HAHAHA
IT'S OVER THE DESKTOP IS FINISHED
http://www.fool.com/investing/general/2016/03/22/intel-corp-officially-kills-tick-tock.aspx
>Yesterday
Something good
>Today
Something worse
God I love marketing
How is Intel going to survive this with the Zen release coming soon?
I don't get it
>>53641319
It means architecture and node lasts 3 years, instead of two.
So you'll be getting
Smaller node > Architecture > Rebrand
probably due to you cant go lower in nm anymore
>>53641357
It's probably because you're a uninformed halfwit.
>>53641357
Pretty much this. They're approaching the limits of what's physically possible with silicon, of course they're having to put more research into each process shrink.
I was promised doubled speed every two years.
EVERY
T W O
W O T
YEARS
>>53641332
We already got that with Kaby Lake
Thanks Intel
>>53641268
97% of the server market. Desktop CPUs are increasingly becoming more irrelevant.
It was finished in 2004 already.
>>53641432
>21% per year
Fantastic joke
>people getting mad at a company for the laws of physics
>>53641432
>that DEC Alpha domination until 2001
>>53641449
they better take a shortcut and JUST create a quantum cpu
>>53641432
The only upside is maybe all the horribly inefficient bloated software and OS will start to get optimized a bit once it's obvious next year's model isn't going to bring a huge single core performance improvement.
Who am I kidding, retard devs are already trying to make entire OS out of node.js
>>53641481
>quantum computing meme
won't do anything for normal computer needs
>>53641481
JUST fuck my physics up senpai
>>53641449
Their job is to fuck over physics, we could have gotten much more, but no, jews had to milk all the cows and wait for the last second to use EUV, FIB or triple patterning.
Or better, use a better material than silicon, there's a bunch of them.
But then again its a major cost to them, consumers and performance be damned.
>>53641432
>that AMD FX dot on the far right
>below the lowest Celeron from the column to the left
>>53641502
Their job is to make profit. Just because they have some theoretical material/method of creating a cpu doesn't mean it's profitable or viable in any other way.
You can make the most amazing bullshit every, but it's only with a billion dollars in RnD money going into making a couple prototypes.
>>53641502
>STOP
>MAKING
>PROFITS
>WAAAAH
>>53641389
We're near the physical limit of electronic processors. Quantum photonics may reboot this trend.
Moore's Law is dead and has been for a long time. We can no longer expect simple exponential scaling in single-threaded performance.
That doesn't mean the desktop is finished. We will still see architectural improvements, both in hardware and in software. We will still see rapid improvements in graphics. We will see more usage of heterogeneous computing. Even CPUs are still (slowly) improving, though the improvements are in core count and vector operations rather than pure IPC.
Now we are refining architectures rather than massively scaling up CPUs. That doesn't mean there is no progress in the desktop space.
And eventually we'll move away from silicon ICs.
>>53641521
>>53641520
>I love 3% performance improvements
That's you,
>>53641432
JUST
>>53641533
logicalfallacies.png
>>53641533
>I'm a communist
That's YOU
>>53641268
They'll survive just fine. Zen will be competitive and will probably undercut Intel on price, but Intel has huge amounts of cash and huge dominance.
>>53641319
It's getting difficult to reduce nanometers, 14 was hard and 10 is going to be even harder so now they're looking into stretching out the schedule to work out all the kinks.
Intel should start looking into using something other than Si for their chips.
>>53641571
>Intel should start looking into using something other than Si for their chips.
Do you think they're not doing that? We've been looking for other shit for decades.
>>53641580
It's already happening, but it's going to take a decade or more before mass production can occur. http://www.zdnet.com/article/ibm-research-builds-functional-7nm-processor/
>>53641614
eh i read on wikipedia intel expects they can do 7mm on silicon.
the incompetence man, sad
>>53641486
That sure would be nice, I'm sure some devs will actually start optimizing once they realize that any extra bloat can't be hidden away due to higher hardware performance
What happens after 5nm?
>>53641725
We finally realize that a 6GHz processor is enough for any task a civilian will ever need
>AMD sticks to Phenom II X4 & X6
>/g/ laughs
>AMD unveils Bulldozer
>/g/ laughs
>AMD can't keep up with competitors fab processes
>/g/ laughs
>Intel fucks you in the ass
>/g/ cries
This the future you chose. You know when Zen comes out, and it wont beat rival Intel parts, you'll laugh again and in a few years Intel will unveil their plans for CPUs to stay stagnant for 5 years at a time.
And you'll look to AMD's corpse for laughs but you wont find any. Only tears.
