• 12 Posts
  • 102 Comments
Joined 5 years ago
Cake day: January 21st, 2020
  • leanleft@lemmy.mltoProgramming@programming.dev•What search engine do you use?English
    1·
    6 months ago

    Not perfect:
    But here are some ideas.

    • reddit (tor)
    • wikipedia (tor)
    • certain niche direct sources (tor)

    More ideal ideas:

    • certain specific private source (maybe tor). Not reliable but high quality.
    • large collection of raw links ( requires labor, skill. And yields imperfect results )
    • searx ( mainly to share with friends and as a fallback. it is a pretty great premade metasearch engine when selfhosted)
    • ready-to-go foss searchengine implementations. (Limited in scope, requires decent amount of “labor” and time. Requires setup phase and light maintaining. Extremely high quality results. Optionally invest money in various ways to supercharge. Perhaps recruit collaborators)

    other stuff:

    • creating private collections and bookmarks
    • not using internet or using rarely or using in cautious way. Or not using www.
    • focusing on distracting self with hands-on projects.

    What am i actually using at this point? (Nothing is set up currently!).

    • Sometimes i use tor 70% of the time. Sometimes i use tor 30% of the time.
    • very frequently non www .
    • duckduckgo when needed. [Often] without visiting the links
    • niche sources (2, …)
    • reddit

    *this isnt perfect! but i think overall i think i dont spend much time traveling to websites for info.

    Historically:

    • searx
    • searchengine
    • dabble in scaling.

    Future:

    • selfhost
    • scaling
    • further isolation
    • IRL
    • other stuff. such as creating new solutions.

    There is certainly room for immediate improvement here.
    Im just lazy.

    I dont need the internet as much as the internet needs me.

  • leanleft@lemmy.mltoLinux@lemmy.ml•Upstreaming Linux kernel support for the Snapdragon X EliteEnglish
    2·
    6 months ago

    Found some background info

    https://www.pcmag.com/news/qualcomm-snapdragon-x-elite-oryon-unveiled

    “detailed the first SoC in the company’s Snapdragon X Elite line, powered by its much-anticipated next-gen CPU core, code-named “Oryon.” [Teased earlier in the month] (https://www.pcmag.com/news/qualcomm-teases-next-gen-snapdragon-x-pc-platform), “Snapdragon X” is the branding for Qualcomm’s newest SoCs for PC compute, and the Snapdragon Elite X is the first issue, positioned as its premium solution.”

    “the punchiest processor for laptops that it has ever produced.”

    “The 8cx chips were built around a CPU core that Qualcomm dubbed Kryo. Oryon is a newer CPU core that will power the conventional compute in Snapdragon X Elite. It was announced at 2022’s Snapdragon Summit and will underpin future Qualcomm initiatives in areas including laptop, mobile phone, automotive, and mixed reality experiences. It’s a custom core (rather than a licensed-from-Arm core) and a product, in part, of the company’s 2021 acquisition of Nuvia,”

    “Oryon (pronounced like “Orion,” the star system) in its initial offering is a 12-core Arm CPU core, custom-designed by Qualcomm, built on 64-bit architecture and 4nm process technology. It’s the successor to the Kryo used in Snapdragon 8cx Gen 3 (5nm process) and earlier 8cx efforts.
    The overall boost clock on these 12 cores is 3.8GHz, with the ability (a bit like Intel chips with their various Turbo Boost and Turbo Boost Max technologies) to boost just one or two cores to 4.3GHz. According to Qualcomm, this limited acceleration should manifest in faster application launch times, better web browsing responsiveness, and snappier UI. The company also points out that, when in boost mode, these cores are the world’s first 4GHz-capable Arm cores. The cores on this initial Oryon effort are clustered into three sets of four. All of them are designated as high-performance cores, in contrast to the “hybrid design” (Intel’s term) of Intel’s recent-generation Core desktop and mobile processors, most of which are divided into banks of Performance and Efficient cores (P-cores and E-cores).”

    "integrated neural processing unit (NPU), dedicated silicon for processing the large data sets associated with AI workloads. (See: Intel’s “Meteor Lake” laptop chips, coming in December, and AMD’s recent Phoenix mobile processors with Ryzen AI.) The Elite X employs Qualcomm’s own Hexagon NPU, which in earlier times was better classed as a digital signal processor (DSP). In mobile designs, this kind of DSP would often be allocated side jobs like image processing to keep workloads off the hungrier CPU; now, AI and machine-learning workloads are in its purview. The Hexagon silicon is rated for 45 TOPS (trillions of operations per second) under INT4. In addition, according to the company, the NPU is capable of handling large language models (LLMs) up to 13B parameters. (LLMs with 7B parameters are also supported. With those smaller models, 30-token-per-second processing is possible.) "

    “Main memory is now LPDDR5x, supporting 136GB per second of memory bandwidth. Capacities to 64GB will be supported on the platform at the discretion of the OEM. The LPDDR5x is backed by 42MB of total cache.”

    “This being a Qualcomm processor, with the company’s pedigree, you’d expect leading-edge connectivity aspects to the platform, and Elite X holds to that. Wi-Fi 7 support is on the menu, as well as, of course, 5G in select SKUs as implemented”