RYZEN IS ON THE HORYZEN!!!

RYZEN IS ON THE HORYZEN!!!

Other urls found in this thread:

zolkorn.com/news/amd-ryzen-sr5-will-be-8c-8t/
twitter.com/SFWRedditImages

inb4 intel destroys them 10 days before they release their shit

the problem with AMD is not that they are bad per se, they are good

it's that they have no Foundries of their own. They cost kazillions.

too late, they fucking waited like 3 years to finally update their fucking architecture

...

we already knew this

>inb4 intel destroys them 10 days before they release their shit
With what? Cannonlake(Wow, dieshrink Skylake and 4 cores!) on desktop in mid 2018 if their yields aren't 1.7%?

Yet Vega comes out in like 4-6 months.

the day when jewtel gets blow the fuck out

Vega 11 is due May, and Vega 10 is confirmed to come out before Vega 11

can u give me salsa?
>inb4 currytech rumour

why are all the tech news sites shilling this rumor about no 6 core ryzen? other than some stupid table on some dumb wccftech-tier chink website i don't see any good reason why there wouldn't be

The low level tech media all run the same stories, its an echo chamber.
CanardPC on twitter were talking to some OEMs had at present at 65w 6 core samples on hand.
Sure its possible that AMD may not have 6 core SKUs available upon launch, but that doesn't mean its impossible for them to make 6 core chips.

Computex schedule leak, mentioned recently launched midrange line. So latter half of May for Vega 11.

Vega 10 earlier because of it passing RRA months before Vega 11, and AMD stating it's 2017 1H awhile back so no room after Computex. My expectation is a spoiler launch in time for GTC.

Fuck knows when Polaris 12 comes out, but that's just an Oland replacement anyways.

thanks user

zolkorn.com/news/amd-ryzen-sr5-will-be-8c-8t/

That's their sole fuckin source. Don't listen to the echo chamber. Wccftech has Zero credability.

Zen is made up of 4-core modules, so depending on how the modules are designed it may not actually be possible to create 6-core versions to begin with. Hopefully it is, however, and the stories currently circulating have no good source.

Doubt it. It shouldn't be like the CMT cores where you can only disable them in pairs. The front end and FPU between cores isn't shared.

No design company needs their own fabs, and the current players have all caught up to Intel. Competition no longer has to work with Big Daddy Core being a node shrink ahead. The major foundries are also matching Intel's node shrink schedules. This isn't the same game you know from the past 20 years.

No they didn't wait you fucking retard, they were designing it all those years, they didn't come up with the idea a few months ago

Why would there be?

I don't believe that. Big players still have to charge a premium, Samsung won't work for free for AMD for example. It is a logical thought but I don't see it in practice making chips much cheaper yet.

But absolutely none of the other foundries can match Intel's density and scaling. They just make up whatever name they wanted to catch up to Intels manufacturing lead.

Intel can also beat them in volume even in the theoretical scenario they were matched in technology (which they aren't). For example, they could go to 6cores or 8cores without breaking much sweat if they were really cornered. Having your own Foundries is a hell of an advantage.

>intels manufacturing lead
>14nm node performs so poorly they had to cancel a whole product line after delaying it for a year
>they'll still be using it into 2018 because of all the problems they had
>their 10nm node will be used for at least 3 years as well

Back end scaling by itself means nothing. Trying to push area scaling to extremes to get more candidates per wafer while causing the side effect of horrid yields is not having a manufacturing lead. Its called falling behind.
Intel is faltering because Kryzanich is a terrible CEO.

> we reach the limits of the quantum mechanical theory
> photolithography is also scientifically at its limits

> Intel's fault
=

Stupid AMDniggerpoojet

it is

Intel will still be king of density and scaling. Garlic eaters can put smaller transistors on the same size interconnects all they want, no one really believes they have any kind of advantage.

You assume they don't have anything in the top secret lab.

lol, ok, we have a badass over here that will defy the laws of quantum mechanics and overcome all the severe photolithography limitations that dozens of Conglomerates can't solve easily.