>>53641739
Will 6GHz processor make my waifu real(er)?
>>53641741
>AMD sticks to Phenom II X4 & X6
>/g/ laughs
was /g/ even a board back then?
>>53641741
So both parties are shit? I miss powerpc
>>53641725
MORE
O
R
E
CORES
O
R
E
S
>>53641784
IBM actually produced a 7nm prototype, they are our hope
if we ever go below 10nm then we will have to work out how to solve quantum mechanics and that means we would have quantum computing anyway. people better start learning quantum theory because developers that don't will be left in the dust!
>>53641814
>implying devs know electrical theory
They'll be fine
>>53641771
Fuck off underage.
>>53641223
>bend over just a little more, goy!
>>53641660
Did you read a troll edit or something? https://en.wikipedia.org/wiki/7_nanometer
>Although Intel has not yet divulged any certain plans to manufacturers or retailers, it has already stated that it would no longer use silicon at this node.[7] A possible replacement material for silicon would be indium gallium arsenide (InGaAs).[8]
>>53641741
This conspiracy theory bullshit about Intel intentionally keeping performance from increasing because "MUH SHEKELS GOTTA MILK DEM GOYS" is fucking retarded. The reason you're seeing performance slow down is <14nm is fucking hard as shit to realize, and it seems Intel is the only game in town.
>>53641843
most devs dont get math or basic physics right there is no reason to belive that there wont be gnu/gcc that allows to compile your unformated and unoptimised shit stained code to a executable blob of vomit like it has always done
>implying optimisation will ever be a thing
>>53641976
>only game in town
That's the whole, tragic problem.
With AMD and Intel competing, they spend less on shareholders and more on research. We double the size of the combined research pool. We have more diverse technology as the two companies attack the same issues from different angles.
It's not a conspiracy theory, it's good business sense.
>>53641976
>guys, really, that multi-billion dollar business is trying their very best. Stop going so hard on them, they can't do any better!
>>53641725
We go home because that is the end of our shift.
If you aren't stupid you saw this coming. Dieshrinking has been getting harder and less rewarding. Skylake got delayed several times, and broadwell fucking vanished since 14nm was so late.
>>53642061
Intel knows desktops are dying anyway. The new game in town is small, power efficient chips that don't need to run at a billion GHz to play angry turds/browse facebook.
>>53641976
but it's true. There's tons of workers also at Intel that admit hardware/software is constructed to only last for x years - after which they get deprecated with new standards.
They could create it so that it lasts longer and is more 'updatable' , but they don't want to.
>>53641268
By having processors way better than AMD.
>Implying Zen will get even Sandy Bridge performence
Top kek
>>53642174
Thermal limits were really the biggest problems. Both power draw and density.
>>53641223
Well, they are already at 14nm and going smaller is incredibly difficult. I would say it was expected.
>>53642235
Constant development requires breakage.
>>53642338
But wouldn't it just be nice if we, at some point, started to focus on making computers actually more productive and programs actually less bloated?
>ree's internally
>>53642327
The reality for many hasn't set in. They don't want to give up their 50% performance gains and are in serious denial. The ride is over, guys.
>>53642360
Give it a few years. Maybe the retarded "make everything slow interpreted garbage" trend will finally start to die out.
>>53641432
>single thread performance
Remember when they started adding MOAR CORES?
>>53641481
>>53641493
>>53641526
>quantum meme
Have fun needing to do everything in the cloud due to the specialized cooling equipment required. You think the botnet and government spying is bad now? Just wait until you have to use a dumb terminal for everything if you want to use the latest software.
Fuck this shit, I just bought Intel stocks.
Does Intel have any plans to bring out better mobile chips?
The desktop market is really shrinking and right now even Nvidia looks better.
Is there any new material in sight than keep on using silicon?
>>53643007
Si-Ge is a good stopgap for now. 7 nm chips will be made on it. How they'll deal with oxidative damage (Ge is shit at being in air) eludes me.
Ok, if performance gains are really marginal let's play the other game, called efficiency.
How about making their CPUs more efficient?
I think this could go well with marketing.
>>53642966
Stock is set to climb this year. Buckle up.
>>53643243
Nope, the quarterly reports are being presented soon, and it's not looking too good due to the shrinking desktop market.
I hope the stocks don't slide down.
Intel really needs to invest into something else than the desktop market.
They tried to get a hold in the mobile market with the smartphones, but so far have failed.
ARM based SoCs dominate the market.
