OH NO OH NO OH NO AHAHAHAHAHAHAHABAHAHAHAHAHAHHAHAHAHAHABA INTEL still a year away from their 10(12)nm AHAHAHAHHAHAHA...

Overreaching

AMD APU's are really good for what they are, and my A4 made for an excellent HTPC/emulator box

Laws of physics are starting to become a problem

Unknown, other than yields are poor. A billion different things can cause exaggerated defect density. Their light source could be too weak, high use of triple and quad patterning in critical areas could be pushing complexity too high. High number of mask layers alone already is a huge hurdle to overcome. Could be that they're experiencing delamination in a back end metal layer for all we know.

Intel's upper management set really aggressive goals that they pushed on their foundry business, and they were unable to meet them. intel has been behind schedule on their own foundry roadmap since 2014.

Thats real retardit, sir.

Doesn't have raytracing so its garbage

Isn’t retarded at all. You’re hitting atomic limits.

Its cute that you're trying, but you should stop. Its embarrassing.
The laws of physics are always a problem. Everything is a problem. That doesn't stop engineers from producing a workable solution. Just regurgitating a popsci headline tier response is a huge red flag that you're an idiot.

The source of intel's current woes isn't "laws of physics" their issue compared to the rest of the industry is trying to bite off more than they could chew while everyone else was refining transitional FinFET processes and collaboratively sharing IP. Intel couldn't manage to shrink their 22nm trigate down to 14nm without enormous delays. Broadwell CoreM launch was a failure. The first Broadwell desktop parts were delayed half a year, then made EOL after just a couple months because yields were so poor they were losing money on every die. It was one disaster and episode of mismanagement after another with them over the last half decade.
Samsung and Global Foundries both work hand in hand with IBM. They're able to to outpace a giant like intel and do it in stride because they have a larger talent pool to work with, they have more research white papers, they have more industry expertise. They're succeeding because they have management who actually know what they're capable of, not up promoted finance heads who are totally divorced from engineering.

Damn. Thanks for explanation

I bought 2 APUs so far: A6-36xx (I think, can't check now), and A10-5700. The problem with them is that their target audience is very niche: graphics is too weak to really compare with even budget discrete adapters, and CPU part is slower than even i3.
They are ok as an alternative, but all in all, it's preferable to buy Intel platform - there are faster as a CPU, and it's easier to find used mobos/CPUs if your current ones dies, or if you decide to upgrade.

is a 7nm threadripper / navi apu possible?

Sure. Likely as a consumer part? No.

The current apus are on am4 and there will be motherboards with good availability untill I'd guess 2022, even if choice dries up a bit at the end. Intels socket a year means that this generally isn't as easy as the years go on.

The target market for a lot of apus generally doesn't need that much cpu power in my opinion, but the raven ridge stuff has it in spades with a graphics solution way above intel's. The niche is smaller you're right but it's there. Only missing point right now is 6+ cores with igp that Intel currently has.

Don’t let this dude meme on you. “Solutions” are marketing bullshit. At atomic near quantum scale the amount of shit you can do rapidly approaches zero. Randomness will destroy an increasing percentage of your chips the smaller you go.

That would be so expensive to produce for such a niche market, there's no point in even trying. I'm sure they could, but it would take a toll on performance on both ends, and would end up being more expensive. HEDT products such as Threadripper aren't worried about doing secondary shit like graphics.

There are atom sized transistors made in labs. Everything is a physics problem that can be solved. It's like saying it's impossible to make transistors smaller than a millimeter or something 60 years ago because the human hand can't produce something that small. I would say there are a 5 to 7 years before that gets adopted.

>I have no idea what I'm talking about but I'll keep posting asspulls anyway
Cute.

Tell me all you know about muh quantum wall while you're at it. I love it when popsci idiots get so blinded by buzzwords that they think current leakage is a magic property.

Transistors are made of atoms, how can you make something smaller than an atom if the atoms making it are bigger than it ?

>I'll just go full ad hominem and expect it to prove him wrong

You're just as retarded mate.

