Slashdot Mirror
Intel Says Chips To Become Slower But More Energy Efficient (thestack.com)
An anonymous reader writes: William Holt, Executive Vice President and General Manager of Intel's Technology and Manufacturing Group, has said at a conference that chips will become slower after industry re-tools for new technologies such as spintronics and tunneling transistors. "The best pure technology improvements we can make will bring improvements in power consumption but will reduce speed." If true, it's not just the end of Moore's Law, but a rolling back of the progress it made over the last fifty years.
Hopefully if this does happen they will keep making the existing products, at least until they *do* manage performance improvements that catch up / exceed older stuff. Where I work we have lots of customers that *need* more processing power, and efficiency be damned.
William George
I can't imagine that there will simply be zero demand for fast, or faster, chips, regardless of the power efficiency. Some applications just demand it. If Intel won't do it, then someone else will, whether that's AMD or some new competitor in China or wherever.
On the other hand, there's certainly a market for more efficiency, especially in mobile devices, so I can certainly see lines of chips designed for that heading in the way described.
Why the Moore's law concern? I didn't get the impression that chips in general will get slower, just they're going to make more slow but efficient chips for specialized stuff while the chips WE use continue to, you know, get faster and eat more power.
Probably not. Going beyond 5Ghz limit has been a problem for the last decade or so. This is why we have multicore processors. It's easier to add more cores than go to plaid.
Optical
:T:R:A:N:S:
A flight from London to New York takes as long today as it did about 50 years ago. But the current planes achieve that more efficiently, with slightly larger windows, and some more pressure and humidity in the cabin. How depressing to think that the computing world might be about to enter a similarly dismal stage as well.
So the plan to make transistors tolerate higher clock speeds by using better materials is not going to happen?
Yet another restating of Moore's Law? The thing gets revised to whatever the latest growth area is.
The original 1965 article it was about "component counts", then it was revised in a later talk to be "circuit density", then revised in 1975 to be "semiconductor complexity", then revised in the later '70s to be "circuit and device cleverness", has been restated yet again when serial devices flatlined in favor of highly parallel chips.
Assuming this goes through the chipset, it will likely be restated again in terms of whatever other factor on the chips continues to grow.
//TODO: Think of witty sig statement
Intel's so far ahead of AMD, they have to roll back the clocks in order to stay competitive.
AMD isn't Intel's competition. Intel needs AMD to prevent Anti-Trust litigation. Intel's competition is ARM and all the OEM's who use ARM based chips. Especially if Microsoft ports full Windows 10 to the ARM. The big draw of ARM is performance/price per watt which is exactly what Intel is shooting for.
"A person is smart. People are dumb, panicky dangerous animals and you know it." - K
Lots of ways to get "speed."
I'm confused - don't Atom and Arm processors and the like already fill this need?
Or is this about significant reductions in power usage, with only minimal reductions in speed?
Gotta say, it would be nice if cutting edge, decently powerful processors no longer required active cooling though...
"Government is like fire; a handy servant, but a dangerous master." -- George Washington
"...If true, it's not just the end of Moore's Law, but a rolling back of the progress it made over the last fifty years."
While Moore's Law was fun to watch and experience while it lasted, this is a bit of a slap in the face when defining progress.
It's kind of like trying to define the transition from the gas-guzzling muscle car era to the fuel-efficient compact car era as rolling back progress.
Regardless of the finite resource, there's plenty of good reasons for humans to be consuming less of it.
Did your mother ever have you tested?
APK likes to ask for responses to the same things over and over. Maybe he just likes the responses?
As long as performance / watt increase can't one just use more of the slower chips?
About the optimal number of transistors in a SoC vs using many discrete components.
https://www.cs.utexas.edu/~fussell/courses/cs352h/papers/moore.pdf
see in particular the "bathtub graphs"
*** Suerte a todos y Feliz dia!
With AMD out of the way Intel can F*** us.
First they cut the pci-e lanes down on a $300-$350+ chip forcing you to pay upped to $350-$400 but then you need jump to $500-$600 to get the same as last gen + a small clock speed boost. This on the server / high workstation side.
On the desktop side they are still on DMI (Now at pci-e 3.0) + 16 PCI-e 3.0 why no QPI to chip set like AMD's HTX?
On the other hand, designs with less energy loss will open up the potential of higher speeds, once the techniques get refined.
One of the (many) limit issues with trying to force current CPUs faster is that the waste energy grows quickly as you increase switching frequency. Energy density becomes a significant problem, and manufacturers are not content with the idea of making all consumer devices use liquid-cooling and/or refrigeration techniques to prevent CPU melt. Take a couple years learning a more efficient set of components and tools, and the cap may raise past something (currently) silly like a passively cooled 8GHz chip.
They are so efficient, they actually generate power!
“He’s not deformed, he’s just drunk!”
Only for embarrassingly parallel workloads.
The mind control seems to be already working, as you are unable to press the Shift key on the keyboard.
Raw Clock speed has been meaningless for the last few chip generations. And actually dropped a few gens ago. All for the sake of efficiency. The as speed goes up power consumption goes up exponentially. But the same works in reverse, by lowering speed a little bit you get a huge savings in power, which allows you to do stuff like add additional cores. So while raw speed goes down total computing power goes up. This just seems like more of the same.
I'd imagine a terahertz chip would have ridiculous power consumption/heat problems. Perhaps they could build a simple design like a 6502 chip and see how that behaves when clocked up. Most of the lab stuff you read about (500ghz switching speeds) is super-cooled and unsuitable for desktop users.
Plus you'd need to use exotic materials like graphene and that would require a whole new industrial infrastructure for chip construction.
Realistically if Intel could push out super-fast chips now they could make lower-clocked/power CPUs for phones that would blow away current Arm devices and thus grab themselves many more billions in revenue.
That they haven't should tell you something.
Intel's been shooting itself in the foot with power vs performance for years. AMD was better, Intel reversed course and then beat AMD down. Now Intel's gunning for ARM because ARM is becoming a real threat to their core business. How many phones have Intel chips? How many tablets? Notebooks are moving towards ARM as well. Imagine an ARM based server farm. ARM is moving up the food chain into Intel's core business, and doing so with a class of processors Intel can't match.
The cesspool just got a check and balance.
because like omg mobile! (and server racks).
Efficiency is good, no doubt. But the electricity to run your computer, tablet, or phone, vs. the rest of your house, is comparatively very little. It's almost trivial even... except for those mobiles devices that are dependent on a battery. And the sloth and complacency of the battery manufacturers vs. the tech industry is what's holding us back. If they were investing into the R&D to keep up with Intel and Moore's law... doubling their capacity every 18 months as well... performance compromises like this would be unnecessary. Every laptop would be useful for a full business day, not just MacBook Airs and Chromebooks; and our iPhones and Androids would last 2 weeks to the charge like our old-school Nokias did.
Imagine all the people...
No, a lot of applications don't scale well across multiple cores / CPUs.
William George
Server 2016 is going per core licencing which means less cores overclocked
http://saveie6.com/
Especially if Microsoft ports full Windows 10 to the ARM
They've been there, done that. MS ecosystem is particularly built upon x86 compiled applications. Sure, they may have ways to have portable stuff, but the stranglehold of Windows is built around legacy applications.
XML is like violence. If it doesn't solve the problem, use more.
And if they're having a significant reduction in power consumption, then adding more cores gets all the easier.