Don't talk out of your ass.
Intel's 14nm node being problematic has nothing to do with any meme science buzzwords. They paid the price of trying to get a node and a half shrink work of scaling out of a single node. They required double and in some cases triple patterning in certain areas to get the resolution required, and they did this so they could present PR fluff to investors. There is no real reason for it. They even developed a variant of the process with a slightly denser back end for mobile chips.
This is an executive single handedly giving orders to make stupid decisions.
Contra revenue was a stupid decision.
Sophia was a stupid decision.
Pressuring the foundry business to meet unrealistic goals so they can blow smoke up the ass of their investors is a stupid decision.

Broadwell CoreM launch was a mess.
Broadwell desktop got delayed twice.
Kryzanich insisted that they were having no foundry issues, then only one week later he admitted they were a quarter behind schedule in where they thought yields would be.
Despite that desktop Broadwell was further delayed, originally only launching with two main SKUs.
Production of the Iris Pro equipped desktop Broadwell chips were soon halted since they were losing money on each die.
They were forced to rush Skylake to market, and it took half a year before they could yield enough high clocking dies to start stocking vendors with desktop Skylake.

This is Kryzanich slowly strangling intel because hes a fucking moron. He needs to be removed by the board.

Stay in Sup Forums, kid.

Back end scaling alone means nothing. Having a denser chip doesn't do you any good when your yields per wafer are still low, and the competition is on par when it comes to electrostatic characteristics on the FEOL.

BUT ITS THE FUCKING JEWS HIDING ALL THEIR SECRET ALIEN TECH
WAKE UP SHEEPLE

Did you miss I also said Photolithography has severe limitations? What is Patterning? Part of Photolithography.

You're just pulling words out of your ass because you don't know what you're talking about.

OK, we wait for you to prove that the Patterning problems you mentioned, aren't Photolithography problems. We'll be waiting all day. All you want to do is to say "shut up, I know those things better".

Yes, you do know those things better.

Here is your prize, fuck off.

>I got caught talking out of my ass
>now I'm going to shitpost like an autist

Why is it so hard for people like you to just refrain from posting about subjects you're totally unfamiliar with?
What compels you dipshits to gleefully spread misinformation?

I don't go on /o/ and shitpost about Wankel engines, so why the fuck do you do this here?

actually it seems like the point he's trying to make is that the CEO is making really bad decisions

YOU brought up the more technical aspects of the conversation and now you're backpedaling because you assumed this user would be as ignorant as you are and you were wrong.

Fuck off and die cunt. Just because you know something better it doesn't mean someone might not know the big picture well.

Answer the question now cunt.

Why is Photolithography easy?

He won't answer it. Patterning is part of Photolithography. He's just trying to be a cunt.

>The token autistic fallacy defense: The Post

Y'all are both faggots

why don't you solder your own CPUs?

> I can't answer the question: the post

the photolithography isn't relevant to the larger picture, which is that Intel basically said "look how wicked cool my teeth are" to investors and then ran face first into a brick wall

You are more concerned with how hard they ran into the brick wall rather than the fact that running into a brick wall is a fucking stupid idea.

He's a cunt. He doesn't want to talk about CEOs and Business, he wants to talk about the technology. I have encountered him in another thread last week.

All he wants to say is that he knows the details better but nobody is allowed to speak.

He's probably a sad little bitch in real life that is scared to speak up.

As long as it's price competetive, like piledriver was, then I'll be happy.

No one claimed any part of fabbing a chip was easy, you're only latching on to this because you have no ground left to stand on.
I called you out for talking out of your ass. For spouting buzzwords that you know nothing about.
Now you're desperately clinging on to this last straw, like the shitposting loud mouth autist you are.

