Research Notes
Research Note: SAP HANA Cloud Vector Engine
SAP recently introduced its new SAP HANA Cloud Vector Engine, designed to bridge the gap between LLMs’ advanced capabilities and enterprise-specific business needs.
Research Note: Intel Gaudi 3
Intel announced its long-anticipated new Intel Gaudi 3 AI accelerator at its Intel Vision event. The new accelerator offers significant improvements over the previous generation Gaudi 3 processor and promises to challenge Nvidia’s current generation accelerators in training and inference for LLMs and multimodal models.
Research Note; Arm Ethos U-65 microNPU
Arm introduced its new Ethos-U65 microNPU (Neural Processing Unit). This state-of-the-art AI accelerator facilitates machine learning (ML) inference in many embedded systems and high-performance devices.
Research Note: AMD Versal Edge Series Gen 2
AMD expands its Versal portfolio with the introduction of its Versal AI Edge Series Gen 2 and Versal Prime Series Gen 2 adaptive SoCs. These next-generation solutions cater to the increasing demands for AI-driven and classic embedded systems, providing a balanced mix of performance, power efficiency, functional safety, and security within a single chip.
Research Note: MLPerf Inference 4.0 Results
MLCommons released the results of its MLPerf Inference v4.0 benchmarks, which introduced two new workloads, Llama 2 and Stable Diffusion XL.
Since its inception in 2018, MLPerf has established itself as a crucial benchmark in the accelerator market. The benchmarks offer detailed comparisons across a variety of system configurations for specific use cases.
Research Note: Databricks DBRX LLM
Databricks launched DBRX, a new open, general-purpose Large Language Model (LLM) that sets a new benchmark for performance and efficiency.
DBRX surpasses the capabilities of existing models like GPT-3.5 while also demonstrating competitive performance with closed models such as Gemini 1.0 Pro, making it a formidable player in general-purpose applications and specialized coding tasks.