AMD zen 2 12c am4 part???

You fucks, I bought a 9700k, you said AdoredTV was wrong, I want my fucking money back

Adored pointed something similar out when looking at another code name for a potential Zen2 engineering sample.

>9700k
I am extatic about CES, but I'm not certain your Intel chip is a bad choice. At all.

ES0 = Engineering Sample, first stepping

Attached: adoredwasright.jpg (294x294, 20K)

NO NO NO NO NO NO NO NO NO NO NO NO NO
NOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

SOMEONE PLEASE PHONE CTS LABS PLEASE

>a literal who pointed something similar
so, fake news? yeah, that kinda makes a bit more sense :^)
lets see if anything worthwhile is mentioned/shown on ces

Anyone who says adored is wrong should have their opinion immediately discarded, even when hes critical of AMD hes always right.

>still mad that adored was right yet again

Attached: 1516796859570.png (552x661, 288K)

Very early chip, likely just to test out the 7nm process to see how it holds up. They'll have several iterations before the final consumer chip, clocks usually go up along the way.

Holy shit I didn't realize people were this assblasted.

someone on >>>/reddit/ pointed out that it could be a base clock that gets multiplied to achieve the true boost clocks

>Friend of mine gave an idea, what if the frequency is the base that is multiplied with the "factor" seen on the russian page?
>i1.wp.com/www.tech-critter.com/wp-content/uploads/AMD-Matisse-Ryzen-3000-series-specifications.jpg

>121 (base frequency) * 39 (factor) = 4.719 GHz - Ryzen 3800x
>121 (base frequency) * 42 (factor) = 5.082 GHz - Ryzen 3700
>121 (base frequency) * 38 (factor) = 4.598 GHz - Ryzen 3700
>121 (base frequency) * 40 (factor) = 4.840 GHz - Ryzen 3600x
>121 (base frequency) * 36 (factor) = 4.356 GHz - Ryzen 3600

>Assuming factor is the clock multipler.

Yeah ok that seems reasonable

(base frequency) * 42 (factor) = 5.082 GHz - Ryzen 3700X
FTFY
My pocket book is ready

that is retarded. REFCLK/BCLK are going to be 100 MHz for pretty much ever thanks to the PCIe spec.

>multiplier base clock
that might actually be the case
>plebbit
oh, nevermind

except the chiplets can have different clock domains than the PCIe block since those are entirely decoupled now.

or its just a VERY early ES

Intelfags trying to damage control their depreciated shit

>Friend of mine
>gave an idea
>what if
that will be a golden yikes from me my good sir

Attached: 1519326608368.jpg (364x364, 22K)

OYYYYY VEEEEEEEEEYYYYYYY

Not sure about this one as the other device ID we saw was 5D0108BBM8SH2_37, which is a matisse (zen 2) AM4 8 core 65w part with base or boost 3.7ghz. So where the 12c has a 1.21 multiplier this has 0.10?

12/12 isn't cores/threads, clocks, or a model number in this case, it's the configuration of the four dies, which is how they're segmenting their models now.

AMD isn't fighting to move product via aggressive segmentation like before, they're fighting to get the best margins off yields on a new process. On GF they needed to move a minimum amount of wafers because of their contract.

Attached: ccx.jpg (3885x2640, 2.19M)

>listens to Jow Forumsents

Attached: kektastrophe.png (400x350, 238K)

>comparing monolithic to nonmonolithic

Attached: 4L_usRmTjaz.jpg (1600x1200, 83K)

that's boost multiplier, 100 is base. still accurate though

R7 3700X 12c/24t 5.0GHz all core boost and
R9 3850X 16c/32t 5.1GHz all core boost confirmed

>4 dies
why would they have 32 cores available for desktop?

You listened to the Intel shills that post here, your own fault.

the GF contract was amended though, and the IO dies would more than meed it anyways.

The C stands for 12 cores, the 4 digits could be anything though, but I think you're right that it's probably something to do with configuration.

It's worth bearing in mind that we first saw threadripper 1 device IDs in mid-late march (techpowerup.com/231798/amds-rumoured-upcoming-16-core-part-to-reportedly-run-at-3-1-3-6-ghz), one in this case was 1D3101A8UGAF3_36/31_N , which is the same ES stage that the newest device is at. The 1950x was officially announced in early July and released in early August, so that's a 5 month turnaround from ES0 to release. Release clocks were 10% higher than on the ES ID too.

Depending on how old this latest leaked device is, this almost confirms a late march-late april release at best IMO. They'll probably announce much earlier at CES this time because they know Intel has nothing that they could release in the time until release that could steal any thunder unlike with the TR release.