Junctions can be made with sub atomic particles. FETs don't have a physical limit in size because fields aren't physically limited.

No you retard, there are transistors being tested in labs right now that can change their value with a single atom. It's completely different from an atom-sized transistor, which is obviously impossible.

?????

Yes you can make atomic sized transistors. The issue is mass production of them. Even with a perfect process you can’t have 100% yield.

NOOOOOOOOOOOOOOOOOO AAAAAAAAAAAAAHHHHHHHHHHHHHHHHH

Attached: 1506977173618.jpg (882x758, 324K)

>amd is on 7nm as of 2H 2018
>intel went from 10nm to 12nm because of process issues
AHAHAHAHAHAHAAHAHAHAHHAHAHAHAHAHAHAAHAHAHAHHAHAHAHAHAHAHAAHAHAHAHHAHAHAHAHAHAHAAHAHAHAHHAHAHAHAHAHAHAAHAHAHAHH

>apu
Did AMD really give up on even trying to fight nvidia?
This is actually pretty sad. I wish they never bought ATI, maybe things would have been different.

With Zen's scalability and connectivity the yields most likely won't be an issue, the CCX thing was genius

Attached: 1521504161165.jpg (1920x2160, 2.1M)

You know it's true user, if ATI was it's own company they would be releasing full retard dual gpu shit like back then and nvidia wouldn't have 90% of the market cap and being on total monopoly mode where nothing comes to their level in terms of performance because they got too much money to R&D.
Look at what that nigger raja did, do you think polaris and vega were remotely close to what they were supposed to be? Hawaii was the last good one from ATI and if you compare it to nvidia release back at the time it actually beat them on high res benchmarks.
Now proceed to post edited wojaks so you get your 0.2 rupees deposited in your Jow Forumsamd account.

iirc their newest cpus were giving worse benchmark results than their prior cpus
meaning not even 1-10% improvement for 2x the cost but -1-10% that is minus 1 to 10

Nah I just enjoy your butthurt, go buy Nvidia kek

yeah but intel is still 2x as expensive and amd is still at 20 bucks a stock

imo they're releasing too fast, why? why is amd not giving time for the market to adapt and for their public image to change? why are they forcefully outdating their current still new product? why are they intentionally wanting to lose money?

>it's premium shit

you should have bought more ATi so they did not become assimilated then

>why don't they do it like nvidia? why don't they make you wait two years for a new, better product, or why don't they do it like intel, which only gives 5% higher performance every year?

Attached: 1527533367941.png (221x250, 21K)

weren't consoles using some form of apu

yes, a custom one though, and they're still using the jaguar cores, which are basically bulldozer ones

>and CPU part is slower than even i3.
You're talking about Bulldozer APUs.

yes, he's talking about the a6 and a10 he bought

>jaguar cores, which are basically bulldozer ones
Jaguar and Bulldozer are two completely different architectures, both of which have been superseded by Zen. Bobcat/Jaguar was there low-power architecture, similar in purpose to (but, I believe, significantly faster than) Intel's Atom.

But he makes it sound as if he thinks it's still true for current APUs.

>25x20
>care about energy efficiency
>not implement LPDDR memory controller
They should try again

both are making more money

Problem is that since companies use AMD only in their cheap lines they'd just end up making AMD perform worse by only using LPDDR in their laptops. A bunch of companies released Ryzen laptops with single channel memory already.

That's not a valid reason for not supporting lpddr in mobile chips

The rumor is that management wanted to compensate for the delays in 14 nm production by reaching further than usual with the 10 nm process, which led to the engineers trying to use quite exotic tricks to accomplish that. Apparently, one of the main troublemakers is using cobalt for the lower levels of metal interconnect, needed to control electromigration at the smaller scales they attempted, but came with its own set of unforeseen problems in pattering, which is said to be what is keeping their yields so abysmally low.

> doing it 100% for the money and not because you can or because you care about giving something to the world
Fuckers like you are why capitalism went straight to hell.