Its always seemed to me that the best approach to processing is to offer a variety of cores and let the scheduler handle what to put where. You can have one or two extremely fast cores, half a dozen moderate speed cores, and dozens or more low speed cores - why insist that all cores be the same in "general purpose" computing?
It's times like this I wish I had a friend named 'The Professor'.
Except they have, in terms of work done per clock (even ignoring multicore). A Haswell 1.2 ghz can achieve the same sort of results as a 3.0 ghz AMD core from 5 years ago in a balanced set of CPU constrained work. It actually comes out ahead in a number of specific workloads. Note I'm comparing to a core significantly older, with less cache for the sake of demonstrating only the senselessness of being fixated on clock, not saying this is a fair Intel v. AMD comparison.
On the other hand, a 1.2 GHz AMD K7 back in the day could beat a 3.0 GHz Pentium 4 of the same time. There's a lot more to processor performance than clockspeed.
XML is like violence. If it doesn't solve the problem, use more.
the software side has been storing up efficiency improvements for a long time. Just get rid of the extras, like bloatware, and hastily programmed apps, and nobody will notice.
I don't see the biggest benefit of reduced power consumption being in the home - have you ever been inside a datacentre? You can barely hear the dude standing next to you because of all the fans and A/C. You can't hear your phone ring. You can't hear PC speaker warning beeps. Now granted, not all of that comes from processor usage, but reducing the power required to run processors would make a massive difference in the cost to run a datacntre.
Not only does the chip consume power, it then requires a fan to move the heat away from the heatsink. The case then usually has more fans to move the heat out of the case. You then need to move the heat to a central cooling system, and then possibly to a cooling tower to be dumped to the atmosphere. For every joule of energy you use on a processor, you need to move 3 or 4 joules of energy around to sustain it. This is likely why, in my experience, power alone accounts for 75% of the cost of a datacentre, once you include the power required to cool the space as well. This could have huge benefits, but true, not really in the home.
"Government is like fire; a handy servant, but a dangerous master." -- George Washington
Except that won't grow their revenue stream. The growth market is the mobile market.
Yeah because you need liquid cool to get to 4 GHz....
"There are processors that work in the Terahertz range "
Source.
" complete bullshit "
Sounds like it.
How about instead of coming up with paranoid delusional conspiracy theories, spend an hour of your precious life and LEARN
https://www.youtube.com/watch?v=NGFhc8R_uO4
Guess what? The universe doesn't owe you continuous growth or eternally faster processors.
Put down the video games, let go of the sci-fi, stop the fantasies, and LISTEN.
And the sloth and complacency of the battery manufacturers vs. the tech industry is what's holding us back. If they were investing into the R&D to keep up with Intel and Moore's law...
And how many trillions would this cost? There's actually massive investments into battery technology. We've come a long ways in the last 20 years. But consider, they're figuring out that we had batteries way back in BC times. The Greeks had them, sort of, they think they were used for electroplating stuff.
But they started entering common use in the 19th century. We've put a huge amount of development work into them. But batteries, it turns out, run into physical laws much quicker than the 'completely new' field of semi-conductor technology. But we're running into the physical laws with semi-conductors now, which is why we haven't seen clock speeds increasing like they used to, and why parallel operations are far more important than they used to be, why we're seeing quad core and even octa-core processors in consumer machines today.
I don't read AC A human right
slower speed is slower speeds, it doesn't matter how many cores you have. It's still the frequency rating that counts. Talk to AMD. Bottom line unless program take advantage or multiple cores - and they don't - you want faster frequencies not more cores.
Yeah because corporations don't want energy efficient machines....moron...
no, because if they did that they they would hold a monopoly on desktop / laptop CPUs. Then they would be regulated as a monopoly, and could no longer get away with their abusive business practices.
Educate yourself on battery technology then post. Long lasting batteries have been the holy grail for just about every application. Research takes time.
Whatever, Intel
-- Tigger warning: This post may contain tiggers! --
What I don't understand is why are last generation parts not dropping in price? For the longest time, whenever new stuff came out, the prices of older stuff dropped. But that doesn't seem to happen anymore.
What's up with that?
Actually, there are some fascinating advances in biological circuitry, which may make transistors redundant.
Real "wetware".
-- Tigger warning: This post may contain tiggers! --
Where I work we have lots of customers that *need* more processing power, and efficiency be damned.
I assume most customers who need extreme processing power have learned over the past 10 years that faster individual processors are not coming. Algorithm design plus parallel processors is going to be the source of perhaps all performance increases in the foreseeable future. Until we move away from silicon that is.
Are there even supercomputers out there which have faster processors than the fastest Xeon processors out there? I may be wrong, but I believe there really hasn't been any non-parallel based performance increases for a long time.
-- All that is necessary for the triumph of evil is that good men do nothing. -- Edmund Burke
Its a constructive use of Reductio ad Absurdum more than Straw Man. He forced the conversation to a ridiculous place because well Garbage In, Garbage Out.
Good-bye
More efficient means less heat which means smaller and quieter devices, so not necessarily meaningless at home.
We hope your rules and wisdom choke you / Now we are one in everlasting peace
MS are playing a very long game because they can afford to. Despite it's well-publicized problems, I find Windows 10 is fast and rock-solid on a desktop and on a Lumia phone. They already have Windows compiled for ARM and they have Office desktop apps compiled for ARM. OK it's a kludged version on the RT platform, but most of the work is done. They are making it easy and attractive (at least in a 'hell, why not?' sense) for new app development to compile for both x86 and ARM. I think one of the reasons why Windows 10 Mobile ('Phone') still exists is because it keeps the ARM branch current and that has sufficient value for MS that they don't even care if the phones never sell.
"Our opponent is an alien starship packed with atomic bombs," I said. "we have a protractor"
Correction, there's been no competition for about a decade now (Barcelona flop)
They could've easily provided tools to let you port WIN32 code to ARM. They didn't want to. Instead they wanted to move to an app store model (just like Apple, duh) based around Metro stuff. Didn't work, and maybe they're kicking themselves now. Or maybe not - there may have been compelling reasons not to support WIN32 code on ARM - but in any case, that's why RT failed.
Posted from my Android phone. Oh, I can change this? There, that's better...
The issue is that a lot of applications people need won't bother to update, and many current applications will forgo the managed runtime upon which MS cross-architecture strategy is based.
Sure, the ecosystem could move, but there's now adequate x86_64 implementations in the space at fairly low cost.
MS' safer bet is to encourage an x86-centric market. Sure, keep ARM port viable and encourage cross-architecture as a matter of course for as many of the developers as they can to hedge their bets, but backwards compatibility is a big selling point for MS platform.
XML is like violence. If it doesn't solve the problem, use more.
Contrary to popular belief, Moore's Law doesn't say that processors will double in speed every 18~24 months. It says that the number of transistors that can economically be put on a single chip will double every 18~24 months. Up until recently, that has translated into a doubling of speed for two reasons: 1) more transistors can be used to optimize the processing of instructions through a variety of techniques and 2) the distances signals have to travel is lessened as the transistors shrink. More transistors contribute not only to power consumption but also more heat, which is another problem with high performance processors. This was partially dealt with by putting multiple cores on a die running at less than max clock rates, thereby distributing the heat and making it easier to deal with. It still may be economical to put more and more transistors on a die, but maybe we don't want to. More transistors consume more power. What's your priority, raw speed or power consumption. Maybe you can't optimize for both at the same time.
