Reality Check
Penultimate Amazing
I klnow that you linked to a IBM Research Report from 2012
Read my post - I know that:
Super Artificial Intelligence, a naive approach
- Thread starter ProgrammingGodJordan
- Start date
Reality Check
Penultimate Amazing
Later you correct the link to an IBM Research Report from 2012 (PDF)
That PDF is at Dharmendra S Modha's Brain-inspired Computing Blog who is an IBM researcher.
His official site at IBM is Dharmendra S. Modha
Reality Check
Penultimate Amazing
High school students learn about dimensional analysis which even works outside of science!
Go to a hardware store and ask for 10 feet of pipe and get given 10 cm of pipe - would you pay the price of 10 feet of pipe?
Get stopped for going 40 mph in a 30 mph - pointing out that your car is less than 30 miles long will not get you out of a ticket
!Synaptic operations per second are not synapses so the two quantities cannot be directly compared.
ProgrammingGodJordan
Banned
An average of two values is the last value - the first value / 2
10^18 - 10^16 / 2 = 4.95 x 10^17. That is roughly 10^17.
The human brain computes on average at 4.95 x 10^17 synaptic operations per second
There are lower estimations, 10^14 to 10^15 synapses.
ProgrammingGodJordan
Banned
Irrelevant, I simply provided a direct link.
The initial link still provides an overview while containing a link to that same direct link.
ProgrammingGodJordan
Banned
High school students learn about dimensional analysis which even works outside of science!
Go to a hardware store and ask for 10 feet of pipe and get given 10 cm of pipe - would you pay the price of 10 feet of pipe?
Get stopped for going 40 mph in a 30 mph - pointing out that your car is less than 30 miles long will not get you out of a ticket
Synaptic operations per second are not synapses so the two quantities cannot be directly compared.
(A)
I had long mentioned, (and also such is the case whether or not I mention) that sops are reducible to synapses.
Your initial statement was that synapses couldn't be compared to any speed. (your quote).
That was shown to be a nonsensical statement, as sops may be reduced to synapses.
(B)
Your dimensionality reference is irrelevant.
I have not violated any laws in expressing that 10^15 synapses is rough for 10^16 sops. (These are within the same range, especially when I repeat many many many many times, that sops are reducible to synapses, and did calculations in the same forms)
For example, I didn't apply 10^x synapses for one side while 10^x sops for another side of moore's law equation, I applied synapse values, for both sides, as seen in early pages in this thread.
(C)
ProgrammingGodJordan said:
Otherwise you are merely "flogging a dead horse", as your prior statements have been shown to fail.
Last edited:
Reality Check
Penultimate Amazing
You mentioned this many times but it is not in the OP and I do not recall the post where you cite or calculate it.
Please link to the post where you cite or calculate the synapses to synaptic operations per second conversion factor for the computer simulation in IBM Research Report from 2012 (PDF).
I do hope this is not the ignorant act taking the rate at which human synapses fire and multiplying applying this to a computer simulation.
ProgrammingGodJordan
Banned
(1)
The synapse values (for 2020 calculation) were already given.
If one looks for a few seconds, one would probably notice that 10^14 synapses for machine from IBM Research Report from 2012 (PDF), and 10^15 synapses for human were but already available...
I did not convert to synapses, as they were already provided.
(2)
The conversion made was from 10^15 synapses for human (already given) to 10^16... sops range mentioned in the original post.
Last edited:
Reality Check
Penultimate Amazing
That is very wrong.
It is not "dimensionality". It is dimensional analysis
We are talking about the difference between one quantity and another quantity with different units. dimensional analysis instantly tells us that they cannot be compared unless we use a valid conversion factor
!Reality Check
Penultimate Amazing
Relevant to your ability to write factual statements
The OP misleadingly linked to a "Brain Power" article. An factual statement would be "search this web page for 10^14". A complete, competent citation would be to the PDF and to that article.
Stating that a persona blog is an IBM web site is at least ignorant about your source.
Reality Check
Penultimate Amazing
Repeating the obvious yet again
!However if you took a few seconds to read your sources (or my posts) then you would know that IBM Research Report from 2012 (PDF) also quotes and cites human synapse counts (~10^14)
!You started with "The human brain computes roughly 10^16 to 10^18 synaptic operations per second". To compare this with computers you need to convert 10^14 computer synapses using computer operations per second per computer synapse. We will then be able to compare computer sops to human sops.
Last edited:
Reality Check
Penultimate Amazing
Repeating the obvious yet again
! I have read the OP. I have read the posts in the thread. I know where you got the OP values from.
Last edited:
John Jones
Penultimate Amazing
Did you forget to abuse the term 'betwixt' again?
ProgrammingGodJordan
Banned
That is very wrong.
It is not "dimensionality". It is dimensional analysis
We are talking about the difference between one quantity and another quantity with different units. dimensional analysis instantly tells us that they cannot be compared unless we use a valid conversion factor
You tend to express words that show shallow mindedness.
(1)
Dimensionality deals with units, as does dimension.
(2)
Once more, I had long mentioned that sops were reducible to synapses.
Your initial comment was that sops/synapses couldn't be compared, which remains silly.
Last edited:
ProgrammingGodJordan
Banned
It still remains that the source, was contained in the relevant url, blog or not.
ProgrammingGodJordan
Banned
Repeating the obvious yet again
However if you took a few seconds to read your sources (or my posts) then you would know that IBM Research Report from 2012 (PDF) also quotes and cites human synapse counts (~10^14)
You started with "The human brain computes roughly 10^16 to 10^18 synaptic operations per second". To compare this with computers you need to convert 10^14 computer synapses using computer operations per second per computer synapse. We will then be able to compare computer sops to human sops.
(1)
Once more, for the calculation of 2020 value, 10^14 synapses for machine (IBM), was used with 10^15 synapses for human (wiki, etc). All already existing, and not needing conversion.
(2)
No comparison needed to be made with 10^16 to 10^18 sops for humans directly, as the above 10^15 synapses for human value was ALREADY AVAILABLE.
Footnote:
You may have noticed the 2020 value. That value emerged from the roughly 10^15 synapses for humans, and roughly 10^14 synapses for machines.
So, I started with 10^15 synapses, as indicated by 2020 calculation and sources.
Last edited:
ProgrammingGodJordan
Banned
Merely an example.
There are other neuromorphic chips.
The tianhe-3 may not be as efficient as these small/low in power consumption as these neuromorphic chips.
For example, if I recall correctly, simulating 240 trillion synapses, ibm synapse chip consumes merely 82.19 kW, (compared to human consumption (~20 W), but still x1542 slower than human brain), while supercomputers like k computer, in a similar year range, simulated a small 10 trillion synapses, at 9.9 MW.
Like k-computer, tiahne-2 ran in the mega watt range, 17.6 megawatts, although at the quadrillion range, above the trillion range of ibm.
This means that architectures like ibm are more efficient, especially the later versions at 70 milliwatts, capable of 46 billion synaptic operations per second.
Last edited:
So many words. So much tripe, because it is all based on this unsubstantiated guess. Until you can justify this figure (here's a tip: you can't), you make yourself look foolish every time you repeat it.
Repeating the obvious yet again
However if you took a few seconds to read your sources (or my posts) then you would know that IBM Research Report from 2012 (PDF) also quotes and cites human synapse counts (~10^14)
You started with "The human brain computes roughly 10^16 to 10^18 synaptic operations per second". To compare this with computers you need to convert 10^14 computer synapses using computer operations per second per computer synapse. We will then be able to compare computer sops to human sops.
....yes, not least because the simulation was "only" running at 1542 times slower than realtime
