Best of Computing — 2025

  1. 1
    Video
    Avatar of lauriewiredLaurieWired·35w

    What if Humanity forgot how to make CPUs?

    A thought experiment exploring what would happen if humanity lost the ability to manufacture CPUs. The scenario traces the gradual decline of modern computing over 30 years, from cloud infrastructure failures to internet fragmentation, highlighting how older hardware with larger node sizes would outlast modern chips due to electromigration. The piece examines how software optimization would become critical, vintage computing would resurge, and society would eventually revert to 1970s-80s era technology.

  2. 2
    Article
    Avatar of sitepointSitePoint·1y

    Learn Data Structures and Algorithms: Complete Tutorial

    Data Structures and Algorithms (DSA) are essential for efficient programming and form the backbone of modern computer science. Mastering both linear and non-linear data structures enables developers to handle diverse scenarios effectively. Techniques such as divide-and-conquer, dynamic programming, and greedy algorithms are crucial for solving complex problems efficiently. Understanding DSA is vital for building scalable applications and excelling in technical interviews. The post also discusses the latest trends like quantum algorithms, AI-driven data structures, and blockchain optimizations.

  3. 3
    Article
    Avatar of hnHacker News·39w

    You should buy a faster CPU

    Modern CPUs have become dramatically faster, with desktop processors offering up to 10x performance improvements over older laptop chips. The author argues that investing in top-tier CPUs like the AMD Ryzen 9950X ($500) provides better value than AI coding subscriptions ($480/year) for developer productivity. Benchmarks show desktop CPUs are roughly 3x faster than laptops, and current generation processors are 3x faster than models from 3 years ago, making build times significantly shorter and development workflows more efficient.

  4. 4
    Article
    Avatar of lonely_programmerLonely Programmer·38w

    Before and After Linux

  5. 5
    Video
    Avatar of lauriewiredLaurieWired·1y

    Pixar DOESN'T use GPUs...

    Pixar primarily uses CPUs instead of GPUs for rendering due to their flexibility and ability to handle large scenes with over 150 million polygons. Their render farms leverage AVX 512 and SSE 4.2 extensions for optimization. While GPUs can offer significant speed gains, they are limited by their memory capacity, making CPUs a more viable option for Pixar's needs.

  6. 6
    Article
    Avatar of phoronixPhoronix·28w

    Intel Core Ultra 7 255H Linux CPU Performance Review

    Comprehensive benchmark results for Intel's Core Ultra 7 255H processor (Arrow Lake H) running on Ubuntu Linux. The 16-core mobile processor features 6 P-cores, 8 E-cores, and 2 LPE cores with a maximum turbo frequency of 5.1GHz. Testing was conducted on a Lenovo ThinkPad P1 Gen 8 in both default and performance modes, comparing performance against 25+ other Intel and AMD mobile processors across multiple generations. The processor supports up to 128GB DDR5-6400 memory and includes Intel Arc 140T graphics.

  7. 7
    Video
    Avatar of computerphileComputerphile·1y

    How CPU Memory & Caches Work - Computerphile

    The post explains the concepts of CPU memory and caches, detailing how computer memory cells work, the difference between static and dynamic RAM, and the role of caches in enhancing memory access speeds. It discusses the importance of volatile memory, the mechanisms of flip-flops and capacitors, and techniques used to optimize memory access times. The post also explores the different levels of cache memory and the reasons behind their implementation to improve overall system performance.

  8. 8
    Article
    Avatar of lobstersLobsters·1y

    Fast(er) binary search in Rust

    Binary search is a fast algorithm but suffers from predictability issues on modern CPUs. This post explores the Eytzinger layout to improve memory access predictability and introduces a branchless version of the Eytzinger binary search. It also discusses the benefits of using software memory prefetching to further enhance performance.

  9. 9
    Article
    Avatar of tailscaleTailscale·1y

    Living in the future, by the numbers

    Over the past two decades, technological advancements have rapidly transformed computation power, web server efficiency, memory capacity, and storage capabilities. Modern CPUs, GPUs, and TPUs have significantly sped up numerical computations, while web servers now benefit from refined programming languages and more efficient hardware. RAM and hard disk sizes have dramatically increased, and SSDs provide unparalleled transaction speeds and reliability. These improvements allow for vastly different and more efficient system designs than those used in the early 2000s.

  10. 10
    Video
    Avatar of bitlemonBitLemon·1y

    How Virtual Memory Actually Works Behind the Scenes

    The post explains how virtual memory works behind the scenes, focusing on the structure and function of page tables and the role of the memory management unit (MMU). It covers how page tables are used to map virtual memory to physical memory, the fields within page table entries, and how the MMU manages memory requests and translations. The post also discusses the translation lookaside buffer (TLB) and its function in speeding up address translation.

  11. 11
    Article
    Avatar of rpythonReal Python·1y

    NumPy Techniques and Practical Examples – Real Python

    The NumPy library is essential for scientific computing in Python, providing multidimensional arrays for complex data analysis. This video course teaches you how to handle and analyze data using advanced NumPy features. It includes 14 lessons, subtitles, downloadable resources, an interactive quiz, and a certificate of completion.

  12. 12
    Article
    Avatar of hnHacker News·34w

    Apple's iPhone 17 chip becomes the fastest single-core CPU in the world on PassMark, beating PC chips and Apple's own M3 Ultra — passively-cooled A19 CPU catapults past power-hungry competitors

    Apple's A19 chip in the iPhone 17 achieved the highest single-core performance score on PassMark benchmarks, surpassing desktop processors from Intel, AMD, and even Apple's own M3 Ultra. The A19 scored 5,149 points while consuming only 4W of power, compared to 44W for Intel's Core Ultra 9 285K and 56W for AMD's EPYC processor. However, the chip's multi-threaded performance remains limited due to fewer cores compared to desktop alternatives.

  13. 13
    Video
    Avatar of t3dotggTheo - t3․gg·1y

    I finally know how CPUs work (w/ Casey Muratori)

    The post discusses the complexity of CPU workings, focusing on how hardware and software interact. It features a conversation with Casey Muratori about the differences between high-level programming instructions and the lower-level micro operations that CPUs execute. They explore why hardware companies keep their internal workings secret, the misconceptions around instruction sets, and the importance of understanding CPU architecture for programmers. The discussion emphasizes the significance of efficient coding practices and the evolving landscape of JavaScript and hardware interactions.

See all Computing archives