Having defeated their upstart competition, they can get back to slacking off. Soo much for "Get ready for thousands of cores!"
(-1: Post disagrees with my already-settled worldview) is not a valid mod option.
The problem is that programmers have gotten lazy (excuse me: "man-power efficient") off of the free speed we've been adding over all of these years. Layers upon layers of abstraction from machine code have made it possible to code in languages which are far removed from the actual code the runs on machines. There may now come a time when efficiency of programming matters to everyone, not just the embedded folks.
Is it just my observation, or are there way too many stupid people in the world?
I don't see how in the world *Windows* is going to break into the mobile market. They have been trying for over a decade, repeatedly without success. Particularly now it seems a pretty cemented Android/iOS landscape. The only hope I could see is Intel getting some hardware makers onboard and that being a platform for MS to push their continuum concept (yes it can work with ARM, but back to square one, a bunch of my enterprise applications are not about to spend money to dust off the build trees and build ARM for the fun of it)
MS mobile strategy is going to have to settle for trying to make money off of iOS and/or Android users/developers. They can (and do) provide hosting, applications, and services. They miss the revenue opportunity of a curated application distribution platform, but I think this is the best they can hope for.
XML is like violence. If it doesn't solve the problem, use more.
Except that Intel has been a licensor of ARM for a very long time, so even if there was some magical shift to ARM in non-mobile ultra-low-voltage devices, Intel would still be able to apply what they know about advancing the state of the art.
Don't worry about Intel, they'll be just fine.
Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
The cortex-A series of chips appears to be catching Intel CISC in some of the raw compute numbers on a per-core basis. Will this possibly rekindle the RISC vs CISC battles of the 90s?
Is it just my observation, or are there way too many stupid people in the world?
Ludicrous Speed....GO!!!
I am not merely a "consumer" or a "taxpayer". I am a Citizen of the State of Texas
Exactly. Apple kept a secret x86 / x64 version of Mac OS X in the closet for 5 years as a hedge against IBM screwing them over on PowerPC. Turns out to be one of the best decisions that they ever made.
Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
It's not AMD. Ever since multi-core started, all Intel had to do was toss in more cores after optimizing a single core for a given process. Since none of the commonly used applications are even adequately parallel (most may at best make good use of 2 cores), Intel is unable to DISPLACE recent Core CPUs at their customers. On the software side of things, Microsoft can force people to Windows 10, but Intel can't force people to, say, go from i3 to i5.
This speed drop is fine if it increases battery life: performance hits are unlikely to be noticeable. The main bottlenecks are download speeds and the loading of videos online, so as long as those don't suffer, Intel will be just fine.
I want a desk warmer that's blisteringly fast. I hope they keep the high end of the bargain too.
The purpose of existence is to make money.
Umm, that's not rolling back. It's a tradeoff.
If Intel sticks to what they've done in the last few product generations, they'll still have higher-wattage higher-performance chips at the upper end for servers and workstations. But the ULV parts have been staying at basically the same performance now for a few years, with drastically reduced energy use. I think the current parts are under 4 watts for the same performance you used to have to spend 18 watts to get.
Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
In addition, lower power per slower transistor does not imply slower per unit area or less work per joule.
I have been involved in the development of things that while smaller and lower power, are in fact faster and lower power per unit area than the larger faster counterparts.
This implies greater performance goes hand in hand with greater parallelism.
I should use this sig to advertise my book ISBN-13 : 978-1501515132.
No, a lot of applications don't scale well across multiple cores / CPUs.
In 2016 they don't. But as chips evolve the applications will as well.
And if they're having a significant reduction in power consumption, then adding more cores gets all the easier.
Its always seemed to me that the best approach to processing is to offer a variety of cores and let the scheduler handle what to put where. You can have one or two extremely fast cores, half a dozen moderate speed cores, and dozens or more low speed cores - why insist that all cores be the same in "general purpose" computing?
If the universe follows the usual scaling rules, I would expect the optimum size distribution of CPUs on a chip for general purpose workloads to be logarithmic.
I should use this sig to advertise my book ISBN-13 : 978-1501515132.
Computer salesperson: "Hey, it's time to replace your old machines." Gov buyer: "Fuck off, they work just fine." Computer salesperson: "But these shiny new Intel models SAVE ENERGY." Gov buyer: "On second though we've got plenty of taxpayer money to blow on 'energy efficiency' projects. Why don't ya' put us down for half million new laptops and two million of those tablet thingies so people can plug them in next to their desktops - I mean 'replace their energy-sucking desktops' - and see if you can't find a new boat for 'my nephew' and a trip to the Caribbean for 'my travel agent' while you're at it."
Nearly every company is thinking about energy efficiency. Every company that moves to a cloud setup is saving energy/power in exchange for a slice of a datacenter that is energy efficient.
Where I work we have lots of customers that *need* more processing power, and efficiency be damned.
I assume most customers who need extreme processing power have learned over the past 10 years that faster individual processors are not coming. Algorithm design plus parallel processors is going to be the source of perhaps all performance increases in the foreseeable future. Until we move away from silicon that is.
Are there even supercomputers out there which have faster processors than the fastest Xeon processors out there? I may be wrong, but I believe there really hasn't been any non-parallel based performance increases for a long time.
Yes there has, but more through architectural changes - new instructions, new modes, bigger better caches, improved offload models, task specific hardware (like crypto, packet moving etc.). This has been enabled by the increasing number of transistors on die and is driven by the mobile and server markets, which have evolving and quite different needs. Xeons today do more work per second than Xeons in the past and the scaling is greater than the scaling of the individual CPU core performance.
I should use this sig to advertise my book ISBN-13 : 978-1501515132.
The vast majority could give a rats-ass about how efficient the machines are. Machines are replaced as they wear out and/or is software-wise obsolete. Cost is a factor in what hardware gets purchased. If a company is really looking to save energy, you start with a GPO (Group Policy Object) that defines power management setting at idle.
Life is not for the lazy.
Er, no I'm afraid they won't.
There are entire classes of problems that do not scale.
Consider, as a thought experiment, any task where the outcome of the first "step" determines the parameters for the next.
There is no way to complete this overall task in parallel, since the other cores are left waiting for the results of the first step.
Yes, there are many tasks that do parallelize well, and these have been well studied for many years. But do not think we can overcome these problems with "progress". We'll be living on Mars first :)
> The big draw of ARM is performance/price per watt which is exactly what Intel is shooting for.
Indeed. Here is an example of interesting hardware:
Parallella: The Most Energy Efficient Supercomputer on the Planet
People may have restated it in many silly ways, but what they actually mean is "Computers become twice as good every 18 months or so." Whether it's multiple cores, or faster clock speeds, or better RAM throughput, that's still what it amounts to: twice as good computers.
As far as I can tell what they are saying is that during the transition period to new technologies there will be a situation where new technologies will not improve and will fall back a little in the area of performance.... which is to be expected. As that new technology improves it will again march reverse and performance will improve. In other words - if power consumption is important to you you will make the leap to the new technology first. If performance is important you will stay with existing technology in the interim. Performance improvements for existing technology is reaching its max, so there will be a stall or slowing down of improvements. That of course was not as interesting so they just spun it to get more readership.
What about people that care more about performance (per thread) than power consumption? Will we be stuck on old technology?
