Donate Bitcoin

Donate Paypal


PeakOil is You

PeakOil is You

Peak Compute

Discussions of conventional and alternative energy production technologies.

Re: Peak Compute

Unread postby davep » Fri 24 Jul 2015, 03:51:54

I am still waiting for you techtopians to...


Which techtopians are these? Or are you building a nice straw man to knock down?
What we think, we become.
User avatar
davep
Senior Moderator
Senior Moderator
 
Posts: 4578
Joined: Wed 21 Jun 2006, 03:00:00
Location: Europe

Re: Peak Compute

Unread postby Outcast_Searcher » Fri 24 Jul 2015, 13:09:36

ennui2 wrote:
pstarr wrote:I am still waiting for you techtopians to project implementation of true natural language processing. Come back to me when Siri can parse this simple string; Time flies like an arrow. into it various permutations.

Computers stink at Go, but are incredibly brilliant at chess. (Go is MUCH harder to describe in terms of math, at least thus far).

Computers have the common sense of a brick, yet Watson winning against humans champions at Jeopardy is a pretty amazing practical natural language processing feat.

Just because computers are only good at narrow domains (and likely will be for the forseeable future) doesn't mean they can't be VERY useful within those domains.

Just as a hammer wouldn't be too good at, say, dusting doesn't mean we should never use a hammer.

To close your eyes and pretend that "computers don't matter" because they can't, say, compose poetry as sensitively as a human is as delusional as those who pretend that the earth has no resource limits since we're not dead yet.

And I'm no techtopian -- I just have some insight through experience into what computers can do as far as solving complex problems. For example, having written a chess program that in 1981 played FAR more like a human (in terms of the types of moves it made and games it played) than any program I saw or read about for a good 10-12 years after that, I am well aware of the strictly limited domain of the game of chess. Thus, I believe the people that believe self-driving cars will be city-street ready in 5 years are highly delusional. However, acting like this means self driving cars will never drive better than typical humans, despite the complexity of the problem, are kidding themselves.
Given the track record of the perma-doomer blogs, I wouldn't bet a fast crash doomer's money on their predictions.
User avatar
Outcast_Searcher
COB
COB
 
Posts: 10142
Joined: Sat 27 Jun 2009, 21:26:42
Location: Central KY

Re: Peak Compute

Unread postby ennui2 » Fri 24 Jul 2015, 13:45:54

pstarr wrote:I never said computers don't matter.


Pstarr, the problem I (and maybe a few others) have is that you seem to want to generally mock, devalue, or degrade technological progress, like the previous thread on Tesla EVs where you were busy trying to killjoy it as much as possible. When you paint with such a broad brush, you just come across like a troll.

It's possible to appreciate technological achievement while still realizing the unintended consequences (like mountains of eWaste, or overpopulation from the green revolution). Don't assume that I or other people here deny or are ignorant of those unintended consequences. We're not babes in the woods. We also realize that under the right combination of circumstances that technology can reach diminishing returns and degrade. But it's not going to necessarily flat-line how and when you'd like it to. There are still some interesting things going on. The vast majority of Graeme's revolutionary-sounding breakthroughs won't amount to anything, but a few probably will.
"If the oil price crosses above the Etp maximum oil price curve within the next month, I will leave the forum." --SumYunGai (9/21/2016)
User avatar
ennui2
Permanently Banned
 
Posts: 3920
Joined: Tue 20 Sep 2011, 10:37:02
Location: Not on Homeworld

Re: Peak Compute

Unread postby davep » Fri 24 Jul 2015, 14:39:16

pstarr wrote:
ennui2 wrote:But it's not going to necessarily flat-line how and when you'd like it to. There are still some interesting things going on. The vast majority of Graeme's revolutionary-sounding breakthroughs won't amount to anything, but a few probably will.
then we agree. :)

And as long as you are willing denigrate Graeme's feel-good techtopian trolling as actively as I do. It is a distraction and tends to dominate any thread by pushing down serious discussion, and off the page where new folks will never see the serious discussion. Get to work. Then we can be friends. :)


So long as you see Graeme's contributions as a "Good News!" aggregator I don't see much of a problem. It's nice to have that news, even if most of it is pretty silly for a lot of us here. Continuously arguing with feel-good news snippets is a futile task, IMO. That doesn't mean we tacitly agree though.
What we think, we become.
User avatar
davep
Senior Moderator
Senior Moderator
 
Posts: 4578
Joined: Wed 21 Jun 2006, 03:00:00
Location: Europe

Re: Peak Compute

Unread postby lpetrich » Mon 13 Aug 2018, 12:53:48

It looks like we have reached the limit of the feasible size of an integrated-circuit component. This means that computer-chip design will have to expand in other directions. Instead of vertical expansion (faster, smaller components), we will have lots of horizontal expansion (more components).

