Matches for ninjashogun, 1788 total results Sorted by newest | relevance
Sun Mar 23 18:27:21 UTC 2014 <ninjashogun> someone should set up an FDI btc - Federal Deposit Insurance of Bitcoins
Sun Mar 23 18:24:47 UTC 2014 <Apocalyptic> ninjashogun, already been posted hours ago
Sun Mar 23 18:02:27 UTC 2014 <ninjashogun> benkay, that is a hilarious observation
Sun Mar 23 18:02:20 UTC 2014 <ninjashogun> benkay, ha!
Sun Mar 23 17:57:41 UTC 2014 <ninjashogun> but we have a theoretical proof that the architecture of the brain is somehow encoded into under 700 MB. (well under that, as that encodes the full human body.)
Sun Mar 23 17:57:11 UTC 2014 <ninjashogun> Of course we have no chance in hell of modelling the development of a brain by "running" the code. We can't even model a few molecules fully.
Sun Mar 23 17:56:20 UTC 2014 <ninjashogun> Also don't forget that we know for a fact that the architecture in the brain is encoded in no more than 700 MB totally uncompressed. That is how much the human genome takes to encode as simple bits (adenine(A), cytosine (C), guanine (G) and thymine (T) encoded as two bits each). With some compression that's even less.
Sun Mar 23 17:54:59 UTC 2014 <ninjashogun> Hardware-wise, if we had the code than a simple cluster of a few tons of CPU and RAM units could probably run it.
Sun Mar 23 17:54:08 UTC 2014 <ninjashogun> Based on the simple calculation that the brain is just 3 pounds, 1.25 liters, and fires at 200 herz, it is extremely likely that there exists ? (latex \exists) code that we could run that would result in a brain running at an appreciable percentage of real-time brain speed, using < $1B of hardware. But we don't have the code, even though it probably \exists
Sun Mar 23 17:52:02 UTC 2014 <ninjashogun> midnightmagic, it's an interesting situation . Rather than being limited by hardware ,we're very possibly limited by software. We don't know what to run - it's not that we can't run it, but we don't have the software written.
Sun Mar 23 17:51:24 UTC 2014 <ninjashogun> midnightmagic, well I agree. Our current understanding isn't even enough for a full model. we wouldn't know what to run even if we had a supercomputer with unlimited resources. we don't have the code.
Sun Mar 23 17:50:28 UTC 2014 <ninjashogun> asciilifeform, exactly, they're not in sync because why would they be. But is that out of phase effect used to do calculation or the equivalent of a pipeline, etc? Probably not.
Sun Mar 23 17:49:42 UTC 2014 <midnightmagic> ninjashogun: Nope. There is not enough understanding to model even small parts of the human brain. The structures we know about are unknown. There are likely other structures we don't yet know about.
Sun Mar 23 17:49:35 UTC 2014 <ninjashogun> asciilifeform, I would think that the brain doesn't depend on sensitive timing information like that, but rather on the pathways that are activated (irrespective of timing).
Sun Mar 23 17:49:09 UTC 2014 <ninjashogun> asciilifeform, I understand. Do you think that timing is used by the brain in this way?
Sun Mar 23 17:48:53 UTC 2014 <ninjashogun> asciilifeform, that is a very good and interesting example. Do you think that the brain relies on these sampling effects?
Sun Mar 23 17:48:01 UTC 2014 <ninjashogun> midnightmagic, this is very true. But, for example, if a supercomputer were to model these interconnects but had fewer "physical" neurons (CPU units), then the same CPU units could be reused 20 million times (4 ghz / 200 herz) to get realtime speed. (This is just an example of how to think about it.)
Sun Mar 23 17:46:13 UTC 2014 <ninjashogun> midnightmagic, the other interesting thing is that the brain weighs just 3 pounds and takes up 1.26 liters. (1400 grams, 1.3 US quarts).
Sun Mar 23 17:44:28 UTC 2014 <ninjashogun> hi, midnightmagic. Well, yes and no. It is meaningless, you are right. But it's also interesting that it means that a supercomputer running at 4 ghz interconnect speed runs at a native speed of 10,000,000 times faster than neurons' native speeds.