If I can cram more cores in a tighter space with less heat and power consumption then I'll call that a performance boost. Bring on the 24 core i5s :)
I was watching some videos on parallel processing. One quote that I remember was that "cores are the transistors of today". Four decades ago, a CPU like a 6502 would have 3510 transistors and was the cheapest on the market, pulling down prices on all the other competitors. A high-end GPU board like an Nvidia Titan will have 2880+ cores. Going by transistors sizes alone, an entire GPU core will fit inside the space of a single 6502 logic gate. It's going to be easier to add more cores as chip sizes get smaller than it is to up the clock speed.
Intel could afford to crank up the juice to get better performance, because their CPU's were always tethered to the mains power supply. Floating-point performance was where their mojo came from. First they managed to get a floating-point unit integrated with the CPU (80486), then they added SIMD instructions (SSE, AVX) and invented massive cooling systems (water cooling with double fans and a radiator grill).
Vintage computer adverts: http://www.vintageadbrowser.com/computers-and-software-ads
I assume you mean licensee.
Check out my sci-fi/humor trilogy at PatriotsBooks.
Imagine an ARM based server farm.
Imagine a Beowulf cluster of ARM based server farms!
For all intensive porpoises your a bunch of rediculous loosers
> Going beyond 5Ghz limit has been a problem for the last decade or so.
Last decade? Uhm, try the last ~40 years. A close friend of mine worked with the military running GaAs CPUs at ~4.7 GHz in late 70's. He also worked on GaAs devices operating up to ~100 GHz. Hey, when you have a nearly unlimited tech budget you can do all sorts of things that the commercial sector won't have access to until decades later.
Anyways, the problem with Silicon is that it needs to be < 110 degrees C. In contradistinction GaAs only need < 175 degrees C.
Hardware designers have known about alternatives for years -- Silicon is just plentiful, dirt cheap, and "good enough." No one wants to pay $100,000 for a 10 GHz GaAs CPU, when you could buy 2,000x Silicon chips instead for the same amount of money.
No, a lot of applications don't scale well across multiple cores / CPUs.
In 2016 they don't. But as chips evolve the applications will as well.
That's what they said in 2006, when CPU clock speeds essentially hit the wall.
Mainstream CPUs started going multi-core back then. Some things parallelize quite well, and the tools are making it easier for them to do so today, but there's still a lot of sequential crunching to do for a lot of jobs. We're not likely to see a 1000 core 200MHz chip out-performing a 2 core 2GHz chip for "average desktop applications" anytime soon.
Here's the thing though. Even if chips remain equally powerful or 10% slower... if they could fit a 40 core Xeon into a 10watt atom power profile that would be a MASSIVE performance increase in mobiles. I'm relatively satisfied with CPU performance these days with a dual Xeon. If it meant I could get a current workstation in a mobile form, great! However I'm assuming that GPUs do keep improving and we finally see openings for specialized chips for physics and raytracing--the last two areas that would really benefit from dedicated hardware. Neither have ever caught on because Intel keeps improving quickly enough that a small specialized chip market can't get to market before Intel outpaces them.
A lot of people seem to think so. I still get pretty strange reactions when my friends learn I'm at 4.5Ghz with air cooling.
Just wanted to say "Thanks!" for the informative video!
My dream is to some day have my computer waiting on me. Unlike today where I am constantly waiting on my computer... even with the fastest CPU, video card, SSDs in RAID, 16 gigs of RAM, a RAM disk for the swap file... I still find myself waiting.
SLOWER TRAFFIC KEEP RIGHT
Why -1 this guy? They're idiots .. They're so late to the BIG DATA party that the dance is almost over, legislations google is facing ala
https://pando.com/2013/12/16/g...
http://www.consumerwatchdog.or...
Is true for ms(with win10) or apple as well.
The whole business model of windows10, big-data-ads up your arse is a dying concept.
When has ms NOT been late to the party? Kind of their catch frase "late to the party" is it not.
Except I don't believe anyone uses Intel ARM chips.
The cesspool just got a check and balance.
> consider, as a thought experiment, any task where the outcome of the first "step" determines the parameters for the next.
> There is no way to complete this overall task in parallel
In fact it's sometimes trivial. Consider this code, in which 'the outcome of the first step determines the parameters for the next':
HasPMI = IsMoreThan80()
PaymentAmount = CalculatePayment (Balance, HasPMI)
If you have 1024 cores, you can easily run CalculatePayment() in parallel with the line before it. You run it for both the true and false case simultaneously with IsMoreThan80. Then when the three threads complete, HasPMI tells you which of the two results to use.
That can also be EVERY IF STATEMENT, every switch-case, etc. On any branch, go ahead and precompute the value for the branch while deciding which branch you'll take. As things move in this direction, functional programming and similar disciplines start to become more valueable, so they will be used more.
A lot of things that wouldn't make sense to run parallel on two cores or four cores suddenly make sense of you have hundreds or thousands of cores laying around. With 4096 otherwise idle cores, it can make sense to calculate 1,000 possible scenarios in parallel and then ignore the 999 options you didn't need. Our way of thinking about problems will change, as will the tools we use to take advantage of the strengths of new systems.
Building up instead of out
Of course, the fundamental problem this presents is that it does *not* automatically result in improved performance.
Architectural changes require that performance code be tuned or re-tuned, which means every at-scale application has to be somewhere between rejiggered and given a huge dedicated rewrite effort (The DOE's upcoming 300 petaflop GPU machine will have exactly ten applications that can run at full scale, each of which will have an entire dedicated team rewriting it to do so). And, of course, Amdahl's Law puts an ironclad limit on the effect that more parallel hardware can have on performance, and some problems simply cannot be parallelized no matter how much we wish otherwise.
Contrast with the effect of improving the serial performance of hardware: All else being equal, double the CPU and memory clock rates and absolutely every program will run twice as fast, full stop. That was the desktop miracle from 1990 to 2003 or so - the same exact code screamed twice as fast every year.
But as processors trend towards slower and wider, everything becomes an exercise in parallel programming. OpenMP parallel, MPI parallel, SSE simd instructions, GPU simd parallel... It's harder to do at all, and harder yet to do *right*, and historically the average programmer has enough trouble working with a runtime that's sequentially consistent.
Rant aside though, I agree you're right - until we move to diamond substrates & heatsinks, we've hit the thermal brick wall (actually we hit it circa 2003) and there will not be any further increases in serial processing speed. Plus, AFAICT, there's a similar brick wall with access rates to DRAM and the fact that it requires a microwave-frequency bus with literally hundreds of pins extending for entire centimeters... so forget that too.
Coke Says Bottles To Become Smaller But Hold Less Soda
Ford Says Engines To Become Smaller But Cars Will Go Slower
O Brave New World, That Has Such Wonders In It!
Just cruising through this digital world at 33 1/3 rpm...
Intel's so far ahead of AMD, they have to roll back the clocks in order to stay competitive. >.>
Lol, if I had mod points, they'd be yours.
Just cruising through this digital world at 33 1/3 rpm...
Not anymore. Intel sold off their ARM business a decade ago.
That being said, Intel was pretty good at ARM when they tried. I doubt they'd have technical problems getting back into it if they wanted. Marketing and branding might not go so well.
Log in or piss off.
Unless MS can convince developers to actually develop for anything other than windows x86, MS ARM will wither and die, just like windows phone/os/metro/etc. I've lost track of the mobile iterations they've done.
The cesspool just got a check and balance.
Trust me, slowing my compiles down would cost the company more than completely eliminating my power consumption would save.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
" The Greeks had them, sort of, they think they were used for electroplating stuff."