This produces a serious programming challenge. With more components per chip and individual processors having the same number of components, a logical route of expansion is more individual processors. This means that one can do more parallel processing per chip.

Each individual processor is a single-instruction-stream, single-data-stream (SISD) system, the sort of thing that one learns early in a computer-architecture class.

For multiple processors, one can make them act like separate individual ones, making a multiple-instruction-stream, multiple-data-stream (MIMD) system. This is the most flexible approach, but also the most complicated one.

One can use the shortcut of having several processors do the same instructions, while operating on different sets of data. This gives a single-instruction-stream, multiple-data-stream (SIMD) system.

In practice, the two are often combined.

Intel has supported SIMD in its x86 chips since the mid-1990's: MMX, SSE, AVX. These chips issue SIMD instructions alongside their SISD ones. This is also done in some other chip architectures, like POWER/PowerPC, Sparc, MIPS, ARM, and PA-RISC.

Multicore chips are essentially MIMD, with each core being SISD though often with SIMD as explained above.

GPU's often use SIMD, for reasons that ought to be obvious -- doing the same thing to lots of pixels.
User avatar
lpetrich
Lignite
Lignite
 
Posts: 365
Joined: Thu 22 Jun 2006, 03:00:00

Re: Peak Compute

Unread postby lpetrich » Mon 13 Aug 2018, 14:48:20

SIMD and MIMD architectures have lots of programming challenges.

MIMD is the easiest to program for. Programming for it is an extension of programming for multitasking and multithreading in general, programming that has been done for some decades now, long before MIMD systems became common. It is done because it is a great convenience for many applications, like supporting multiple users and multiple active apps and multiple Internet-client requests. Going from SISD to MIMD only requires some changes to OS kernels, for distributing tasks and threads among several CPU cores.

SIMD is more difficult to program for. It is designed for problems where one does the same calculations on different sets of data, like rendering operations for different pixels in an image. Arithmetic and logical operations are straightforward for it, but its big difficulty is flow of control. For if-then-else statements, a common approach is to calculate both branches for each stream of data, then calculate a condition flag for each stream of data, then select which of the branches for each stream using its condition-flag value. Loops are even worse. Fixed numbers of loops are no problem, but if one wants to exit from a loop, there is the problem that different streams may want to exit after different numbers of loopings.

So in practice, SIMD processing is implemented as "slaves" of SISD or MIMD "masters".

I think that we will be seeing more of SIMD processing, since that is a good way of doing some AI operations, like artificial-vision ones.
User avatar
lpetrich
Lignite
Lignite
 
Posts: 365
Joined: Thu 22 Jun 2006, 03:00:00

Re: Peak Compute

Unread postby aadbrd » Thu 06 May 2021, 22:28:17

Sorry for the thread bump but this was timely.

https://www.bbc.com/news/technology-57009930
User avatar
aadbrd
Permanently Banned
 
Posts: 510
Joined: Sat 26 Dec 2020, 16:09:06

Re: Peak Compute

Unread postby Outcast_Searcher » Sat 15 May 2021, 17:12:28

aadbrd wrote:Sorry for the thread bump but this was timely.

https://www.bbc.com/news/technology-57009930

VERY nice. Yet, in fact of very successful parallel computing, vs. all the wrong claims of the past, it's not like this is even needed.

Outfits like Nvidia continue to make VERY meaningful progress in powerful chips. And I don't care whether it's nanometer process size for chips or parallelism or both.
Given the track record of the perma-doomer blogs, I wouldn't bet a fast crash doomer's money on their predictions.
User avatar
Outcast_Searcher
COB
COB
 
Posts: 10142
Joined: Sat 27 Jun 2009, 21:26:42
Location: Central KY

Re: Peak Compute

Unread postby mousepad » Sat 15 May 2021, 17:40:20

Outcast_Searcher wrote:Outfits like Nvidia continue to make VERY meaningful progress

Not at all. They make progress. But is it meaningful?

What is the goal? What should be achieved? Happiness?
I was no less happy 40 years ago then I am now. I was able to laugh love and be merry. All this technology did not increase my quality of life one bit. Yes I'm much richer now, stinking rich compared to 40 years ago. But happier? Not one bit. Are you?
mousepad
Tar Sands
Tar Sands
 
Posts: 799
Joined: Thu 26 Sep 2019, 09:07:56

Re: Peak Compute

Unread postby AdamB » Sat 15 May 2021, 20:56:29

mousepad wrote: Yes I'm much richer now, stinking rich compared to 40 years ago. But happier? Not one bit. Are you?


Sure, based exactly on your time frame. I was young then, poor, single, and stuck in Appalachia. "Stuck" being the part that most represents "unhappiness".

