false negatives. The result is more-accurate detection that increases profits by reducing the expenses associated with making in-person security checks. With the proliferation of high-resolution cameras and IoT devices, and increasing data volumes, CVEDIA believes the notable shift towards using AI on edge devices offers clear advantages ..read more

Co-written with Prakruthi Deepak Heragu If you’ve started working with Android Virtualization Framework (AVF), then you probably know that it offers a set of APIs so you can configure and launch protected virtual machines (VMs). Many app developers have use cases they want to move out of main Android into isolation and protected VMs with the help of a hypervisor, and AVF makes that possible. You’ve probably also worked with the protected kernel-based virtual machine (pKVM), which is the reference hypervisor for AVF. It’s well-suited to computing use cases, but in the Qualcomm Innovation Center ..read more

Co-written with Elliot Berman Instead of copying and parsing the complete RAM dump after a device crash, what if you could collect only the debug information you want? It would reduce the time and resources needed for transfer and storage, and you could study the information much more efficiently. Qualcomm Technologies’ infrastructure for RAM dump collection, residing in proprietary boot firmware, supports minidump on our system-on-chips (SoCs). The minidump driver developed by Qualcomm Innovation Center, Inc. (QUIC) provides for collecting either a complete RAM dump or a minidump with only th ..read more

How do you maintain backward compatibility with userspace apps in the Linux kernel? Most maintainers rely on code review and pushing changes out for testing. But with experienced reviewers and testing farms in short supply, it can be helpful to add automated tooling at build time to check for compatibility. At the Qualcomm Innovation Center, Inc. (QUIC), we’ve developed a shell script that performs application binary interface (ABI) analysis of patches. We use it in house to detect at build time whether a particular change we’ve made to the Linux kernel will break compatibility with the usersp ..read more