There's no solid evidence for this. There are some artifacts that look like they might have been batteries, but were more likely just very well-constructed storage jars for archiving documents. The lack of any documents referring to the process or of any electroplated artifacts puts the electroplate theory on very shaky ground indeed. Very few archaeologists consider it even plausible.
This certainly doesn't represent "a rolling back of the progress it made over the last fifty years". For the vast, vast majority of people performance/kw is more important than raw performance in a single core. If my process gets a core to itself instead of sharing one that's 10% faster, that's progress.
I'm sure there are applications out there that must be single threaded, but I'd be surprised if Intel stopped making hot, high performance cores for people in that boat. It's just going to be a niche market.
Or more exactly how many crunches (2 bit NAND operations per j/s)*? If that number is growing at an average rate (over 5 years) similar to what Moore predicted the details of the technologies used do not matter.
*Yeah I made that up, got a better metric than the crunch?
At this point I think they are looking for business models that are more annuity-like, with recurring revenue. A transactional purchase of an OS is becoming increasingly less interesting because fewer upgrade cycles. So for them, the strategy was 'app store or bust!'.
XML is like violence. If it doesn't solve the problem, use more.
You can have strong AI in ~20W, because that's what our brain uses. Each neuron is really, really slow like 100Hz and below, but when you have absurdly many it works. The problem is understanding the programming model, because it's nothing like our one list of instructions.
Live today, because you never know what tomorrow brings
People may have restated it in many silly ways, but what they actually mean is "Computers become twice as good every 18 months or so." Whether it's multiple cores, or faster clock speeds, or better RAM throughput, that's still what it amounts to: twice as good computers.
I think that's pretty much failed, then, for general purpose computers. At one time, I actually used to upgrade about every 18 months, and would see a really nice boost in performance. That's not so much the case anymore, it takes more like 3-4 years.
"Somebody has to do something. It's just incredibly pathetic it has to be us."
--- Jerry Garcia
At this point power consumption matters a heck of a lot more for ubiquity than pure performance gains.
I'm sure the fire-breathing dragster edition of current silicon technology (with a pin count to match) will continue to exist at an upscale price for those willing to pay for it.
That uncomfortable rush in your stomach? It's from clinging to yesterday.
is "new technology" isn't guaranteed to be adopted. I'm sure a lot of people would like better battery life. But for desktops,small server installs (or ones dominated by massive per core licensing), or just plain people that don't give a crap they want more power: they don't have to go for the new tech. It could happen Intel converts a fab or two over to the new tech and people keep buying the old model for years forcing them back (or a competitor) to the old tech. Everyone is different but if I get > 3hrs battery on anything I'm good to go. All but say two flights a year I'm not further than that from plugging my junk back in. Would it be good to not have to bring my cords around? Sure. But if the system gets significantly slower to do that it is a trade off I don't want to make. waiting for a compile is bad enough already.
I would have thought these days for 'Continuum' it's just a checkbox in their IDE to target a different processor architecture, with compiler warnings as to why this C code is non-portable.
[X] i686
[X] amd64
[X] arm v7
[X] arm v8
(I guess I should give visual studio a download instead of making uninformed comments!)
That was the power of the emerging wintel duopoly that even IBM purposely blinded themselves to because they thought they'd lose sales from their mainframe business if they promoted desktops as anything but clients to mainframes.
The cesspool just got a check and balance.
The Ghz race is pretty much dead unless we invent better transistors.
But there are other ways to raise the speed, like well actually adding more pipelines to the CPU and making it run more instructions per cycle, but this is not as "efficient" as adding more cores.
I did read somewhere that every extra instruction per cycle per core doubles the core size, while adding an extra core double the performance (if people can use it).
There also other paths, like decreasing the pipeline size, which make the CPU take less time to recover from stalls, but the smaller the pipeline, less clock you can put on the chip before the transistors actual speed limit come in and crash the party.
You can also try the VLIW thing again, that allows you to create those "superinstructions that do several things at once" and increase the chip size a lot less than actual extra pipelines, but then its not x86 compatible, and its very, VERY hard to create a compiler that use it well.
Finally, there is the golden goose of getting several cores to act like a single, more powerful core, but that's a nutjob dream i think.
Good video!
Small Intel CPUs are on par with ARM for efficiency, but the platforms as a whole are not competitive yet. For the servers, Intel is blowing ARM out of the water for ops per watt.
Absolutely.
Some organisations I've worked at with hundreds of employees on site means big savings if they switch. And they don't necessarily own said depreciating hardware assets in the first place but instead lease the desktops. e.g I worked at one place for 2 years and had 3 workstations in that time.
$3/month/desktop might not sound like much but let's say you have 300 desktops on a 24 month lease - that's a $1/4million off your power bills.
Other than Intel being faster and consuming less power. Especially on high end servers. My cousin had to purchase 4 $60k servers, two Intel builds and two AMD builds. Ran them all through a slew of tests for both Linux and Solaris, everything as optimized as possible. Not only did Intel win in nearly every benchmark, but it consumed nearly 1/2 the power. This was important because the mains was only able to handle 10 megawatts and he was able to get their electrician to jimmy rig another 2 megawatts, but they still had several $500,000 racks turned off. You get to do silly things when you have your own 150 megawatt power plant that doesn't charge you any money. State funded University research center.
On the plus side, your fencing skills would improve.
No, progress won't be rolled back fifty years. Even if performance is rolled back 50%, that's only one or two years of progress "rolled back."
And if you're not running the newest build of Win 10, your programs will likely run on the slowest cores.
https://en.wikipedia.org/wiki/Inverted_totalitarianism
They sold off a specific ARM-based product, one which doesn't get much press these days. Why do people think this means that Intel doesn't sell *any* ARM-based products?
I'd argue that it's also the case that most computers for the past decade have been ridiculously overpowered for what most average consumers are asking of them. That's partly why the market is moving to mobile. For many common tasks, a tiny mobile computer is still more than enough to do the job just fine. And in the case of Windows, the required minimum specs for an OS hasn't jumped nearly as substantially since Windows Vista, as MS focused quite a bit on performance optimization rather than letting things keep bloating up. If you had a reasonably powerful computer that could run Windows Vista when it first came out, you could almost certainly still run Windows 10 on it.
Vista recommended specs:
1-gigahertz (GHz) 32-bit (x86) processor or 1-GHz 64-bit (x64) processor
1 GB of system memory
40-GB hard disk that has 15 GB of free hard disk space
Windows Aero-capable graphics card w/ 128 MB of graphics memory (minimum)
Windows 10 minimum specs:
Processor: 1 gigahertz (GHz) or faster processor or SoC
RAM: 1 gigabyte (GB) for 32-bit or 2 GB for 64-bit
Hard disk space: 16 GB for 32-bit OS 20 GB for 64-bit OS
Graphics card: DirectX 9 or later with WDDM 1.0 driver
Note that I'm comparing recommended to minimum specs, but it's still fairly impressive given the time between these two OS releases. In general, I just think there's less market pressure to keep creating faster and faster CPUs.
Irony: Agile development has too much intertia to be abandoned now.
"Except I don't believe anyone uses Intel ARM chips."
The communication processors in XMM standalone modems.
At the time Intel owned them, StrongARM and XScale were pretty much *the* mobile processors, alongside MIPS. Nothing Intel has done with ARM since has been done at the same scale and it'd debatable whether they've made anything viable with it.