You tried to make that argument that, at the 14nm node, intel was coping with some intrinsic scaling limitation. You woefully tried to mention a quantum mechanic force being a reason, which is still wholly untrue. The force you undoubtedly wanted to mention is uncontrollable electron tunneling. However this isn't an issue anyone is having to deal with. Leakage of FinFET parts is still lower than all prior planar parts. Intels yield issues had nothing to do with this, and everything to do with management pushing engineers and tooling well beyond a single node shrink in one leap for completely superficial reasons.

You literally have autism.

>You literally have autism.

I'm going to be honest. This is projection. You barge in demanding nobody to have an opinion.

You have to relax. Everyones sees that you know stuff on CPUs. You don't need to demand full attention.

Hence you might want to look into your tendency to demand in a tantrum like a 5 year old for everyone to shut up for you to exist.

>He's probably a sad little bitch in real life that is scared to speak up.

your name is John right? and you hate every single one of us?

dude you're a fucking joke

see

$700 for a CPU that gets beat by a $300 Intel in all games? I'll pass.

Autist, my posts are all well thought out and coherent.
You however do nothing but shitpost and use reddit spacing. Having an opinion about something is not the same as talking out of your ass and pretending to be a factual voice of authority. You got called out for talking out of your ass, and that hurt your little autistic ego.
Get over it, and don't post unless you know what you're talking about. Which means you should probably stop posting all together.

>reddit spacing
Dude, yours posts make a lot more sense than his, but LOL.

For all your supposed superior intellect you have to google some terms as homework. Autism and the Asperger Syndrome is typical of individuals that may also be high in Intelligence but with severe problems in social interaction. Regardless of what I am, look into your mirror for a second, and see that your behavior is criminally stupid.

You barge in, you see someone that knows something not in much detail compared to you and you try to shut them off completely from your existence.

I fear you might also have criminal tendencies. You remind of ReiserFS. The guy that was super intelligent in filesystems, and killed his wife.

Vega 11 is small vega, just like polaris 11 is small polaris. If this quote is reffering solely to vega parts that would mean vega 10 is the one to release in May with 11 coming sooner.

Which seems unlikely to me. They haven't even said anything official on vega 11, and it would seem more likely for it to come later like polaris 11 did.

How do you know they're not reffering to polaris itself with that quote? AMD themselves consider the 470 and 480 midrange cards.

can you fucks use trips, I can't fucking tell who is who anymore

I agree with the fuck who was talking about decisions on the business side of things being responsible for the poor 14nm process.

Read: recently launched midrange line

Vega 11 is midrange, Vega 10 is in no way midrange. Is Polaris 10 recently launched?

Theres your answer

I'm going to let jessica know how she's doing.

Too late. I already did. I thanked Jessica for helping out multiple individuals at once through Sup Forums.

Technically yes, polaris is recently launched as it is AMD's most recent architecture launch. Regardless of the fact it would be nearly a year later, it would be "recent" for AMD.

For all we know they'll start a new series and just carryover Polaris, launch Vega 10 as their flagship, and bring out Vega 11 later to fill in the gap.

They did exactly this with the 200 and 300 series.

Will 2017 finally be the year of EUV lithography?

>falling for the multiple patterning meme

People honesty believe Ryzen will be better or cheaper.

It has to be, AMD goes out of business otherwise

I think it will be cheaper and almost as good.

Problem is i they are even 10% off the mark they are still generations behind because of the taper in performance gains.

Nobody wants a brand new cpu that benches the same as an ivy chip no matter the cost.

I still don't buy the whole "it'll be $350 for the 8c/16t part guys!" BS. This is from a single quote and is honestly open to interpretation. Plus it wouldn't be the first time a CEO didn't choose their words wisely when confronted with a reporter.

Lisa Su is a baddass business woman for sure, but nobody's perfect.

I withhold belief the cheapest the 8/16 part will be is $500 and wouldn't be surprised if it came in a good $100-200 above that. That part is, afterall, targeted more at prosumer/ workstation audiences.
4c8t parts are where gamers should be interested in as that would in effect be an i7 7700k for probably less than an i5 7600k.
8c/16 ~$500
6c/12t ~$350
4c/8t ~$200-250
Would be more than fair.
There's also maybe the possibility of parts with SMT disabled but some user im another thread claimed that would be very unlikely with the design of th architecture, but I'm nowhere near knowledgeable enough to touch on that.

