• 1 Post
  • 71 Comments
Joined 1 year ago
cake
Cake day: July 8th, 2023

help-circle





  • Infinix Note 30 has an option where when plugging the charger and using the phone at the same time electricity would be used to feed the phone directly, instead of charging the battery (since charging the battery (33w fast charge) while using the phone causes it to heat and lose from its life) thus it stays cool during simultaneous usage and charging: now one could do work on a plugged phone without the risk of ruining the battery






  • zoe@lemm.ee
    cake
    toVideos@lemmy.worldCPU vs GPU vs TPU vs DPU vs QPU
    link
    fedilink
    arrow-up
    1
    ·
    edit-2
    11 months ago

    silicon is really inefficient: too much power wasted in form of heat (Joule effect), on the other hand, quantum computing functions in absolute zero conditions, so no inefficiencies. in a perfect world, one would outsource all its calculations to a quantum instance, but the thing is too costly and centralized; if everyone could afford low latency internet our phones wouldn’t require cpus and gpus, instead we could run them on a quantum computer, but since its too centralized, the issue of privacy would prevail. Would be nice if we had quantum instances scattered around, like lemmy instances, but donations won’t probably cut it to cover costs, so we r stuck with sand computers for now


  • the video explores different processing units used in computers. GPUs are essential for tasks like rendering graphics and training deep learning models. TPUs, on the other hand, are designed specifically for tensor operations required in deep learning and can save significant time and cost when training models. DPUs, or data processing units, are optimized for moving and processing large amounts of data in Big Data Centers, relieving the CPU from such tasks. Finally, there is the potential for the development of QPUs, or quantum processing units, which operate using qubits instead of classical bits and have the potential to revolutionize computing with their ability to handle complex algorithms exponentially faster. However, the threat they pose to encryption and security is also a concern.


  • i had a tab s2 with 28nm soc, 3gig lpddr3 ram running android 7 and the app quickly crashes, now i use an infinix (random asian brand, but way better than the rest) with a 12nm soc 4 gig of ram (4 doesnt seem enough) and 10w slow charging, the next phone i consider buying would run android 13 with 16 gig of ram (swap included, well, unlocked at least), also includes 33w fast charging and a 6nm soc, kinda finding myself reasons to upgrade, but as u said i am waiting this out too: i wanted to run the app on my pc but also i would have to upgrade my laptop ram so my os could support the android emulator without hiccups. i considered running my own instance for this sake but kinda extreme…lemmy is addictive not gonna lie


  • I use Connect: i filter alot of undesired coms and instances so i could browse only interesting stuff. I filter alot that i get a runtime error, due to server overwhelmed with requests. Connect Dev said that the app will be optimized in a way that u could browse infinitly without compromising said filters or encountring such error. I feel my only bottleneck now is android ram, since browsing down the feed causes the phone to become sluggish and cause other background run apps to crash: probably considering getting a phone with extra ram (8gig with swap enabled? )