AI workloads Archives

Tech Optimizer

March 19, 2025

7 Reasons PostgreSQL is a Great Choice for AI Projects

PostgreSQL, or Postgres, is increasingly recognized as a leading choice for AI projects due to its strong integration capabilities, cost-effectiveness, and scalability. It supports vector similarity search essential for AI tasks through extensions like pgvector, which simplifies storage and querying of vectors. The latest pgvector version 0.8.0 introduced enhancements such as iterative index scans and improved cost estimation. PostgreSQL optimizes query performance with various index types, including B-tree, Hash, BRIN, GiST, and SP-GiST indexes, and allows for custom index creation. It also features native JSON and NoSQL capabilities, enabling efficient handling of semi-structured data. Parallel processing and query execution are supported, allowing faster data processing on multi-core machines. Scalable and distributed computing options are available, including Multi-Master Asynchronous Replication and Multi-Master Sharded PostgreSQL, catering to the growing demand for AI applications. PostgreSQL ensures AI data security and compliance through Access Control Lists, Row Level Security, and Transparent Data Encryption. Its open-source nature allows for flexibility and integration with AI frameworks, making it a cost-effective alternative to proprietary databases. PostgreSQL was recognized as the Most Popular Database in the 2024 Stack Overflow Developer Survey, reflecting its strong adoption and evolving capabilities in AI projects.

Winsage

March 6, 2025

I get to play with the world’s most powerful gaming tablet, and it’s totally awesome

The ASUS ROG Flow Z13 (2025) is a 2-in-1 convertible device designed for gaming and creative tasks, featuring a Windows-powered tablet with a large touch-enabled display, an adjustable kickstand, and a keyboard that doubles as a protective cover. It includes the Ryzen AI MAX platform, which integrates a powerful CPU, NPU, and advanced integrated GPU, allowing for exceptional performance without additional components. The device starts at ,099.99, with configurations and pricing that can be complex, but broader availability is expected. The unboxing experience is premium, featuring a sleek design and a substantial 200W charger. The tablet has a 13.4-inch display, a solid kickstand, and a responsive keyboard cover. It is equipped with a Ryzen AI MAX+ 395 CPU, a Radeon 8060S GPU, 1TB of PCIe Gen 4.0 SSD storage, and can be configured with up to 128GB of RAM. Battery life is decent, with a 70Whr battery allowing several hours of use, even while gaming. The ROG Flow Z13 caters to a niche audience seeking innovative hardware in a portable form factor, though potential buyers may find traditional gaming laptops offer better value.

Tech Optimizer

February 25, 2025

APAC enterprises shifting towards open source in database strategies

Enterprises in the Asia-Pacific region are increasingly transitioning from proprietary databases to open-source alternatives, with PostgreSQL being favored due to high licensing costs and vendor dependencies. IT leaders are reassessing database strategies to meet performance, cost, and scalability needs, particularly for applications requiring transactional, analytical, and AI workloads. Hybrid IT models are gaining traction, especially in countries like Singapore, Indonesia, and the Philippines, while Japan and Korea show strong on-premises demand, and India is moving away from legacy systems due to cost pressures. PostgreSQL is evolving to support hybrid IT and emerging workloads, including AI and analytics, with enhancements like lakehouse architecture for managing structured and unstructured data and the introduction of pgvector for AI applications. Data sovereignty regulations are prompting enterprises to deploy PostgreSQL in hybrid cloud environments to balance control and scalability. EDB has developed a hybrid control plane for unified management across on-premises and cloud deployments. Many enterprises are opting for a gradual transition to PostgreSQL due to legacy dependencies and operational risks, often starting with specific use cases before migrating legacy workloads. An example includes an insurance company in Hong Kong that migrated from Oracle to PostgreSQL, achieving reduced costs and improved performance. Successful migration depends on database mobility and application portability, with EDB planning to integrate GPU support and develop tools for faster AI application deployment, with expectations for rollout by the second quarter of 2025.

Tech Optimizer

February 21, 2025

Multi-tenant vector search with Amazon Aurora PostgreSQL and Amazon Bedrock Knowledge Bases

The article discusses the implementation of a multi-tenant vector store using Amazon Aurora PostgreSQL-Compatible Edition and Amazon Bedrock Knowledge Bases. It outlines a scenario where users request home surveys, and surveyors update findings stored in an Amazon S3 bucket. The process involves converting documents into vector embeddings for enhanced natural language queries through the Retrieval Augmented Generation (RAG) technique. Key steps include: 1. Ingesting data from S3 into Amazon Bedrock Knowledge Bases. 2. Using an embeddings model to convert documents into vector embeddings. 3. Storing vector embeddings, data chunks, and metadata in Aurora with pgvector. 4. Submitting natural language queries that are transformed into embeddings for retrieval from the vector store. 5. Forwarding relevant documents to a large language model for response generation. To set up the vector store, prerequisites include access to an Amazon Aurora PostgreSQL-compatible database, Amazon S3 for document storage, and Amazon Bedrock for managing knowledge bases. The article provides SQL commands for creating the necessary schema, vector table, and index in Aurora, as well as instructions for ingesting data and ensuring multi-tenant data isolation through metadata filtering. Best practices for deploying a multi-tenant vector store include optimizing chunk size, evaluating embedding models, and monitoring query performance. The article emphasizes the benefits of fully managed solutions to reduce operational complexity.