I expect Intel still dabbles in ARM like Microsoft dabbles in Linux. They "do it", but there's not a whole lot of love, and you maybe don't want to get your core business too dependent on how they approach them.
Log in or piss off.
Well, I suppose you can call it air since 80% of air is, but using liquid nitrogen and calling it "air cooling" is a little bit misleading, don't you think? ;)
That's only because most people are too unaware to realize they perform much worse with a mobile device than they would with proper keyboard, monitor, and software. Then they export their subpar output to others. Who wants to sit there and read badly 'corrected' txt gibberish in an email concerning a critical issue?
The day all the 'workstation' work requires paying up the ass in 'service' fees for access to remote clusters is the day computing for the masses truly dies.
"it's not just the end of Moore's Law, but a rolling back of the progress it made over the last fifty years."
How about if it's giving the consumer what they want? I don't need more powerful chips. I need more battery life! (and less wasteful software, but that's a different issue)
Don't get me wrong... I've always lambasted the pundits who seem intent on declaring the PC "dead" - that's only true for people who don't actually do any work on a computer. Mobile devices are best at consuming content or *very* light work. Only idiots would argue otherwise. But let's face it - that's the bulk of what most people actually *do* with their personal computers outside of actual work.
And I'm not saying that there isn't still a need for high-powered workstations. It's just that the market for those machines isn't nearly as big as it used to be. And I think PCs have reached a tipping point where, at least outside of gaming or specialized jobs, there's less pressing need for them to be more powerful, so I think that's also contributing to the slowing market.
Don't worry - PCs and workstations aren't going anywhere anytime soon.
Irony: Agile development has too much intertia to be abandoned now.
"and invented massive cooling systems (water cooling with double fans and a radiator grill)."
I can guarantee you Intel did not invent such a cooling system design.
Source: Go look at any car around before Intel was even conceived.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
If they were investing into the R&D to keep up with Intel and Moore's law... doubling their capacity every 18 months as well...
Li-technology batteries hold about 25% of the energy of a similar mass of TNT explosive at the moment. If the manufacturers kept up with Moore's Law as you would like then within three years or so they'd be equivalent to TNT in energy density. Wouldn't that be fun?
This thing looks quite powerful.
Sadly the question "can it run x86" is the first to be asked, and i would guess either intel or AMD are probably trying to answer this one.
A lot of tasks intrinsically don't scale, or scale only up to some limit. Some people are running into this already in the HPC world, were we have big parallel machines that they can't take full advantage of. Their simulations simply don't scale above a certain number of cores.
This problem is becoming steadily worse, since people want to make models with more detail (that tends to not parallelize well), and simulate much longer timeframes than before. If you're simulating protein interactions over one millisecond, then it might not matter if it takes an hour or two. But if you want to use that to understand LTP in neurons and simulate a second or two, then it becomes a very major problem if your model can't parallelize further and the per-core speed stays put.
Trust the Computer. The Computer is your friend.
John Cook (put his blog in your RSS feed if you don't already have it) made a very good point recently: The speed gains from Moore's Law are dwarfed by the speed gains from algorithmic improvements. And unlike Moore's Law, we're not yet seeing a limit approaching for better ways to solve stuff. The post in question: http://www.johndcook.com/blog/...
Trust the Computer. The Computer is your friend.
I know IBM had water cooled mainframes. My first workplace had to call in a plumber to get the cooling system fixed. One homeless tried to build a shelter around the compute room cooling fans in the car-park. Figured he could build a litte hut and capture all the warm air.
Vintage computer adverts: http://www.vintageadbrowser.com/computers-and-software-ads
There are processors that work in the Terahertz range and its complete bullshit that these processes cannot be scaled up. Why is Intel pushing to arrest processor speeds??? Have classified systems hit a wall??? Can the gap no longer be maintained?
Not even superconducting logic families can run at THz frequencies, that is a total and deluded fantasy. Even increasing processing speed to 50GHz with the fastest technology available in the next 10 years or so is extremely problematic, 500GHz? Perhaps in 30 years using supraconducting logic at very low temperatures - assuming a number of scaling problems can be avoided.
The problem isn't the raw switching speed of components, even transistors used in modern processors have a theoretical maximum switching speed far beyond the frequencies that systems implementable using those transistors reach. Reasons: RC delays in interconnects, leakage current at each transistor etc.
The Ghz race is pretty much dead unless we invent better transistors.
But there are other ways to raise the speed, like well actually adding more pipelines to the CPU and making it run more instructions per cycle, but this is not as "efficient" as adding more cores.
I did read somewhere that every extra instruction per cycle per core doubles the core size, while adding an extra core double the performance (if people can use it).
Both of those are trivially false. Wider execution scales as n^2 in theory (though a bit better in practice) but doesn't affect all parts of a core so while some stuctures can double in size the core itself will not. Adding an extra core doesn't double the performance for several reasons: serial parts of a program will limit parallel execution (Amdahl's law), synchronization effects will in practice be worse than that stated by Amdahl and two cores will load the shared resources more than one processor.
There also other paths, like decreasing the pipeline size, which make the CPU take less time to recover from stalls, but the smaller the pipeline, less clock you can put on the chip before the transistors actual speed limit come in and crash the party.
You can also try the VLIW thing again, that allows you to create those "superinstructions that do several things at once" and increase the chip size a lot less than actual extra pipelines, but then its not x86 compatible, and its very, VERY hard to create a compiler that use it well.
Transmeta (and Nvidia Denver) uses translation from the target architecture to an internal VLIW format so virtual x86 compatibility is possible.
Finally, there is the golden goose of getting several cores to act like a single, more powerful core, but that's a nutjob dream i think.
It's not (e.g. Federated cores: http://www.cs.virginia.edu/~sk... ) but thinking that it is a general solution for performance surely is.
You bring up two important points. First, you said "we" want power efficiency. The article says Intel is going to provide efficient CPUs. It does not say that everyone will always prioritize efficiency over speed for all tasks. "We" (many people) will continue to want many tasks to run quickly. In many cases, speed will be more important than efficiency, and that's what this sub-thread is about. We're talking about what to do when you want speed.
Secondly, it just so happens that in the vast majority of cases, over 90% of CPU time is spent in a very small section of code called the "inner loop", which is the little chunk that runs many times.
Suppose you're adjusting a video in some way, maybe resizing it or changing the brightness. The video is a bunch of frames, each frame is a bunch of pixels, and each pixel is three color values, red green blue. There are 256 possible values for each of R, G, and B. The code looks like this:
for each of 200,000 frames
for each of 800,000 pixels
pixel.red=CalculatePixel(pixel.red)
pixel.green=CalculatePixel(pixel.green)
Pixel.blue=CalculatePixel(pixel.blue)
CalculatePixel() gets called 320,000,000,000 times. (320 billion times). Each time, it's passed a value from 0-255 and returns a value from 0-255. Which means that the value for CalculatePixel(0) gets recalculated about a million times. Compare this code:
For x in 0-255
Answer[x] = CalculatePixel(x)
for each of 200,000 frames
for each of 800,000 pixels
pixel.red = Answer[pixel.red]
pixel.green = Answer[pixel.green]
pixel.blue = Answer[pixel.blue]
If you're in the habit of speeding it up by calculating all possible values for your inner loop, you code to take advantage of that fact. Here we can see that it's much more efficient to do the calculation 256 times rather than 320 billion times. This concept is generally true for most programs- the bulk of the CPU time is spent doing whatever the program does repeatedly. I routinely make other people's software faster amd more efficient using this type of approach.