Attached: 1520806764946.jpg (499x497, 50K)

>everyone is not doing it 100% for the money
what lala land do you live in?

Capital was supposed to be the means, not the end, yet greedy fucks came and fucked it up, that's all I'm saying.

>7nm apu

Attached: apustus.png (657x527, 352K)

Imagine being this retarded

It's probably a "pipe cleaner" for 7nm. The yields will not be great, but for products that doesn't sell in incredible amounts it will be enough.

This is a tough time for intel

Attached: 5753977823906.png (400x400, 109K)

>yet greedy fucks
im sorry but im neither a banker nor i have anything to with the fed, sadly

WHO is not doing it for the money retard?

>doing it for the money
>doing it solely for the money

Diamonds are fleeting, a good rape is eternal.

yeah they're doing it to pay for their next holiday or car or dinner or whore or whatever
how can you be this retarded

>want to build a new pc
>can't find a mini atx mb or a boxy matx case
reeeeeeeeeee

>mini atx
itx

Vega 7nm is the pipecleaner, announced for the second half of 2018.

anandtech.com/show/12910/amd-demos-7nm-vega-radeon-instinct-shipping-2018

But it will only be for machine learning and server farms, so the gaymer will need to suck more dicks to afford nvidia's RTX cards.

Attached: manofculture.jpg (448x401, 20K)

Attached: Screenshot_2018-08-22-10-51-54.png (1080x1920, 204K)

Because they want to completely eclipse their competitors. With such a huge mind share for Intel (every normie knows about an i7 and almost none about Ryzen), AMD has to be significantly better and cheaper to actually be noticed. If AMD was only 10% better than Intel, the vast majority of people wouldn't even think about switching.

Oh no no NO NO I will be only buying it by the end of the week.

>AMD answer to nvidia is a 7nm APU
>A fucking APU

ayymd strikes again, can't wait for anons with buyers remorse saying this shit is faster than a 1080ti just like they used to say bulldozer was faster than nehalem when it only had half the IPC lmao. Praise mommy su.

Holy fucking shit Boyz I was right about to buy the 2400g this is fucking insane

I LOVE YOU AMD

Attached: 1528763804263.jpg (553x441, 39K)

They tried 2.7 density Vs 2.4 of skylake/kabylake

And saw very poor yields

no you stupid nigger
the 2400G was released in Q4 last year, it's only logical they'd release next gen apus 1 year after it

wait i get it - they have a good core - zen - and they want to provide as many buying options for purchase as possible

which one of you did this

Attached: nvidia-or8.png (370x92, 6K)

2400g APU was quite literally insane delivering performance better than an i7 haswell with one of the most powerful igpus in the buisness with Vega 11 which is equal to the 1050 now if I am not mistaken.
If the next 7nm APU could deliver a better performance especially in single core with an igpu that stands to to toe with the likes of 1060/970 then we have a real good game changer over here.
igpus are the future people.

Attached: Justincaseanyonewantsit_4848ccf6cb62a1143e020d320dba3913.jpg (1000x750, 206K)

literally me

>Vega 11 which is equal to the 1050 now if I am not mistaken.
you are mistaken, v11 is only slightly faster than the 1030

For AMD to compete against Nvida they first need to make themselves big enough in the CPU markets. That means EPYC, Threadripper and desktop comes last. Then comes laptops another large area to be exploited that is ruled by Intel. Their GPU line is aimed squarely at the pro markets. If they can get their foot in the door of all these markets then they can start making inroads into the consumer GPU market again. They will spend time refining their GPU for consumers when they have something worthwhile and the market is stagnant (Nvidia being a monopoly driving prices through the rood might piss off enough fanboys that when AMD comes out with something cheaper but just as good or better then consumers may at last see the light).

no it was me :)

no

I wouldn't get too excited about this particular one, being a ryzen 7 would mean it's gonna be pretty expensive.

But I'm enjoying my 2400g for the most part. I haven't used it in an extremely demanding game yet, but it still holds over 30fps in downtown boston in FO4 in 1080p. It's crazy how big of a performance hit that place brings.