Core dies from TSMC getting married to IO dies from GloFo on 16c/32t CPUs at 5GHz that aren't even HEDT
What a time to be alive

It'll only be confirmed when AMD says so. Rumors aren't confirmations. As much as AdoredTV has been right in the recent past, I'll believe it when I see it.

Clocks don't come out of thin air; they need to be divided down/multiplied up from some other signal or generated from a sufficiently reliable mechanism.

In practice, host motherboards have a 25 MHz crystal that gets 4x multiplied in the chipset then sent to CPU(s) and PCIe devices for further up-multiplying as needed. AM4/SP3/etc. platforms don't generate clocks in-socket, and the specs wouldn't even allow an egress channel for a Ryzen 3xxx with that capability even if they did.

As for chiplet-based Zen 2 in particular, I would be shocked if AMD strayed from the previous general clock domain structure, with the IF clocked synchronously with the memory controller between L3s and having domain boundaries at the point of the IOMMU/PCIe complex and within CCX L3s. And again, all those domains should be created by multiplying up from the 100 MHz BCLK.

>Literally who?
Fake news. Intel's frame rates are objectively better than ayymd's. Besides, intel 10nm is better than amd 7nm because intel uses premium, quality wafers. Amd cannot reduce ccx and die to die latency with infinity fabric. And we all know that applications never use more than one core.

2 of 4 would be dummy dies like the original threadrippers had, if they were going to do that.

INTLEL FAGS LITERALLY LIVING IN 2014 STILL LMAO

Attached: 1546137572176.jpg (690x720, 77K)

the whole point of that was pressure balancing under the lid itself, there wouldn't be much point to it on a much smaller socket and lid.

not to mention they don't even have the room for 4 chiplets + the IO die, even cut down to 1/4th as large as the Epyc IO die.

>you now remember the accidental leaks from MSI about their motherboards supporting 8 cores and up from the middle of last year
techpowerup.com/246100/msi-drops-first-hint-of-amd-increasing-am4-cpu-core-counts

It's all falling into place

Attached: easter.png (420x420, 7K)

>intel 10nm is better than amd 7nm
I guess that's why their entire product stack is 10nm now, right?
vaporware can never be better than production silicon.

Seething

>2019
>anything less than 8c/16t is pleb tier
t-thanks lisa

Attached: 1537156853826.jpg (1024x697, 60K)

Bro this guy said applications still dont use multiple cores and 10nm is better when it doesnt even exist yet. Hes delusional

Tim from hardware Unboxed is saying that AdoredTV made up the rumour

AMDISCOUNT DRONES BTFO

Attached: 2019-01-04 16_42_03.png (1907x179, 22K)

bless you user

Fuck TURBO AND FUCK BOOST CLOCKS

>1 core gets high frequency
>when no program or game in the last decade only use 1 core

he never said that

based

Attached: 1529567124804.jpg (757x627, 87K)

>a literal who made up the rummor
he didnt, he was just too dumb and ecstatic about it and chose to believe in the lies. for fuck sake, the idiot even believed in the blatant plebbit poster that faked some excel sheet to troll a friend (with the prices clearly all cattywampus)

why don't they say (((which company)) they're getting their info from?

youtu.be/r3y_cQww-cU

>user believes the fake reddit post about the leak
kek

aint clicking that link. forgot to install windows defender

>you know realize that MSI has the best B450 VRM designs because they are future proof

How did they do it? Are they BFF with AMD because MSI was the first motherboard manufacturer to spit on Intel with the first graphic designed AMD motherboard box?

us.hardware.info/reviews/4741/3/the-20-most-bizarre-and-innovative-motherboards-1999---2010-msi-ms-6167-1999-first-proud-amd-board

You have material ready for when it's all true?

Literally no CPU from AMD this year will beat 9700k in games.
Maybe next year too.

>>a literal who
>>Literally who?
>>a literal who
Do they pay you per post or per word user?

>Literally no CPU from AMD this year will beat 9700k in games.

source : your ass

10nm is bigger than 7nm you fucking brainlets. Bigger = better. Enjoy your 3000 series while Intel is on 9000. Maybe you'll catch up in 6 generations.

Imagine when amd goes full nuclear on Intel and makes a full 8 core apu + igpu on par with Vega 56 in a 100w envolope on a 5 with ddr5 on 7nm+ or 3-5nm?
Gonna be as powerful as current gen midrange pcs and high end consoles
2020s is gonna be glorious

>t-there's no way my source wouldn't tell me literally everything they know, or that they're too lowly to get real information this early about a new product

what the fuck is he on about, he says amd won't be launching anything at CES (probably true) but that they won't be announcing anything about ryzen 3000 or navi either, so I guess he thinks they're going to talk about enterprise and science shit (AI, ML, DL) for an hour at a fucking CONSUMER electronics show?

