Targeted at ADAS and autonomous vehicle (AV) applications, the TDA4VM88TGBALFR processor family is based on the evolving Jacinto™ 7 architecture and builds on TI's extensive market knowledge gained from over a decade of leadership in the ADAS processor market. A unique combination of high-performance computing, deep learning engines, and dedicated accelerators for signal and image processing in a functional safety-compliant architecture makes TDA4VM devices ideal for a wide variety of industrial applications, such as: robotics, machine vision, radar, and more . TDA4VM provides high-performance computing for traditional and deep learning algorithms with an industry-leading power/performance ratio, and a high level of system integration, enabling advanced automotive platforms supporting multiple sensor modes in centralized ECUs or standalone sensors. Scalability and lower cost. Key cores include next-generation DSPs with scalar and vector cores, dedicated deep learning and traditional algorithm accelerators, the latest Arm and GPU processors for general-purpose computing, integrated next-generation imaging subsystems (ISPs), video codecs, Ethernet network hubs and isolated MCU islands. All of this is protected by automotive-grade security hardware accelerators.

Overview of the main high-performance cores

"C7x" next-generation DSP combines TI's industry-leading DSP and EVE cores into a single higher-performance core and adds floating-point vector computation capabilities, enabling backward compatibility with legacy code while simplifying software programming. Operating at the typical automotive maximum junction temperature of 125°C, the new "MMA" deep learning accelerator can achieve up to 8TOPS of performance in the industry's lowest power envelope. A dedicated ADAS/AV hardware accelerator provides vision pre-processing as well as distance and motion processing without compromising system performance.

General Computing Kernels and Integration Overview

A standalone dual-core cluster configuration of the Arm Cortex-A72 facilitates multi-OS applications with very low software hypervisor requirements. Up to six Arm Cortex-R5F subsystems can manage low-level timing-critical processing tasks and keep the Arm Cortex-A72 independent of the application. The integrated "8XE GE8430" GPU delivers up to 100GFLOPS of performance, enabling dynamic 3D rendering for enhanced vision applications. TI's 7th-generation ISP builds on the best existing ISPs, with the flexibility to handle a wider sensor suite, support for higher bit depths, and features for analytics applications. Integrated diagnostics and security features support operations up to ASIL-D/SIL-3 levels, while integrated security features protect data from modern attacks. For systems that require large amounts of data bandwidth, a PCIe hub and Gigabit Ethernet switch are provided along with CSI-2 ports to support the throughput of numerous sensor inputs. For further integration, the TDA4VM family also includes an MCU island, eliminating the need for an external system microcontroller.