Posts

How I built AI Deep Learning Workstation

This article explores the process of building a custom AI and deep learning workstation , delving into its benefits and potential challenges. It provides a detailed rationale for creating such a system, emphasizing its value for those passionate about the hardware side of AI, seeking local development environments, or conducting budget-conscious research. The article covers essential technical considerations, such as choosing a powerful GPU, a compatible CPU, and a high-performance motherboard. It also highlights the importance of adequate RAM, a spacious case to accommodate hardware, and a reliable power supply to ensure the system runs efficiently. These details help readers understand what’s needed to build a workstation capable of handling the demanding workloads of AI and deep learning. Beyond component selection, the article addresses the costs of high-end hardware like GPUs and the technical know-how required to assemble a system. While noting the availability of free platfo...
Post a Comment
Read more

NVIDIA GPUs with 12 GB of video memory

This article offers an in-depth look at NVIDIA GPUs with 12 GB of memory, diving into their specifications and standout features. From architecture and memory type to bandwidth, CUDA cores, Tensor Cores, and power requirements, every detail is covered to give you a clear understanding of what makes each model unique. Beyond the specs, the article explores how these GPUs excel in various applications, including deep learning, scientific computing, gaming, and rendering. Whether you're building a powerhouse for AI research, crafting stunning visuals, or optimizing for gaming performance, you'll find practical insights to guide your decision. With a focus on balancing performance, versatility, and budget, this guide helps you discover the perfect NVIDIA GPU to meet your needs and achieve your goals. Listen to the podcast generated based on the article by NotebookLM. In addition, I shared my experience of building an AI Deep learning workstation in ⁠ another article ⁠. If...
Post a Comment
Read more

NVIDIA GPUs with 16 GB of Video RAM

This article takes you on an exciting journey through the specifications and performance of NVIDIA GPUs with 16GB of video memory, showcasing their remarkable capabilities for AI and deep learning. Packed with insights, it highlights key features like memory type, bandwidth, CUDA cores, Tensor cores, power requirements, cooling systems, and supported data types, giving you a clear understanding of each GPU’s strengths. Discover how these GPUs excel in machine learning tasks, from training massive models to delivering fast and efficient inference. The guide also provides recommendations tailored to specific applications—whether you’re tackling intensive deep learning workflows or seeking a versatile solution for general-purpose computing. If you’re gearing up to supercharge your AI projects or fine-tuning for inference optimization, this comprehensive guide offers everything you need to choose a top-performing GPU and unlock new levels of productivity and innovation. Listen to ...
Post a Comment
Read more

NVIDIA GPUs with 24 GB of Video RAM

This article delves into a lineup of high-performance NVIDIA GPUs featuring 24 GB of video memory, perfectly suited for deep learning, AI, and high-performance computing (HPC). Each GPU is analyzed by architecture—spanning Maxwell, Pascal, Turing, Ampere, and the latest Ada Lovelace—with detailed insights into key specifications, including memory type, bandwidth, CUDA cores, Tensor Cores, power consumption, and system connectivity. Beyond showcasing the strengths of each model, the article provides a balanced perspective by addressing potential limitations. This comprehensive guide empowers you to make an informed decision, helping you select the ideal GPU to meet your unique needs and drive your applications to new levels of performance and efficiency. Listen to the podcast generated based on the article by NotebookLM. In addition, I shared my experience of building an AI Deep learning workstation in ⁠ another article ⁠. If the experience of a DIY workstation peeks your intere...
Post a Comment
Read more

Nvidia GPUs with 48 GB Video RAM

This article provides an engaging guide to NVIDIA GPUs with 48GB of video RAM, purpose-built for the high demands of generative AI and deep learning applications. Organized as a timeline, it takes you on a journey from earlier models to the latest cutting-edge releases, offering a comprehensive look at each GPU's standout features. For each entry, key specifications are spotlighted, including architecture, memory, CUDA cores, Tensor cores, compute compatibility, power consumption, cooling system, and support for NVIDIA NVLink. The article highlights how NVIDIA's GPU technology has evolved, showcasing the increasing power, speed, and efficiency of newer generations. These advancements make handling complex AI tasks faster and more flexible than ever before. Whether you’re a researcher, developer, or part of an organization scaling AI infrastructure, this guide is an invaluable resource. It simplifies the process of choosing the perfect GPU to fuel your AI and deep learning ...
Post a Comment
Read more

The most powerful NVIDIA datacenter GPUs and Superchips

This article provides an in-depth overview of NVIDIA’s datacenter GPUs , categorizing them by architecture—Pascal, Volta, and Ampere—and interface types such as PCIe and SXM. It highlights essential features, including CUDA cores, memory bandwidth, and power consumption, for each model. Special attention is given to the differences between PCIe and SXM interfaces, with SXM standing out for its ability to enable faster inter-GPU communication—an advantage critical for training large-scale AI models. The article also guides readers in choosing the right GPU by considering specific requirements such as memory capacity and precision support. The discussion moves on to NVIDIA’s flagship GPUs, including the A100 (Ampere architecture) and the latest H100/H200 series (Hopper architecture). Detailed specifications such as memory size, bandwidth, CUDA cores, and power consumption are outlined, along with interface options like PCIe, SXM4, SXM5, and NVL. The article also introduces NVIDIA Super...
Post a Comment
Read more

Binary classification with Tensorflow 2

Image
This post uses Tensorflow with Keras API for a classification problem of predicting diabetes based on a feed-forward neural network also known as multilayer perceptron and uses Pima Indians Diabetes Database from Kaggle . A Google colab notebook with code is available on GitHub . Exploratory data analysis The dataset consists of 8 numeric features each of which does not have any missing values. The database contains 768 records from which 500 correspond to negative outcomes and 268 to positive. There are no features that strongly correlate to each other. Building a model We split the dataset into the training part which constitutes 80% of the whole data and the test part of 20%. A sequential model consisting of 6 layers. The first one is a normalization layer that is a kind of experimental preprocessing layer used to coerce it inputs to have distributions with the mean of zero and standard deviation of one. The model contains two three fully-connected layers, two wit...
Post a Comment
Read more