Cured all 4 of those conditions in 40 years. :)
Plant Thu 27 Jul 2023 "Personally I think the IEA is exactly right when they predict peak oil in the 2020s, especially because it matches my own predictions."

Plant Wed 11 Apr 2007 "I think Deffeyes might have nailed it, and we are just past the overall peak in oil production. (Thanksgiving 2005)"
User avatar
AdamB
Volunteer
Volunteer
 
Posts: 9292
Joined: Mon 28 Dec 2015, 17:10:26

Re: Peak Compute

Unread postby mousepad » Sun 16 May 2021, 06:18:41

AdamB wrote:
mousepad wrote: Yes I'm much richer now, stinking rich compared to 40 years ago. But happier? Not one bit. Are you?


Sure, based exactly on your time frame. I was young then, poor, single, and stuck in Appalachia. "Stuck" being the part that most represents "unhappiness".

Cured all 4 of those conditions in 40 years. :)


And all because of technologies progress? Or because you buckled down, did the work, studied, had a goal and did what was needed? What do you think?
mousepad
Tar Sands
Tar Sands
 
Posts: 799
Joined: Thu 26 Sep 2019, 09:07:56

Re: Peak Compute

Unread postby AdamB » Sun 16 May 2021, 14:13:52

mousepad wrote:
AdamB wrote:
mousepad wrote: Yes I'm much richer now, stinking rich compared to 40 years ago. But happier? Not one bit. Are you?


Sure, based exactly on your time frame. I was young then, poor, single, and stuck in Appalachia. "Stuck" being the part that most represents "unhappiness".

Cured all 4 of those conditions in 40 years. :)


And all because of technologies progress? Or because you buckled down, did the work, studied, had a goal and did what was needed? What do you think?


Studied, did work, did harder work, found a good woman, moved, etc etc. Can't say technology was involved much.
Plant Thu 27 Jul 2023 "Personally I think the IEA is exactly right when they predict peak oil in the 2020s, especially because it matches my own predictions."

Plant Wed 11 Apr 2007 "I think Deffeyes might have nailed it, and we are just past the overall peak in oil production. (Thanksgiving 2005)"
User avatar
AdamB
Volunteer
Volunteer
 
Posts: 9292
Joined: Mon 28 Dec 2015, 17:10:26

Re: Peak Compute

Unread postby Outcast_Searcher » Sun 16 May 2021, 20:30:20

mousepad wrote:
Outcast_Searcher wrote:Outfits like Nvidia continue to make VERY meaningful progress

Not at all. They make progress. But is it meaningful?

What is the goal? What should be achieved? Happiness?
I was no less happy 40 years ago then I am now. I was able to laugh love and be merry. All this technology did not increase my quality of life one bit. Yes I'm much richer now, stinking rich compared to 40 years ago. But happier? Not one bit. Are you?

If you can't deal with the word meaningful, in context, like an adult, then try "significant".

And, who said better chips would make YOU happy? :?

And yet at the end of the day, computers and chips continue to get MUCH more powerful and cheaper on an objective scale (price vs. compute cycles), whether you point at rainbows or not.
Given the track record of the perma-doomer blogs, I wouldn't bet a fast crash doomer's money on their predictions.
User avatar
Outcast_Searcher
COB
COB
 
Posts: 10142
Joined: Sat 27 Jun 2009, 21:26:42
Location: Central KY

Re: Peak Compute

Unread postby aadbrd » Sun 16 May 2021, 20:54:10

mousepad wrote:They make progress. But is it meaningful?


I don't think the central thesis of the OP is about that. It's simply about whether computing power has peaked. It started flattening out but did not peak. How increased computing power impacts society is hard to discern. More power, at the very least, translates into efficiency, doing the same amount of work for less energy input. That is a positive in my book.
User avatar
aadbrd
Permanently Banned
 
Posts: 510
Joined: Sat 26 Dec 2020, 16:09:06

Re: Peak Compute

Unread postby jedrider » Sun 16 May 2021, 22:42:17

aadbrd wrote:
mousepad wrote:They make progress. But is it meaningful?


I don't think the central thesis of the OP is about that. It's simply about whether computing power has peaked. It started flattening out but did not peak. How increased computing power impacts society is hard to discern. More power, at the very least, translates into efficiency, doing the same amount of work for less energy input. That is a positive in my book.


That's why virtual currencies were created: We peaked on useful networked compute power.
User avatar
jedrider
Intermediate Crude
Intermediate Crude
 
Posts: 3106
Joined: Thu 28 May 2009, 10:10:44

PreviousNext

Return to Energy Technology

Who is online

Users browsing this forum: No registered users and 37 guests