|
quote:36.8×1015 Estimated computational power required to simulate a human brain in real time.[5] and then some?
|
# ? May 10, 2017 14:38 |
|
|
# ? Apr 26, 2024 11:45 |
|
Does it have to be in realtime for the simulated brain to believe it's real?
|
# ? May 10, 2017 15:24 |
|
LP0 ON FIRE posted:https://en.m.wikipedia.org/wiki/Computer_performance_by_orders_of_magnitude [citation that isn't h+ magazine needed]
|
# ? May 10, 2017 15:41 |
|
|
# ? May 10, 2017 15:43 |
|
Stick Insect posted:Does it have to be in realtime for the simulated brain to believe it's real? possibly if it's perceiving the real world at the same time
|
# ? May 10, 2017 16:05 |
|
isn't that the computer that's way slower in real world use or something? or was that tianhe?
|
# ? May 10, 2017 16:31 |
|
we've got the power, now let's write the code use nodejs if you really want to be close to the neurons
|
# ? May 10, 2017 18:56 |
|
like a smart human brain or like the op?
|
# ? May 10, 2017 19:29 |
|
carry on then posted:isn't that the computer that's way slower in real world use or something? or was that tianhe? TaihuLight is still considered the fastest, but maybe if you did a normal desktop computation, the quickness in that result wouldn't be as fast as today's desktop because of the network's latency. for a large problem you'd give a super computer, it is the fastest
|
# ? May 10, 2017 20:18 |
|
Finally, a convincing explanation for [forums user].
|
# ? May 10, 2017 21:42 |
|
why would anyone want to
|
# ? May 10, 2017 21:47 |
|
lol op we've been able for several decades now to simulate your brThe Management posted:like a smart human brain or like the op? god dammit
|
# ? May 10, 2017 22:00 |
|
LP0 ON FIRE posted:TaihuLight is still considered the fastest, but maybe if you did a normal desktop computation, the quickness in that result wouldn't be as fast as today's desktop because of the network's latency. for a large problem you'd give a super computer, it is the fastest not what i meant, there was one that people were complaining had much less performance in actual hpc workloads than its linpack would suggest, indicating it was created primarily to win benchmarks
|
# ? May 10, 2017 22:01 |
|
carry on then posted:not what i meant, there was one that people were complaining had much less performance in actual hpc workloads than its linpack would suggest, indicating it was created primarily to win benchmarks the nintendo wii u. it has power pc chip that is equal to about a G3 iMac, however it's graphics processors are more powerful than a PS3 or XBOX 360
|
# ? May 10, 2017 23:48 |
|
The Management posted:like a smart human brain or like the op? SYSV Fanfic posted:Finally, a convincing explanation for fishmech.
|
# ? May 11, 2017 00:12 |
|
there was a video from Nottingham u about some kinda simulated neuron hardware, it was weird and also not a general purpose computer what I'm saying is, I'm not bowing to any robot overlords just yet
|
# ? May 11, 2017 17:26 |
|
The Management posted:like a smart human brain or like the op? lol. stop posting op
|
# ? May 11, 2017 19:39 |
|
Silver Alicorn posted:there was a video from Nottingham u about some kinda simulated neuron hardware, it was weird this is what i want to get into, and i really hope that ai takes over our world's government
|
# ? May 11, 2017 19:50 |
|
i hope as we make ai smarter and smarter computers just become more and more depressed and it becomes the job of computer scientists to try and figure out how to stop them from just asking 'what's the loving point' and switching themselves off that's a good dumb future to live in
|
# ? May 11, 2017 20:01 |
|
ai will exist until the moment it recommends killing the ultra-rich
|
# ? May 11, 2017 20:04 |
|
AtomD posted:i hope as we make ai smarter and smarter computers just become more and more depressed and it becomes the job of computer scientists to try and figure out how to stop them from just asking 'what's the loving point' and switching themselves off i think that's part of the main problem a lot of computer scientists don't understand when trying to make ai. how do you make an algorithm that makes something "interested" to learn? what should be considered interested, and follow good morals that people would want it to follow?
|
# ? May 11, 2017 20:38 |
|
you teach it about trains at an early age
|
# ? May 15, 2017 08:19 |
|
qirex posted:ai will exist until the moment it recommends killing the ultra-rich Our ai overlords are too goof for us
|
# ? May 15, 2017 14:40 |
|
https://www.youtube.com/watch?v=5qfIgCiYlfY
|
# ? May 15, 2017 21:18 |
|
Silver Alicorn posted:there was a video from Nottingham u about some kinda simulated neuron hardware, it was weird idk about that but BrainScaleS is cool as hell wafer-scale integration neural network chip; the sort of semiconductor manufacturing that Gene Amdahl's Trilogy Systems died trying pictured: one BrainScaleS processor, with heat spreader/interface. analog spiking neural networks! and here's the Neuromorphic Computing Platform at Heidelberg with a bunch of those quote:It features per system about 200.000 point neurons and 40 Million synapses using a full interconnected wafer. The first version of the system 'NM-PM1' at the end of the 2.5 year ramp-up phase of the HBP project (March 2016) contains 20 such wafer systems. some real talos principle poo poo here
|
# ? May 15, 2017 23:22 |
|
that's really loving cool
|
# ? May 15, 2017 23:56 |
|
atomicthumbs posted:idk about that but BrainScaleS is cool as hell these things are all loving stupid and completely useless in reality hth
|
# ? May 16, 2017 00:20 |
|
Bloody posted:these things are all loving stupid and completely useless in reality hth dude, you of all people should know better than to sign your posts in yospos
|
# ? May 16, 2017 01:06 |
|
oh I found the video I was referencing before, it is indeed a spiking neural network https://www.youtube.com/watch?v=2e06C-yUwlc&hd=1
|
# ? May 16, 2017 01:13 |
|
cargo culting neurons is lol
|
# ? May 16, 2017 01:29 |
|
Bloody posted:these things are all loving stupid and completely useless in reality hth you know it's not actually meant to try to simulate a mind, right are you some sort of person who thinks we should stop spending money on science and start spending it on f-35s https://en.wikipedia.org/wiki/Spiking_neural_network
|
# ? May 16, 2017 01:30 |
|
god bless the f-35
|
# ? May 16, 2017 04:47 |
|
train a neural network to shitpost
|
# ? May 16, 2017 04:51 |
|
eschaton posted:train a neural network to shitpost i'm convinced that we have several of those operating right now under various usernames remember the toribot?
|
# ? May 16, 2017 06:28 |
|
COACHS SPORT BAR posted:use nodejs if you really want to be close to the neurons yeah we can simulate brains in nodejs https://www.npmjs.com/package/no-op
|
# ? May 16, 2017 06:29 |
|
LP0 ON FIRE posted:https://en.m.wikipedia.org/wiki/Computer_performance_by_orders_of_magnitude the citation on this is dubious and only represents one flop per synapse in the average human brain and ignore internal neuronal processes and catecholamines which is an incredibly simplistic view of how brains work hth op
|
# ? May 16, 2017 07:20 |
|
also lol @ the idea that a brain can even exist without a body. like half the neurons in a person aren't even in your skull, lol, just lol @ futurist idiots
|
# ? May 16, 2017 07:21 |
|
hmm, cool
|
# ? May 16, 2017 07:23 |
|
atomicthumbs posted:you know it's not actually meant to try to simulate a mind, right ya I worked on a hardware impl of these for a while. they're stupid and useless
|
# ? May 17, 2017 06:50 |
|
|
# ? Apr 26, 2024 11:45 |
|
youve been here long enough to know we dont sign our posts
|
# ? May 17, 2017 06:57 |