>>53641268
HAHAHAHAHAHAHAHAHAHAHAHA
if they (intel) could make a high end i7 in the mainstream segment a six core 90 watts and below i believe they would of already done it. they appear to want to keep mainstream below 100 watts. adding two more cores to the mix along with ht would put them over that. i don't think intel will increase core count above four in the mainstream segment until they can reduce power consumption further without having to lower stock frequency. they quite like having that 4ghz processor.
i honestly believe this is why amd went with their cmt bulldozer architecture originally. they sorta knew (lets face it, both knew then), they were going to hit a wall in single threaded performance and that multi cores were the feature. they just didn't think cmt was going to suck that bad at release and provide them with worse single threaded performance than their previous "stars" architecture.
probably why zen will only match haswell / broadwell single threaded performance. there is only so much you can extract.
>>53641223
I expect Intel to do Process>Optimization>Arch>Optimization after Cannonlake. There's going to be some fuckups and will delay things while Intel figures stuff out.
>>53643353
Bulldozer was supposed to be a revolutionary new architecture for single thread performance. There were two cores per thread so moar corez seemed like a good idea. When AMD discovered that was a disaster, they ran out of ideas and just added moar corez to beat Intel at highly parallel benchmarks and hope people would be hyped enough to buy their dead architecture.
>>53641402
I'm actually looking forward to Kaby Lake
Full HEVC hardware decode without needing an Nvidia card
>THE DESKTOP IS FINISHED
what?
It just says intel moves from tick toc to tic tac toe process.
>>53643741
kek
>>53641408
This. It's all dependent on the server market and Intel has damn near 99% marketshare.
>>53643741
Tick tock was 2 years. Then they doubled it to 4 years. Now they made it 6 years. Soon, it will become 8 years. Etc.
>>53643075
Airtight CPU
>>53644227
Which means you can't delid anymore. RIP extreme overclocking.
>>53643671
>Full HEVC hardware decode
Processors do that fine right now without acceleration.
What we need is hardware encode.
>>53644784
>What we need is hardware encode.
Polaris is promising transparent UHD@60 fps h.265 encode, and it'd be realistic to expect Pascal doing the same.
We'll see how well they actually do.
>>53643297
http://markets.ft.com/research/Markets/Tearsheets/Forecasts?s=INTC:NSQ
So now let's hear your opinion, since clearly you are a financial analyst
>>53643557
>There were two cores per thread
No. Bulldozer was never a VISC architecture. Ever. Originally the module was wider, with 3 ALUs per core and 256bit data paths in the FPU. Test chips were produced in 45nm silicon, but they never went past engineering samples.
>>53643075
You don't know what SiGe is, or how its used.
SiGe is already in use, future nodes will use it in the channel instead of just the wells. It is not a substrate material itself.
>>53641792
>IBM
You mean Global Foundries.
>>53641223
Since we're CPU talking, about to finish a new build. Last part needed is a CPU cooler for a 6700k in an itx case.
what do?
>>53645542
I have a noctua nhl12 on my 6600k. Works for me. There are better performing models out there, but it fits my case perfect, and seems to handle my workload adequately.
>>53645486
No. IBM and SUNY worked together to develop carbon nano tube transistors
>>53645652
I've got a Hyper 212 Plus I lapped years ago I can re-use, and it seems like CPU cooling tech hasn't fucking changed much in 5 years since I built my last one.
Most review sites have CPU coolers on their roundups/benches not even available for purchase anymore in their top 5.
>>53645692
Again. You mean Global Foundries.
IBM no longer owns a fab business. All of their tooling, process IP, and most all of their facility staff belongs to GloFo.
>>53645741
It would probably be just fine. My case has a 140mm fan that sits @2-3cm above the vertically- exausting noctua. Perfect for my situation, although YMMV of course.
>>53645743
Jesus fuck you are dense. It was built at SUNY. I know gloflo, and Samsung are IBM partners but flofo didn't make this one. Yet.
>>53645829
Global Foundries, Sumsung, and IBM are not partners.
Global Foundries and Samsung are partners. IBM's fab business no longer exists, it is entirely the property of Global Foundries.
Stop being a retard.
>>53641223
The result will be indistinguishable from the previous cycle which has been stagnant since Core2.
>>53645880
I had to go and find the press release. Here:
https://www-03.ibm.com/press/us/en/pressrelease/47301.wss
>>53645940
Congratulations on not being able to read.
Global Foundries owns IBM's fab business. Get this through your malformed subhuman retard skull.
>>53641223
>uMatrix has prevented the following page from loading:
>http://www.fool.com/investing/general/2016/03/22/intel-corp-officially-kills-tick-tock.aspx
No thanks, take this tracking shit back to its grave.
>>53645975
It was made at SUNY you dense fuck