A recent case was a security scanner, which did this:
For each IP 192.168.1.1 - 192.168.1.255
For each port
For each vulnerability
CheckPortForVulnerablity()
You can see that CheckPortForVulnerablity() was called over a billion times.
5 GHz is a pretty high speed, and physics come in play. At that speed, a signal can travel less than 6 cm within a single clock pulse (almost 6 cm based on vacuum). At die sizes of around 10x20 mm, the signal takes a significant part of a pulse to reach its destination after which the transistors still have to make the switch.
This is even more of an issue for the communication between the CPU and the memory, which is often located further away. Distance becomes an issue, even at light speed, at those short intervals.
You seem to have no idea how much Microsoft R&D has contributed to many aspect of technology, especially around computers. A lot of modern GPUs were made possible because Microsoft had many kinds of exotic custom hardware architectures created, then create several custom OS kernels with completely different designs from mainstream, just to see what the best way forward would be. This allowed AMD, Intel, and Nvidia to prototype many ideas.
It's pretty funny in a sad way that today we're only discussing x86 for "high end" servers, when x86 pretty much sucks at it. Take a look at the Top500, Graph500, and Green500. You'll see an interesting pattern very quickly, and it's not pretty for Intel. It also doesn't matter, since Intel has priced out much of the competition at this point and all formerly viable architectures have been overwhelmed by the "good enough" intel chip. Just don't go thinking it's a great chip, because it's not. It's a cheap chip that can do the job, not as well, but good enough. Think of it as the VW Bug of chips in a world of Lambourghinis, McLarens, and Ferraris. The bug sold a lot more even though it's not as good in any sense, but it gets the job done.
The cesspool just got a check and balance.
A well developed & deployed os+compiler
well, that answers the question about Windows portability....
Do you think the hills in China really look all that different from the hills of Nebraska?
Depends, even hills in one part of Nebraska can look look different than in other parts, provided Nebraska actually has "hills" ;)
The cesspool just got a check and balance.
No and here is why.
For a CPU of a given complexity, a specific area is needed for transistors, routing, etc in a given process. If the process density goes up, then the power has to be lowered to maintain the same power/area because the area largely determines the thermal resistance and for the past few generations, high performance CPUs already operate with the junction temperature as high as is reliable. So power is proportional to chip area and higher density processes yield smaller chips so power has to be lower.
You can see this trend in Intel processors since about the Core2. The highest power models all have a power rating proportional to area and since more recent models are smaller, they have lower power ratings.
This is also why stacking memory on top of logic is not going to happen for anything except low performance logic.
Take your tree-hugging chips and jump in the lake, maybe you'll float.
60 different CPUs, it's worse than Ben and Jerries, or Star Bucks.
Give me a fast, high core count, CPU. For a real Man's Computer.
I don't know if I would say Cyrix was faster than Intel. They could beat Intel in some benchmarks, as they had a very good integer unit. But their FPU was garbage, and anything FPU intensive Intel absolutely stomped Cyrix. Cyrix liked to use the performance of their integer unit to calibrate their "PR" system. Sure, maybe their 6x86 PR200 could perform as well as a Pentium 200 in some situations, but in others it struggled to keep up with a Pentium 75. I know my Cyrix 6x86 PR200 would struggle to do anything else while playing back a MP3 file in Windows. The K6-200 could play back the MP3 file with hardly an impact on anything else.
You also have to deal with the speed of light. At 3GHz, light will travel about 1cm. You'll note that's on the order of the size of a CPU die. So go much faster than that, and you run into problems with the time it takes signals to travel across the die and keeping the whole CPU synchronized. So you either have to deal with those issues, or continue to shrink the CPU die.
Marketing and branding might not go so well.
They would be selling to manufacturers, not consumers. I'd guess 99% of consumers have no clue which chip is in their phone, and 90% don't know which is in their laptop.
In terms of performance/watt, a Core i7 stomps all over ARM. Intel just needs very low power x86 chips that also have acceptable performance. They aren't quite there yet, but they're getting closer.
AMD also managed 64-bit architecture that was backwards-compatible with 32-bit, while Intel was trying to say it wasn't doable and pushing pure 64-bit
Performance/watt also has this tendency to be missing another factor. Performance at WHAT (or rather what measure of performance)? Some examples from history include
* iops
* flops
And stuff that may account for above but also has optimizations for:
* triangles/sec
* physics
* fluid dynamics
* lighting models
* etc
That's why we still have PC with fast CPU's that would suck donkey-balls for games without additionally fast GPU's, and why we also have things that are a hybrid (APU) as well as a bunch of edge-cases, optimizations, etc
So yeah, you might have the biggest, baddest spreadsheet processor around, and still have a machine that overall performs more like a Ford Fiesta than a Ferrari when it comes to certain types of media or computations.
Vista at only 1GB of RAM was *NOT* a pleasant experience...
The big draw of ARM is performance/price per watt which is exactly what Intel is shooting for.
I'm too lazy to Google my citations but I remember reading somewhere that Intel CPUs and SOCs actually perform competitively if not better per watt when compared with the better ARM implementations. But that's when the CPU or SOC is actually doing what computers are supposed to be doing, computing. The problem is that the typical Intel desktop CPUs, which the Intel mobile SOCs are still partly based on, are quite poor at doing nothing. In the desktop, there's a clear distinction between a CPU in "sleep" mode and a CPU in active operation. ARM SOCs don't have a distinct sleep mode. They just greatly reduce power consumption when the screen of your smartphone or tablet goes dark. This area is where Intel needs to catch up to, not in terms of raw processing performance.
Even selling to manufacturers wouldn't be easy. There's always going to be that concern about whether or not Intel is "all in" with the product roadmap or whether they're just biding time until they can drop the product for a more internally palatable x86 (or Atom, or whatever they'd brand it) version.
Log in or piss off.
It's not as simple as how many flops can you do. The PowerPC Top500 are created specifically certain types of workloads and are designed by to ran at 100% 24/7. The idle power consumption of those platforms is ridiculous. Super computers are nothing like computers in normal datacenters. Datacenters have large swings of computational usage and do a lot of VMs. Super computers have constant load and run bare-metal for the most part with highly optimized OS's with kernel schedulers that schedule time slices in quantums of minutes instead of milliseconds.
And yes, AMD has great peak flops and can compete with Intel in this area, but AMD's design is biased more like a GPU than a CPU and takes a big hit any time you need to do thread synchronization. Higher latency to flush a cache-line, which is highly detrimental to synchronized performance(Amdahl's law). AMD wanted a large exclusive cache instead of a smaller low latency inclusive cache. ARM is great if you only need a bunch of weak CPU cores. Below 1Ghz, ARM is slightly ahead of Intel. Around 1ghz, ARM runs into efficiency issues and suddenly they draw more power than Intel and only similar performance at best. ARM does beat Intel in some niche workloads, but the same can be said about pretty much any CPU.
It's not as simple as how many flops can you do.
This is why I quoted 3 sets of tests. The Top500 is pretty much flops focused, a very specific test for a very specific workload, which is what all supercomputers were originally targeting back when that benchmark started. While Intel can compete in this arena, as soon as you move to what we might call more realistic workloads, Intel's weaknesses spring out everywhere. You speak of latency - Intel's x86 base architecture has huge issues with process/thread switching compared to any of the RISC entries. Those effects are what kill Intel in the Graph500 list. The Green500 is just a bonus for showing how horrible these processors are, yet as of today, they are the most likely hardware most of us will run. It's kind of like being tied to the current set of inherently dangerous nuclear reactors when a better design has existed for decades, but no one wants to spend the extra cash to get one operational.