Yep. If you search an archive for the words 'tech illiterate retard' you will see that he has been shitposting and shilling for his favorite company (AMD) for like two years.

If AMD isn't lying about the performance then they will be pretty much the same as Intel's pricing. If they are (and AMD are masters of making up bullshit hype) then I would expect delusional levels of pricing like the first bulldozer chips were until they're forced to discount them a few months later.

2017? No. TSMC has stated that they will start ramping up a 7nm node utilizing some early EUV machines in 2019 though. Some fanboys on Anandtech are currently flinging shit at one another over whether or not they'll get volume production before 2019 or not.

GloFo has some EUV machines though, and they've stated they're going to start work getting their 7nm SOI FinFET process up and running later this year. Barring EUV light source the process relies on absurd quad patterning, so its yet to be seen how that'll play out for them. They might have some fancy IBM IP for negating the exponentially increasing complexity of adding a 4th pass, but thats some pretty abstract guess work.
2018 for early samples is more realistic, but still on the optimistic side for a high performance part like a CPU or GPU.
Global Foundries having the first EUV parts on the market would certainly make for good PR, especially after acquiring IBM's foundry business and merging with it. Its still very much an emerging technology that will take years to really iron out.
450mm wafers would be a bigger cost saver, but as far as I know every foundry shelved their plans for them until 2018 or later.

How much of an idiot do you have to be to think a common insult thrown around here is all one person?

it's not going to be anywhere near LGA 2011 prices that's for sure

complete suicide otherwise especially if the lower core count chips can't match the i5/i7 clock speeds

Seriously. The other dude may be a shit flinging retard talking out of his ass but there's nothing wrong with good sentence structure and the use of paragraphs.

Sometimes I wish Sup Forums supported indentation just to clean up longer posts a bit and make them less annoying to read.

Same posting style and same self-absorbed attitude. It's very obvious to anyone who isn't an autismo retard.

I think theres totally an autist here with an attitude problem. Its this guy >
Really, the fuck is that spacing? Its like some imgur meme or something.

FUCK YOUR DRAMA BULLSHIT GET HYPED

It would be difficult for them to lie and get away with it through their presentation method. Afterall it was presented to investors and lying to them is a very dangerous move. Not to mention an unbiased third party achieved some respectable performance figures on an engineering sample at a humble 3.1-or-2-ish clockspeed.

They'll still have to offer a commendable price/performance to overcome the mindshare intel enjoys, though I agree they'll have to test the waters and probably shoot for higher than people are ultimately willing to pay.

Too many sources either confirming or nodding to the performance shown in AMD's two demos. Before CanardPC published their magazine preview a bunch of OEMs on various sites have been all but breaking NDA. A while back some guy even showed off his own OEM sample, a 3.2ghz base, 3.5ghz turbo chip.
If there were any huge performance discrepancies they'd be well known. The type of publicity that would get would be worth breaking an NDA for to publish early and get web traffic from the entire internet.

>spent the last 45 minutes reading about EUV
>becoming more sceptical with each article

Honestly believe that SAQP will be the standard at 7nm unless ASML get their EUV source to 250 W+ & at high availability soon.

It's called sentence structure. You learn it in writing portions of English classes from pretty much the time you start writing up until you graduate high school.

Calm down buddy I'm agreeing with you here but there's absolutely nothing wrong with separating groups of sentences from others in the form of paragraphs. It actually makes the body of text more attractive and easier to read.

This is off topic though, let's get back to talking about Zen cause FUCK YEAR SHITWRECKER KELLER GONNA TAKE INTEL OUT BACK LIKE OL' YELLER.