Tech Optimizer

February 20, 2025

EDB Postgres® AI Significantly Outperforms Oracle, SQL Server, MongoDB, and MySQL in New Benchmark Study

EnterpriseDB (EDB) has released findings from a benchmark study by McKnight Consulting Group, showing that EDB Postgres AI outperforms Oracle, SQL Server, MongoDB, and MySQL in various workloads, including transactional, analytical, and AI tasks. Key performance metrics include being 150 times faster than MongoDB in processing JSON data, 4 times faster than MySQL in handling insert operations, and outperforming Oracle by 17% and SQL Server by 30% in processing New Orders Per Minute (NOPM). EDB Postgres AI also offers 7 times better price performance than Oracle and 6 times better than SQL Server. The study highlights the challenges enterprises face with legacy systems consuming 55% of IT budgets, which hampers modernization efforts. EDB Postgres AI aims to address these challenges by streamlining data infrastructure, reducing total cost of ownership, and facilitating AI capabilities in a secure environment.

Tech Optimizer

February 4, 2025

EDB Earns Top Industry Honors with EDB Postgres AI, Appoints Nancy Hensley as Chief Product Officer

EnterpriseDB (EDB) has introduced EDB Postgres® AI, the first unified data and AI platform for transactional, analytical, and AI workloads. Nancy Hensley has been appointed as Chief Product Officer (CPO) to guide the company through its next growth phase. Recent studies show that 35% of enterprises are considering Postgres for database initiatives, with 65% integrating AI in hybrid environments. EDB has achieved several milestones, including recognition as a top database solution and expanding its customer base to over 1,500 globally. EDB is the largest contributor to PostgreSQL, accounting for approximately 30% of code contributions.

Tech Optimizer

February 4, 2025

EDB Bridges Open Source Innovation and Enterprise-Grade Impact with CloudNativePG’s CNCF Milestone

EnterpriseDB (EDB) has announced that its open-source Kubernetes operator for PostgreSQL, CloudNativePG (CNPG), has been accepted into the Cloud Native Computing Foundation (CNCF) Sandbox. This acceptance underscores the significance of Kubernetes in enterprise strategies and EDB's commitment to innovation. CloudNativePG is recognized as the most popular PostgreSQL operator for Kubernetes on GitHub, with over 5,000 stars. EDB provides a data and AI platform that supports over 1,500 customers globally, including major government agencies and leading companies in various sectors.

Tech Optimizer

December 24, 2024

PostgreSQL vs. the Cloud for GenAI? 4 Things To Consider

Generative AI (GenAI) is expected to transform industries by improving productivity and revenue. A Deloitte report states that 80% of enterprises with advanced AI capabilities plan to increase AI investments. Companies must modernize legacy systems and tackle data challenges to implement GenAI, but many still use outdated technology, leading to costly transitions and downtime. Nearly half of US generative AI decision-makers expect returns on their investments within three years, which may underestimate the complexities of AI implementation. While cloud computing offers scalability, managing costs can be difficult, prompting many enterprises to adopt financial operations frameworks (FinOps). Concerns about third-party cloud services are leading businesses to consider on-premises and hybrid solutions, with predictions that by 2025, half of enterprises will use on-premises models for AI development. Choosing the right database is crucial for GenAI development, with PostgreSQL being a strong option due to its extensibility, seamless integration, open-source benefits, and deployment flexibility. Successful GenAI strategies focus on rapid development, cost-effective failures, and effective scaling, with Percona offering support to help organizations achieve these goals.

Tech Optimizer

December 11, 2024

EDB Brings Cloud Agility and Observability to Hybrid Environments with Sovereign Control

EnterpriseDB (EDB) has introduced advancements to its EDB Postgres® AI platform, enabling enterprises to implement secure, adaptable AI-driven applications within a sovereign, hybrid framework. The platform offers a unified interface for transactional, analytical, and AI workloads, facilitating quicker AI project development. A recent EDB study found that over 56% of organizations are deploying mission-critical workloads in a hybrid model. The EDB Postgres AI platform features a Hybrid Control Plane for centralized management and automation, enhancing application performance and database health monitoring. The AI Accelerator allows enterprises to test and deploy generative AI applications within the Postgres environment, automating the data pipeline. The Analytics Accelerator improves high-performance analytics and offers near real-time OLAP capabilities. PostgreSQL version 17 is now available across EDB products, supporting various deployment environments. EDB serves over 1,500 customers globally, providing high availability and compliance controls.

Winsage

December 6, 2024

Phi Silica, small but mighty on-device SLM

The Applied Sciences team has developed the small language model (SLM) Phi Silica, which enhances power efficiency, inference speed, and memory efficiency for Windows 11 Copilot+ PCs using Snapdragon X Series NPUs. Phi Silica is designed for on-device use and supports multiple languages, featuring a 4k context length. Microsoft announced that developers will have access to the Phi Silica API starting January 2025. The Copilot+ PCs can perform over 40 trillion operations per second, achieving significant performance improvements when connected to the cloud. Phi Silica utilizes a Cyber-EO compliant derivative of Phi-3.5-mini, and its architecture includes components such as a tokenizer, detokenizer, embedding model, transformer block, and language model head. The model's context processing consumes only 4.8mWh of energy on the NPU, with a 56% improvement in power consumption compared to CPU operation. Phi Silica features 4-bit weight quantization for efficiency, rapid time to first token, and high accuracy across languages. The model was developed using QuaRot for low-precision inference, achieving 4-bit quantization with minimal accuracy loss. Techniques like weight sharing and memory-mapped embeddings were employed to optimize memory usage, resulting in a ~60% reduction in memory consumption. Innovations such as a sliding window for context processing and a dynamic KV cache were introduced to expand context length. The model has undergone safety alignment and is subject to Responsible AI assessments and content moderation measures.