AMD also suffers from the process/thread switching costs as originally they were x86 based and I'll be honest that I've not kept up with what they've done since the developed their RISC like core, so can't comment on to what extend they're suffering from those effects today.
The cesspool just got a check and balance.
Funny, because you keeps using the same debunked argument. Also, delusional people talk about themselves in the third person, it isn't a good idea to intentionally lump yourself into that group.
You have yet to submit to a code review from anyone but your friend. No, I don't trust that he has thoroughly assessed your software.
You are terrified someone will steal your software if you publish the source code, this must stem from the fact that you are stealing other people's work in your code (you didn't write the hosts files, you just steal them and collate them).
You have yet to address the issue of name resolution performance of anything not found in your hosts file. This is a serious issue when the hosts file is so large, as Windows has to step through the entire file before it fails out. DNS outperforms your hosts file solution several fold, so why not just run your own DNS server? Oh, resources eh? But you have no problem running 100k copies of the hosts file in a domain instead of using DNS, like that is even a solution.
When you have an argument tougher than a wet paper bag, let me know, for now, your argument is debunked, now go and learn something about how computers work, since you seem to have missed that class.
APK likes to ask for responses to the same things over and over. Maybe he just likes the responses?
Only if I get some action on my request for company swords. The request for a ball pit in the office space we're moving to soon didn't get approved either.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
Same arguments again and again, all already refuted:
http://slashdot.org/comments.p...
So, eat your words as you say. You definitely post enough of them, since you enjoy spamming so much. I counted 4 posts on the Adblock Plus article, which just goes to show, you are as bad as all the advertisers you try to block.
APK likes to ask for responses to the same things over and over. Maybe he just likes the responses?
So, still the same claim, yet again, refuted here: http://slashdot.org/comments.p...
No one is going to steal your software from the code, as you already give it away! If you are that scared of people seeing your code, it isn't because of a concern about theft, that is sure as there is nothing to steal. Copyright covers your code like it covers all other code, no one can legally copy your code without your permission.
I don't "steal" anything (projecting YOU DO)!
So, now you accuse me of theft? But you don't take other people's work? So, where do you get your hosts files since you aren't taking them from other people's work?
By placing users FAVORITE SITES where they spend 95++% of their time online @ the TOP of hosts files cached in RAM, I get them to sites FASTER & MORE RELIABLY than a more-than-potentially REDIRECT POISONED DNS SERVER (99.999% of ISP DNS aren't patched vs. the kaminsky flaw, stupid).
So, you lie and redirect. The name resolution performance issue was clearly stated to you, but you have to dodge it to make your solution look better. So, how long does the resolution of items not in your hosts file take? It doesn't matter how many favorites you add, there are always things missing, and adding a static entry for something doesn't fix the issue as many are dynamic in some way.
APK likes to ask for responses to the same things over and over. Maybe he just likes the responses?
No it doesn't (see using hardcoded favorites above) - & DNS outperforms hosts in GOING DOWN (does a lot) OR poisoning users via redirect poisonings!
So in other words you are a shitty sys admin since your DNS server has uptime issues? My DNS server is only down when I take it down for maintenance, and that is why there are multiple DNS entries in the network configuration. How do redirect poisonings effect a DNS server you run yourself?
DNS resolution far outperforms hosts files as it uses a branching algorithm instead of stepping through a long file. Even for entries you redirect to 0.0.0.0, it will be faster to do the resolution in DNS.
Yes, more resource consumption + moving parts complexity AND POWER USE doesn't = a GOOD solution vs. hosts by using redirect poisoning exploitable DNS locally w/ only a few systems @ home.
So, you are running on a pentium 3 with 256MB of ram? Why are you so concerned about resources, can't you afford a recent computer? The power usage of a DNS server is most definitely less than your shitty hosts file in actual use, but I am sure you made measurements to determine this?
It works easily migrated by central admins via scripts or chronjobs/scheduled tasks with less moving parts complexity, room for exploit & breakdown, OR power usage.
So, slowing down 100k computers is so much better than DNS? A DNS server that is required in an environment like this? I'm glad I didn't work in any company you did, it must have been painful to use those computers.
Yes I have to a seasoned security pro AND coder himself.
AGAIN. One person reviewing your code is NOT a code review. Code reviews are performed by many people to prevent mistakes, and subversion of the process.
APK likes to ask for responses to the same things over and over. Maybe he just likes the responses?
You know, posting things in all caps doesn't suddenly make them valid or true right? You can keep beating that strawman all you like, but you are still wrong. DNS beats Hosts every single time when an entry isn't in the hosts file. DNS beats hosts files after about the 5th record for speed.
Still beating another strawman there too, one person reviewing your code =/= a code review, as I have repeatedly stated.
And you CLAIM you're a security guy? No, no way.
Apparently I am way more of a security expert than you, because I actually understand these things, something you have repeatedly shown you do not do.
he'd have KNOWN if I 'stole others code' WHICH YOU FALSELY ACCUSE ME OF you disgusting little reprehensible PIECE OF SHIT!
NEVER did I accuse you of stealing code. I said your software uses other people's work in that it coolates copies of other people's hosts files, not your own file you created by hand. Now, who is the reprehensible piece of shit here? since you can only attack strawmen, I totally expect the EXACT same argument yet again, as you can't beat the actual statement.
LOL - you, a FUCKING LIMITED IN RANGE OF SKILLS IN COMPUTING menial, lol (who has no code the protects others & speeds them up online as I do which is HIGHLY RECOMMENDED by notables in the art & science of computing)... apk
Again with the insults? Can't take anyone actually questioning your product and pointing out its numerous flaws, you have to try and claim you are so much better than anyone rather than actually addressing the issues.
APK likes to ask for responses to the same things over and over. Maybe he just likes the responses?
You are hilarious. Where is your actual testing? Come on, you have to have tested it to claim that yours is faster and uses less resources.
Annihilation would actually require you to have a single response to anything I said, since you still are fighting strawmen it can only be presumed you have no responses to the actual issues, therefore you have already admitted that you are wrong and have no responses but attacks. It is lovely to see you finally admit that you know nothing about security as you can't even make any kind of response to the actual known issues I have brought up with your method.
APK likes to ask for responses to the same things over and over. Maybe he just likes the responses?
You added remote DNS to the statement. I never said the DNS was remote, why would I use someone else's DNS I don't trust?
How is it faster to query a Hosts file (which has to be stepped through in its entirety for records not found in it) than a DNS server running in memory (which uses branching tree algorithms to do resolution)?
FAVORITE SITES @ THE TOP OF HOSTS vs. REMOTE DNS
Oh, so you are still attacking a strawman, I see. Stop talking about favorite sites in the hosts file, as that isn't what was said, it doesn't solve the problem of slow resolution of things not in the hosts file.
* YOU FAIL, limited menial... lol, you fail.
It must make you feel great to keep defeating those strawmen. You sure are the mental midget you claim everyone else to be.
APK likes to ask for responses to the same things over and over. Maybe he just likes the responses?
Pretty much, intel can build fab buildings on the whim and could buy AMD outright but whats the point if they're not really competition and will get hit with monopoly practices for x86