ATTENTION: MOORE'S LAW IS DEAD!!!
Intel finally retires their Tick-Tock development process.
They plan to cuck the consumers by milking the same product multiple time from now on.
>>53720353
Main article link:
http://arstechnica.com/information-technology/2016/03/intel-retires-tick-tock-development-model-extending-the-life-of-each-process/?comments=1
It's just changing to Tick-Tock-Tock. Calm the fuck down.
old news
>>53720515
More like Tick-Tock-Cuck
slap some linux on this, and it's true
>>53720353
Not really, we've just hit a physical/practical limit on silicon, need to start making hybrid shit, or maybe graphene chips.
>>53720353
Past:
Tick - Tock
Present:
Tic - Tac - Toe
Future:
Eeny - Meeny - Miny - Moe
>>53720552
>More like Tick-Tock-Cuck
10/10
Every tick and tock has been delayed. We have been cucked for years by Intel.
>>53721248
Oooh graphmeme, while I'm at it maybe I'll just pull quantum super computer out of my ass and it will all be fine. Easier said than done faggot.
What makes you think any of this shit will actually come to fruition and not just turn into research branches that reached a dead end?
>>53720353
History is filled with motherfuckers thinking there's nothing left to improve or invent. Join the club.
>>53720353
Thanks AMD
Luckily ARM exists
>>53721352
/the truth
>>53720552
well memed.
>>53721352
We're going to be stagnant for many years on tech due to this limit, at the very least AMD has a bit of a chance to catch up now
>>53721379
Still not enough productivity software support nor performance
>>53720353
They'll still release new processors, but they will be cheaper, and more power efficient. Not much faster. Possibly slower for mid-range or low-end.
>>53721516
>Intel
>releasing cheaper processors
>>53721379
ARM processors are just not designed for desktops and will never reach the same level of performance as x86. Just like x86 will never reach the same level of power efficiency.
What ever happened to mainstream quantum computing? I remember that was the big buzzword back in the mid-2000s but I don't hear much about them anymore.
>>53721593
that's buzzwords for you.
>>53721593
The current technology is still limited to research labs and extremely specific use cases.
What you're suggesting is like saying that a bunch of 1940's transistors be used to make a general purpose reprogrammable supercomputer.
We're looking at 5-10 years before we seen quantum computers being used in financial computing, let alone household.
>>53721693
>We're looking at 5-10 years before we see quantum computers being used to make all modern encryption obselete
Fixed for you.
Its not Moore's Law thats dead, its x86-64.
>>53721539
I know what you mean, but they kind of have to or they will just be priced out of the market by cheap-ass ARM and AMD SoC/APUs. The market for big-ass desktops with super powerful dedicated CPUs is shrinking, and there's no signs of that stopping any time soon.
>>53721764
Yes, we're all waiting for that brand new shiny architecture that does away with x86 forever and has amazingly optimized software written for it. Except it will never happen unless there's decades of no progress.
>>53721593
Many people learned eventually that the advantages of quantum computing can only be exploited for narrow class of problems. Otherwise they are no better than a regular computers.
The hype was partially due to marketing and overoptimistic people who wanted to exaggerate and believe that the future will be miraculous.
Eventually reality hit everybody on the head and they realized that they live in a dark universe where the miracles they hoped for don't happen.
>>53721764
All hail Loongson MIPS!!
>>53721786
I know that it is pretty much impossible to replace it, at least not within a decade, but still you can't just milk the architecture for centuries expecting to never hit a brick wall performance wise.
>>53721593
We still know nothing about quantum physics and the use cases of quantum computers are very limited.
>>53721593
Quantum computing units! To accelerate certain calculations!
Now in your nearest retailer near you for $30 000
/tpg/ been saying this for months. Sandy Bridge is still king.
>>53721804
You joke, but /g/ is probably one of the largest groups of users for this architecture outside of the targeted Chinese market.
>>53722218
The performance is good enough, feature/efficiency wise its hopelessly obsolete.
Thinkpad fags only deny this because everything >*20 series is shit.
>>53722267
>feature/efficiency wise its hopelessly obsolete.
what features?
>>53722337
DDR4, PCI 3.0, HBM, etc.
>>53721593
but 2500 hasn't even happened yet
>>53722337
Do you want me to spoonfeed you the spec sheets?
>>>intel.com
>>53722373
irrelevant for laptops
>>53722439
no. want you to back up what you're saying. if that is unreasonable then don't post on a discussion thread.
>>53722488
I need to back up the claim that a 5 year old platform lacks behind in features?
Like seriously, if you don't know how the mainstream platform developed in the last couple of years there isn't even a point in discussing this.