Power output is the limitation, but without it we're looking at the average cost of a chip probably going up. It only 20% of a 7nm wafer are good candidates then low end parts either have to be heavily subsidized by the sales of other parts, or they just don't get sold in the low end price bracket.
With tech entering mainstream in 2~ years one would hope that this sort of thing would be viable. Its still pretty sketchy.

Single sentences aren't paragraphs, and constantly breaking up posts into three lines seems like its either a meme, or some mental problem.

>tfw my portfolio is literally around 40% $AMD, 40%$NVDA, and 10% $INTC.
>tfw when I'm beating the S&P meme 500

Thanks Sup Forums, memes have made me around 4k so far. Better than betting it all on red or some stupid shit

Single sentence paragraphs are generally frowned upon, yes, but their ultimate purpose is the separation of ideas. Separating chunks of sentences which work together to convey meaning from chunks of others that lead into a different idea helps give a body of text flow.

I admit my paragraphs probably look retarded on desktop versions of the site but I'm mobile posting and this shit is so ingrained in my head I can't write any other way. This paragraph is an example of what I was talking about, by the way, as I've gone from explaining what a paragraph is to why mine might look odd from another perspective.

>but I'm mobile posting

:^)

I flat out stated that AMD was guaranteed to hit $10 per share before the end of the year, and they might even hit $12. A bunch of shitposters scoffed at it. AMD did reach $12 per share. Their loss.
Now lets wait and see if they hit $8-10 billion in revenue for fiscal year 2017. Should be a great year for them.

Single sentence paragraphs while also autistically shitposting is a double no no.

>3 biggest companies in their industry

Yes geez you sure needed Sup Forums expert opinion on that.

for amd to not have 6 cores, that would mean there is an architectural failure that would demands all or nothing form half the cpu, as incompetent as amd can be at times, I don't think keller would make something that shit. honestly wondering if there are going to be steps in between like a 7/14 core, of if amd is going to put them out like the phenom series where a dual could unlock one or two more, that could be a selling point if they do.

>mfw Zen will become the 8350 of this generation
Count me in.

Allegedly the CCXs just have to remain symmetrical.
IE: A six core chip has one core binned off in each of the two CCX structures.

THERE Will NOT BE 3C OR 6C VARIANTS!

500$ max for bottom binning, with i7 price math more likely due to amd even saying the 350$ range was on the table publicly.

I can buy that, but there needs to be a good reason, as a 7/5 core could give them more ranges they could hit, I don't say 3 because lets be honest, if you fuckinged a chip to the point 7 cores are dead its scrap. though a 3 core 6 thread distitute build does have its place.

It that is the case, CCXs needing to maintain symmetry, it could indicate a nuance about the L3 cache interconnect, and the buffer used to ensure the consistent average latency they talked about.

The frogs already benched the worst case scenario cpu, a 3.15ghz throwing errors piece of shit, it's better than intel's 6 core cpu, and with just the clock up comes dangerously close to intel's 8 core, without throwing errors may match or surpass it.

at ces, amd said that the range for the 8/16 thread was either 500$ or matching i7. Now considering the size of the cpu, they could match the i7, considering what they had to leave out, avx512 and quad channel memory, the 8 core just cant command the same price as the 8 core or 6 core because if you price it there, you are in the range where people need those features, there may be a black edition or some binning for a 500$ chip but i have doubts it will be there. and then you have the last bit, all enterprize chips are gotten from form the same design, just mcmed, meaning amd likely will give consumers the shit binned chips, so there we have all the reasons why it may fall to the i7 just on chip reasons alone.

for market reasons, amd can sell it at i7 prices and make profit, if they sell it at i7 prices, that WILL convince people to buy it apposed to the i7, and then the full line up below it, which will hit each of the i lines will be VERY compelling if not calling you retarded for going with the i chip instead.

but amd marketing could easily shit the bed and say 'teasing the i7 price match point was eye catching, and putting it out at 500$ is good enough for people as its a solid drop in the price of performance' which lets be honest, it is, but that effects EVERY chip in lineup by pulling the price up with it, and making it less compelling to buy when you can get the known quantity that is intel.

