Jump to content

Bananasplit_00

Member
  • Posts

    18,032
  • Joined

  • Last visited

Posts posted by Bananasplit_00

  1. 8 hours ago, hishnash said:

    In the end Siri like workloads are best done on device given your want low latency and you have so many possible devices that could be making requests that doing this server side would just cost a small fortune.   Just piping Siri to a cloud based LLM would have so many downsides from a load perspective since the Siri load is likly has lots of spikes throughout the day, and you cant pipe requests half way round the world to have a single server handle that load due to massive added latency so you would end up with a lot of compute spread out around the world with much of it only be used at peak times for that timezone. 

    Being able to do as much as possible locally on the device, even possibly letting that on device model figure out how to query remove data sources (even remote LLMs) would filter out most of this load... it would be a huge wast of $ to have an huge cloud based LLM handle people adding groceries to thier shopping list, or tuning on and off lights in the house (this sort of things makes up most of the Siri requests)... people are not asking Siri to write them a 100 page essay. 

    "Hey Siri, turn off the lights"

     

    *phone catches fire*

  2. 17 hours ago, PacketAuditor said:

    The UIDs and keys can all be updated! And if EMV gets approved, any payment cards will be tokenized much like Google/Apple pay, and are objectively better than a traditional card with just static data.

    Neat, then it's mostly my phobia for needles that's getting in the way of me getting one 😂

    17 hours ago, PacketAuditor said:

    Haha, there actually are some people who install magnets in their tragus, the little flap in front of the ear. While they can't hear the radio you can hear power supplies, chargers, power stations, etc.

    Personally I like fingertip sensing magnets and think they're a little more useful.

    That sounds awful, I can already hear most powersupplies coil whine along with tinnitus 💀

     

    Got pretty close to doing the small magnet in my finger some years back but the primary source of the magnets was ebay and I wasn't old enough to drink so I'd have absolutely nothing decent for painkillers 😂

     

    Might look into it again, in a not potato and scalpel in a shed way this time

  3. Those are some ridiculously low minimum specs if you have 3k to spend 😄

     

    I'd first look for battery life and RAM. At the very least 16GB, I'd probably go for 32GB.

     

    Next up is a nice screen and keyboard. Definitely going to be a really nice thing to have seeing as you are likely going to be using this laptop for most of the day almost every day for the next 3-5 years.

     

    My recommendation is to go for something fairly slim and 14" or below, I am extremely happy to not be lugging around a thick and heavy 15.6" during my days.

     

    The graphics might be tricky, seems like they want you to have a dedicated GPU, any discrete GPU you can find in a laptop from the last five years or so will probably be good enough.

     

    Buying used can get you a heck of a lot more value too. I got my surface laptop 2 for €300 when my pro 4 died, you have about 6x that budget so you should not have a hard time finding something suitable in your price range.

  4. If you really want to know why, id say you should become and electrical engineer with a focus on semiconductor design for integrated circuits. Some useful topics to look into if you want a real answer are digital design and semiconductor design for integrated circuits. There are way too many for me to explain and its not like i know every limitation either but a few of them are:

    Gate charge, the controll terminal of a MOSFET acts as a capacitor that needs to be charged before it can effectively switch state

    Gate oxide thickness, thicker gate oxide means the controll terminal can handle more voltage which can speed up switching

    Critical Path delay, how fast the slowest thing you are trying to do with your transistors is

    C2Q delay, how fast your memory elements can ready the next output(kinda, not really)

    Package power dissipation, if its not enough the silicon structures will break

     

    These are VERY simple explenations of a few of the limitations that came to my mind, its not as simple as "make it cold -> more go fast"

     

     

×