>>53722488
>>53722578
I sense two people who have gotten rustled today.
>>53722578
>5 year old platform lacks behind in features
of course. but this doesn't makem it "hopelessly obsolete"
>>53722459
How is HBM and DDR4 irrelevant for laptops? Nice try.
>>53722774
and PCI3.0
>>53720353
>Intel is forcing consumers to buy the same product multiple times
Pretty sure the consumer is the one who makes that decision.
It's the same thing they did with Haswell. This isn't really news.
>>53720548
/thread
>>53722774
/v/
>>53722825
lol
>>53720353
>ATTENTION: MOORE'S LAW IS DEAD!!!
Good. Maybe we can focus on software efficiency instead of falling for the "faster CPUs will fix it" meme.
>>53722639
No this guy just thinks there has been some world-changing laptop revolution in the last 5 years. Because my family was telling me at Easter dinner how they would shoot themselves if they read one more story about new laptops; everyone was dumbfounded when I brought up self driving electric cars, artificial intelligence via siri, cortana, alexa, and 4k pressure sensitive cell phone screens.
>>53722337
Meme shit they think makes a real difference for their grandma-tier use cases when most applications won't even take advantage of it.
>>53722439
you'll spoonfeed every gamertard who comes knocking from /v/ his next leet build for free, why don't you spoonfeed us something
interesting for once?
>>53721248
Circuits made with 2D nanosurfaces are years upon years away. I'd be surprised if there were ones available in 2020.
Also, graphene is only the easy part of such circuits.
>>53720353
>Moore's Law
Did any of you faggots actually think this was a 'law'?
Fucking preposterous.
>>53723417
I was thinking they could move to a new CPU architecture with even more cores and multi processor builds would become more popular, but that works too (and would probably be necessary to work with all those cores).
>>53722218
Sorry, but I prefer Ivy Bridge.
The chip no one wanted :'(
>>53723894
youre entitled to it. the intel 4000 is pretty nice for my kodi laptop
>>53721562
>Just like x86 will never reach the same level of power efficiency.
x86 has already surpassed ARM in power efficiency. ARM is now officially a meme cpu architecture especially since android can run on x86 just fine.
>>53721562
Arm is fast enough for desktop.
X86 is more efficient.
Your post is so full of shit I feel like I'm in India
>>53724114
it's moores law v2
>>53723832
Dual CPU 64 core ARM desktop when?
>>53724004
It can't though. Android still needs some tweaking to work on x86.
>>53724114
>Arm is fast enough for desktop.
Nope, not really. Simple things like 1080p VP9 decoding on youtube require something with about the performance of an Athlon 5350 (4 x86 kabini cores @ 2GHz).
Given that x86 silvermont atom cores have about 25% less IPC compared to x86 kabini cores and x86 silvermont cores pretty much match ARM krait 400 cores, you would need 4 ARM krait 400 cores running at about 2.5GHz constantly (no throttling) or 8 Cortex A53 cores running at about 1.4GHz constantly (nothrottling) to decode 1080p VP9 smoothly.
This is for 30fps 1080p VP9 streams btw, not 60fps 1080p VP9 streams.
Point is ARM has no chance of entering the desktop market especially since it can't run on x86 software.
>>53724185
>what is the zenfone 2
>>53724359
>GPUs don't exist
>>53720353
And who cares? I'll still use my 3770K then. I think it will be still i3 tier in 2020. I won't upgrade unless they give me a significant performance increase or more cores. Next big for me thing is 8-core i7 below 400 €. I already need 2K to upgrade this with a mainboard GPU and watercooling.
>>53724387
Not for VP9 decoding though.
>>53723417
efficiency is a meme
speed is everything
200ghz when
>>53724443
when you have 0.035 IPC
>>53724375
Nice argument, faggot. Android itself can run on x86, but not optimally yet. There's even a book about optimising android for x86. Just Google it.
>>53724524
>Nice argument, faggot. Android itself can run on x86, but not optimally yet. There's even a book about optimising android for x86. Just Google it.
Well the zenfone 2 beats phones with energy efficient A53 AR SoCs in battery life so it looks plenty optimized to me. But better battery life would be better desu, maybe it will be enough to trash all these angry bird processors.
>>53724575
That's because Asus isn't Samsung, they did their best to optimize it. That doesn't mean every company will do the same. But Asus really needs to step up their design and naming. The zenfone looks, sounds and feels shitty, even though it functions very well.
>>53724632
Not him but do you think android in general could be trashed if x86 phones ran desktop versions of windows and included precision stylus input?