.......so what has avx512 ?

Source?

See the problem here is a few fold, amd may have the compute to play with intel's 1000$ lineup, but dual channel memory, no avx 512, if you are looking at amd or intel and you can use either of those, you are going the 6 core or bite the bullet on the 8 core, possibly even write it off as a business expense, so the 8 core is not apealing.

amd does not have the better process currently, so intel likely clocks higher, and that matters more then an extra 2 cores in most games, so slit your wrists on the 6 core price,

and because its a known quantity, you are likely going to buy an i5 over an amd 4/8 at that price.

-------------

playing that price match game for amd is dangerous, because intel can rush a new lineup saying 'thanks to improvements in the process, we now have a way to give you more for less, and drop everything down a tier' would intel like to do that? Fuck no, but if amd forces them, they can easily do it, and then with cannon lake increase the price a bit if not out right releasing cannon lake asap

but if amd drops the 8 core at i7 prices, intel has fucking nothing they could drip their entire e line down to the 200-300 range and leave the i7 at the 150 price point, and then have their 10 core be 1000$ but would they do that? if they go on business like usual, they wont have an answer with cannon lake either, this would be a multi generation win for amd, and that's assuming they don't bring out zen+ around the same time frame of when intel's next chips should drop.

this is just the consumer side which amd has written off, enterprise they will have mcms that have quad to octa channel memory and enough compute power to crush intel and from the demos, at a better power point too.

512 bit extensions for the avx2 instruction set.

a 'leaked' slide
only souce are clickbait sources and places running with the rumor because amd is hot now.

You are so full of shit. You know nothing more than some wccftech article.

There is no consumer chip with avx512

Your analysis is absolutely worthless

No one pay any attention to this LARPing faggot

$150-$200 four core i7 performance and I'm sold

>inb4 intel destroys them 10 days before release
Not happening. Intel's next line (Coffee Lake) will be cheap as fuck though. Unfortunately it will be the last on 14nm though. The 10nm line (Cannonlake) will come out right after Coffee

If the 570 is >470 perf with the Vega arch, it'll be good.

The most concerning thing is that they might be artificially creating the 8c / 8t chips for product and price differentiation.

>CanardPC on twitter were talking to some OEMs had at present at 65w 6 core samples on hand.
Hope so.

>8c/16 ~$500
>6c/12t ~$350
>4c/8t ~$200-250
I doubt the 4c / 8t variant will sell for more than an unlocked i5, which is currently at about 230USD. The 6v / 12t (if it exists) will probably run at about i7 prices, aka 300USD.

I very much doubt that they won't have something somewhere in the $75-150 segment

They will, but that's Raven Ridge, the APUs

the entire 6 core + intel line have it.
amd prices at 500$, they are playing with the prosumer lineup, and coming up short.
amd prices at 350 with a binned chip at 500, no one really pairs it up with an e line, just the best i7.

If amd do go for the 500$ price point or 350, intel will likely respond by helping pro software incorporate avx512 support into their shit, so amd is playing a game they are going to lose in the long run if they price high. not only would intel be able to effortlessly price match them, with an instruction set they don't have and memory throughput they don't have the capability of outside of mcm parts, and without amd having the sphere processing power to brute force avx512 it would fall short of everything in the e line.

look, amd has one chance to launch this at the right price, and that right price is i7 match at minimum with a binned at a higher point, this is the case of better to sell 10 at a 300% profit rather then 2 at 500%

then outside of that, how many gamers look at intel's i7 and think, fuck that, the i5 is just as good, or in many cases, why get the i5 when the i5 is just as good? amd made a public gaming event showing off the ryzen 8 core 16 thread.

They knew the price right there, and then they teased it again at ces. at this point im id be willing to bet money on the minimum 8/16 being in the 300-400$ price range, if not just straight price matching the i7 to fuck with intel.