|
Shifty Pony posted:likelihood of it happening is some crazy equation with a lot of exclamation marks in it that I don't feel like calculating out but let's just say that as you shrink things the effect isn't noticeable until it very suddenly is a huge problem. got me a new 5nm chip in my computer and 2+2 apparently equals “suck my cock jizz turd” alert the masses, we’ve been doing math wrong all these years
|
# ? Jul 9, 2018 15:36 |
|
|
# ? May 5, 2024 21:11 |
|
Shifty Pony posted:atoms jiggle when hot. sometimes they jiggle just right to trade places with the atom next to them. when your transistor is half the size of what it used to be the atoms that make it work only have to trade places half as far before they are someplace where they are going to gently caress things up. everybody loves jigglin
|
# ? Jul 9, 2018 16:58 |
|
old chips best chips I want a 68000-derived equivalent to today’s Intel Core i7
|
# ? Jul 10, 2018 09:33 |
|
or like a 65816 but running at hundreds of MHz to put in my Apple IIgs
|
# ? Jul 10, 2018 09:36 |
|
why can’t one just apply cooling and run a CPU super fast, assuming other components can be made to keep up what’s the physics behind needing a different implementation to hit higher clocks or is it really just a matter of practicality and susceptibility to error?
|
# ? Jul 10, 2018 09:38 |
|
eschaton posted:why can’t one just apply cooling and run a CPU super fast, assuming other components can be made to keep up essentially, when a gate switches on or off, the output doesn't transition instantly - it takes a (small) amount of time to transition to the point that the next transistor in line will see the change, if that causes a flip then it takes a little longer for the next one to see it, and so on. the fixed clock rate essentially allows the signal to propagate all the way through before moving on to the next stage. if you make the clock rate too fast, the signal can't propagate fast enough, and your cpu ends up doing totally random stuff and giving you back total junk. (this is why the p4 was such a piece of poo poo by the way - they designed it to have really fast clocks, which required making every stage really short so that the signal could go all the way through in time. but having lots of really short stages meant that if your branch predictor made a mistake, it had to throw away a ton of work and start over, instead of only having to throw away a little bit of work like in modern cpus)
|
# ? Jul 10, 2018 10:01 |
|
Midjack posted:use the turbo button judiciously. funny, urmom said the same thing last night
|
# ? Jul 10, 2018 11:36 |
|
eschaton posted:why can’t one just apply cooling and run a CPU super fast, assuming other components can be made to keep up you can dunk your cpu in liquid nitrogen and run it at its maximum rated frequency continuously. above that you end up with propagation delay issues as described in the post before this one.
|
# ? Jul 10, 2018 15:43 |
|
there's some research into superconducting logic that's been going on, which has absurdly fast switching speeds due to superconductors having zero resistivity. When you're on supercomputer scale you can get computing power / power consumption savings even taking into account the energy needed to cool the circuit into superconducting state
|
# ? Jul 10, 2018 16:34 |
|
speed of light is a bitch. same with discharging capacitors
|
# ? Jul 11, 2018 03:58 |
|
stick a 68000 soft core in any modern fpga and you can prob clock that poo poo high
|
# ? Jul 11, 2018 03:59 |
|
condense my posts and liquid cool your cpu with them
|
# ? Jul 11, 2018 04:43 |
|
Bloody posted:stick a 68000 soft core in any modern fpga and you can prob clock that poo poo high that’s what the Apollo Vampire accelerators for Amiga are
|
# ? Jul 11, 2018 05:33 |
|
eschaton posted:that’s what the Apollo Vampire accelerators for Amiga are "apollo vampire accelerator" is way too cool a name to be wasted on something as mundane as a faster processor for a dead computer system
|
# ? Jul 11, 2018 05:39 |
|
Bloody posted:stick a 68000 soft core in any modern fpga and you can prob clock that poo poo high once your core frequency passes a higher multiple of the bus than it was designed for then it is just going to spend those extra cycles blocked on data access.
|
# ? Jul 11, 2018 06:57 |
|
that’s why the accelerators generally have their own RAM (they also usually include some modern I/O but that’s more for convenience)
|
# ? Jul 11, 2018 07:37 |
Sagebrush posted:"apollo vampire accelerator" is way too cool a name to be wasted on something as mundane as a faster processor for a dead computer system what on earth is the market for people who want to play modern digital video on an Amiga?
|
|
# ? Jul 11, 2018 15:48 |
|
Shifty Pony posted:what on earth is the market for people who want to play modern digital video on an Amiga? Nerds
|
# ? Jul 12, 2018 19:59 |
|
feedmegin posted:Nerds dsyp
|
# ? Jul 12, 2018 23:22 |
|
eschaton posted:that’s what the Apollo Vampire accelerators for Amiga are man the amiga fan base is so weird in that it just will not ever ever die
|
# ? Aug 10, 2018 18:03 |
|
C.H.O.M.E. posted:man the amiga fan base is so weird in that it just will not ever ever die in 2099 they will still be trying to force a motherboard originally forged in early 1985 to interact with the holonet.
|
# ? Aug 10, 2018 20:03 |
|
Damnit, should have called this "Fab-gab: Chips 2HOT4TV (56k: )". Mods, plz do the needful.
|
# ? Aug 11, 2018 13:41 |
|
C.H.O.M.E. posted:man the amiga fan base is so weird in that it just will not ever ever die an Amiga is a Mac in a fursuit
|
# ? Aug 11, 2018 16:10 |
|
glofo just threw in the towel on 7nm. https://www.anandtech.com/show/13277/globalfoundries-stops-all-7nm-development
|
# ? Aug 27, 2018 22:00 |
|
are they really cutting another 5% in addition to the 5-10% that got let go a month or two ago
|
# ? Aug 27, 2018 22:17 |
|
countdown until tsmc announces a delay
|
# ? Aug 28, 2018 05:09 |
|
PyPy posted:Classic cool. drat I had one of those when I was a kid they were apparently introduced literally the month I was born
|
# ? Aug 28, 2018 11:15 |
|
C.H.O.M.E. posted:glofo just threw in the towel on 7nm. dragon enthusiast posted:are they really cutting another 5% in addition to the 5-10% that got let go a month or two ago i hope glofo burns. they owe me $20k.
|
# ? Aug 28, 2018 12:29 |
|
|
# ? May 5, 2024 21:11 |
|
Tokamak posted:countdown until tsmc announces a delay what i thought only intel was having any issues, don't stop now, pleasurekevin is almost there
|
# ? Aug 28, 2018 13:55 |