I've always been hoping PDAs would make a comeback.
>>53724632
>not everyone will optimize
It's open source software, are you fully fuckin retarded?
You've gone from "it won't work" to "it won't work well" to "yeah, well not everyone will do it."
Just fuck off already and next time you're wrong fuck off some more.
>>53724715
not him but to be fair some companies do go out of their way to make android run like dogshit (ie samsung).
>>53724692
This is why Microsoft is working on Continuum. It's why Windows 8 shipped with a touch UI. It's why USB-C is being pushed really hard. We've been going in this direction for over 10 years, but semiconductor fabrication is only just now getting to the level we need to make it a reality.
I daresay that Microsoft is best prepared in terms of tech to make great devices, but their phone market share is so tanked, I don't know if it matters.
>>53721353
The simple fact is that ISAs need to change with the computing needs of a given time period. They usually do, such as CISC x86 becoming RISC with a native recompiler.
People also misconstrue where things are going. Performance per watt is essentially as important as IPC and the move towards one will still benefit the other in some way. The only difference is that, as with any very mature technology, further gains on x86 will be difficult and wrought with failure along the way.
So how long do you think an i5 6600k will last for gaming?
>>53724839
Well the x86 instruction set was horrendously complex and had a lot of historical baggage, nowadays RISC can express more or less the same things and apply optimizations that you previously have had to use a specialized x86 instruction to get
>>53720552
R O A S T E D
O
A
S
T
E
D
>>53724840
Hard to say. We finally have game development APIs that leverage multicore processors properly, but we don't know where the sweet spot between utilizing the hardware and ease of development is.
We also don't know how much work Intel has put into new semiconductors. Intel started building their own fabs a while back and they probably did that to experiment with new semiconductors. They may have new tech ready next year. If and when a new semiconductor tech comes to market, all bets are off until the capabilities of said tech are fully understood.
>>53720552
Perfect. Just perfect.
A doubles or triples would have made this post even more perfect than it is, but that part is just luck and not the main point.
I thought the "E" series was supposed to represent optimization? is Kaby Lake going to basically be Skylake-E or something different?
>>53724991
Intel appears to be optimizing Kaby Lake for mobile devices. They've paid special attention to sleep state management on Kaby Lake, and have also added a USB 3.1 controller to the SoC.
That screams to me that they want to compete with ARM, and they certainly will.
>>53725154
>That screams to me that they want to compete with ARM, and they certainly will.
Not him but since google started the push for chromebooks there's been a large focus on making notebooks that sleep the same way tablets and phones do. It turns out that people quite like that they don't have to wait to get back into things.
>>53725260
My laptop wakes faster than I can open it doe
>>53724943
kekd
>>53724359
>ARM can decode VP9
>so ARM cant decode VP9
bro wut the fuck r u smoking
>>53724430
Good thing you just listed a bunch of ARM CPUs that can decode VP9
>>53720552
>>53724692
>Microsoft
fuck off fag
>>53723832
I doubt we would abandon x86 due to the massive amount of legacy software, though I supposed it could happen. Without question we could increase performance by using a RISC architecture like PowerPC or ARM. Pure RISC (not "hurr durr decode CISC to RISC on chip") just allows you to make more efficient use of the silicon. Right off the bat some of the space that goes to x86 decoders can go to additional units or cache since the RISC decoders will be simpler. You'll also have fewer stalls in the pipelines and less dependency between instructions, and some of the logic to deal with that can go to something else (caches).
But there are still limits. At best PowerPC was, for a moment in the 90's, about 2x as fast as x86 for similar die size / process / clock speed. I don't think we could hope for more then that, and that's at the expense of compatibility.
I'm halfway through "What Every Programmer Should Know About Memory" by Ulrich Drepper. His paper answered something that was bothering me, namely why don't we already have more cores? Contention for the bus to memory and the caches is the reason. "MOAR COREZ" helps with some tasks, but stalls on most, even if you have software that's written to try and take advantage of the cores.
If feature size is going to stall for a while, I can't help but wonder if we should be focusing on memory. Not more main memory, but faster main memory. I'm not clear on the penalty (in terms of density) for going to a different memory cell design, but if programmers would stop wasting memory we might be better off with, say, 2GB of very fast RAM then 16GB of the DRAM we have now. The penalties when you miss the cache are huge even if there is no traffic on the bus. DRAM is just fucking slow compared to the CPU.
>>53724359
Bruh.
Arm chips are as fast as core 2 quads now.
Just because it can't decode your one special codec doesn't mean it'll be any worse for average use like web browsing and multimedia.
>>53720552
le XD