Tim from hardware unboxed spitting in his viewers' faces

I stopped my Patreon subscription to Hardware Unboxed because of this. I like Steve from hardware unboxed but little timmy probably suffers from Aspberger's Syndrome or atleast some variation of autism spectrum disorder (which would explain why he can't figure out how to link to a youtube comment because he literally linked to their own video (yes, i tried manually copying the URL like he said))

fucking moron

Attached: 2019-01-04 16_40_46.png (1765x562, 67K)

Imagine getting an Xbox one X GPU with 8 core 4GHz Zen2 for $250.

Attached: fnttan.gif (250x250, 992K)

I'm the ass, I can confirm
t. his ass

So Zen 2 leaks pretty much confirmed. Damn.

I wonder what's in store for Zen 3 then? 10% high clocks? chiplets on 7nm+ and I/O die on 7nm for lower TDP? PCIe 4.0? what else?

Best part a about that video was trying to discredit the rumours by looking at TDP. Specifically looking at two different parts with the same TDP as if that hasn't been the way they do it forever.

4ghz at 100w with Vega 64 7nm tier perf would be pushing it power wise even at 3nm tho
7nm definitely not
But yeah future is very bright till we reach the theoretical limit of nm/silicone at around 1-2nm in a decades time or even less
Thank fuck we won't have Intel + nvidia holding us back with shit rtx (amd already had their own superior dxr in the works just not for gaming) it hardware and gaming system building will actually be fun again I missed the crazy 80-00s when shit was actually moving at break neck speed not waiting 3+ years for shit to barely improve 30% over overpriced shit quad cores and dumb shit like the titan xp and v

And Tims "source" thinks that AMD is making no announcements at CES even though Lisa Su and the CES keynote page confirm that AMD will make product announcements, and thats beside the fact that a 12 core on AM4 is now 100% confirmed

ddr5

Intel is best, nobody needs more than single core processors

AMD pretty much confirmed both Ryzen and Navi will be revealed at CES (CES interview of Lisa Su stating that she's very confident in the upcoming MCM products), a release wouldn't be improbable.

Respect for Steve
Dislike for timmy

Dunno
Zen 4 is gonna be mental ddr5+ 5-2nm maybe?
Hnnng imagine Zen almost twice as fast + barely any latency on ddr5 6400+ my dick

>Zen 2 leaks pretty much confirmed
Absolutely not user. It's less than a week now Goddamnit. Just fucking pace yourself.
It's starting to seem probable, but this "confirmed" shit just sounds like false flag hype.

>Navi
That one is almost a little too exciting

IS IT CES YET GOD FUCK

What a fucking retard, Steve shouldtell that timmy faggot to stop posting

2019 is shaping out to be an amazing year for some upgrades.

There won't be DDR5 until a new socket comes to replace AM4.
Zen 3 will be on AM4, PCIe 4.0 may be supported but not DDR5.

lower voltage at higher frequencies, and a slight clock bump. basically what zen to zen+ was

at best, the 12nm APUs are released or are released within a month, if there was going to be 7nm release of anything but Vega (which we know exists and is being produced as MI60) then there would be more leaks and more details as products were being prepared to enter distribution channels

2020 is am5 ddr 5 tho that's barely a year away now

>DDR5
I doubt it. Faster PCIe though

Zen 3 on AM4 was confirmed by AMD too.
No DDR5 until Zen 4

how about a lane wars instead of core wars?

I want ATX motherboards to have 7 fucking full lane slots (if PCI-E 3 = full 16x slots and they don't share bandwidth with anything else)

>tfw i'll be moving from a trash beema A8 to R5 3600
just how big of a jump are 8 real cores?

Attached: 1546434443991.jpg (540x540, 20K)

Zen 2 will support PCIe 4.0 but only on a new board.
Screencap this post.

Can't really double core counts going from 7nm EUVL to 5nm EUVL. The might implement some of AVX512 (which imo it was smart of them to avoid so far, it's way more complicated than AVX2, really only the extra instructions for 16-byte and 32-byte SIMD registers are useful if the read/write bandwidth is limited to 32-bytes) or maybe just go with a smaller chiplet surface area for better yields/cost to compete with ARM.

However, once they get down to TSMC's 3nm, they'll be able to double the core counts across all products again, plus there would be an extra 15% surface area that could still go towards AVX512 if they decide to implement it in Zen 4 (or 5) on 5nm.

Of course only the newer chipset will support PCIe 4.0, not a breaking news.

98% of consumers only use single GPU setups though. Look at Steam HW survey. Those lanes are pretty much wasted for mainstream segment.