http://www.guru3d.com/news-story/am d-ryzen-processors-reach

Also note that going all the way to A3 for retail is very unusual, guess AMD wanted those high clocks at any cost, even releasing 6 months later.

>>58234946
Sounds pretty nice desu! How do you think it will scale? (regarding clock speeds)

>>58234946
source on the stepping info?

muh dick and will it run every OS, also shit like mac os x?

>>58234946
>inelite
Man, that brings back memories.

Awful, awful memories.

Damn AMD!

Release Bristol Ridge and AM4 enthusiast motherboards already. Then you can take your time and release actual Zen chips 6 months later...

>>58234946

Too much hype, not much basis.

File: inelite.png (90KB, 1683x697px) Image search: [Google]

90KB, 1683x697px

>>58236329
oh man

>>58238048
>tfw always right and almost always wins threads

>>58238070
you can put your trip back on, chris

>>58234946

>muh gigahertz

Because a 3ghz P4 is the same as a 3ghz 2600k

>>58238161
>3.15ghz engineering sample with 3.4ghz boost is able to match an intel chip with up to 3.7ghz boost that retails ~$1000
>said intel chip has trouble getting much higher clocks on all cores
>implying it wouldn't be incredible for a chip that already performs better at a lower clockspeed to also be able to run at speeds higher than the competition can hope to reach, especially on air

It aint all about "hurff durff much ghz," it's about how BFTO intel would be if this turns out to be true.

Can everyone just FUCKING STOP with the "COMPANY WINS" or "COMPANY BTFO" shit?
This supposed "GPU war" is bullshit.

No one really believes that AMD will take the performance crown from Intel in the CPU market or from Nvidia in the GPU market.
The point of AMD (at the moment) is that they keep the others in check - pricewise as well as servicewise.
They have not done that for the last ~3 years, but now with Zen as well as Vega coming up (Polaris actually being a decent stepping stone), we may finally have good competition again.

>>58236250
Charlie.

>>58236329
IN Elite shitposting was art. Now you have dumb frogposter redditshits.

>>58239286
I miss SilliconDoc too.

File: 1.png (872KB, 1192x9543px)

872KB, 1192x9543px

>>58239363
best times of /g/ shitposting

File: 2.png (1MB, 1192x9774px) Image search: [Google]

1MB, 1192x9774px

>>58239421

File: 3.png (1MB, 1192x9150px) Image search: [Google]

1MB, 1192x9150px

>>58239434

>>58239363
Red Roosters and communists, right?

>>58234946
Source on the revision.

https://forums.anandtech.com/threads/zen-es-benchmark-from-french-hardware-magazine.2495505/page-20#post-38655642

CanardPC had an A0 sample to play with. First batch out of the fab, buggy, yet stable enough to play Doom on it back in June.

AMD is on their third revision of Zen at the moment, A2, with A3 on the way with further fixes and tweaks. Bug fixing, tweaks here and there, and clocks going up. This is why launch was pushed forward from Q4 2016 to Q1 2017, these things take time to get fabbed. AMD must've had their good reasons to do this.

CanardPC's numbers, again, are based on an A0 sample. That's old, and performing like that.. imagine where AMD's jaws were when they started testing the first engineering samples. Latest samples are doing "higher than expected" frequencies according to bits n chips.. who've had a good track record on Zen so far.

File: C0zYKC1WgAA8RVw.jpg (125KB, 849x712px) Image search: [Google]

125KB, 849x712px

>>58234946
Also to cement the reliability of these frogs.

From March.
the binary translates to "Intel GPU = AMD"

Which they published months before we even got word of Intel licensing AMD's GPU tech.

For what reason did I buy a fucking Core i7-6950X if Zen can overclock!? FUCK YOU AMD SHITTY FUCKS FUCK FUCK

>>58239674
>AMD's at fault because I have more money than sense

>>58234946
It'll be interesting to see what clocks they can achieve for a 2018 refresh. I bet they'll have base clocks near 4ghz inside of the same 95w envelope.
Zen+ is going to be ridiculous.

AMD is SHIT and GloFo is SHIT.
There's no way this is true, it's just another disappointment like the last 15 years.

>>58239977
I have no doubt Kabylake-EX will be close to a 4.0 base as well.

>>58239992
It would be insane if it was clocked that high.
If you look at the technical side of things and extrapolate how much power each core is pulling at a given clock speed things are getting pretty ridiculous. I'd like to see voltage/clock scaling curves start to flaten out and become much more linear. If we can push clock speeds higher without sacrificing IPC in unduly long pipeline designs then we'll see big gains in serial performance for at least another generation or two.

>>58234946
>This is the same magazine that posted those ES benches a week or so ago.
ES?

Help a nigga out here.

AYYYYMD IS FINISHED AND BANKRUPT

>>58240046
ES stands for Engineering sample.
CanardPC got their hands on a super early test chip and validation mobo from a contact they have inside a mobo manufacturer. They ran a bunch of benches on the hardware and published them since they weren't under NDA.

>>58240066
Thanks anon.

>>58239612

It is suc ha shame that as a rule /g/ is too stupid to understand what that deal is all about.

My personal conspiracy theory as to why Intel has gone this route (do note that this doesn't cover everything - on the gpu front AMD and Nvidia have basically all the patents) is that Intel has realised the sort of resources required to catch up to AMD and Nvidia and have said fuck it, we'll just license some of the tech they are willing to share and save us the headache of reinventing the wheel.

>>58240055

Tripled their stock value in 12 months and its still rising - proof that bankruptcy is around the corner.

>>58240131
All bubbles burst

>>58240120
Intel just don't have chance to get into GPUs, almost every avenue has been patented by either AMD, or nVidia, or is shared between the pair of them. The only way to get into the game is to license the tech from them.

>>58240165

If you think AMD's stock is going to crash you really need to read their portfolio more. AMD have secured quite a few contracts securing them long term income, they recently wiped out a huge portion of their debt and all in all investor confidence is high.

Then again what do people who invest millions into AMD know? /g/ clearly has more insider knowledge than these professionals.

>>58240186

Pretty much - it is a lost cause for Intel despite the effort they have put in (and equally its what makes zen based APUs potentially very dangerous for Intel).

File: 7v5BT5G[1].png (92KB, 1653x931px) Image search: [Google]

92KB, 1653x931px

>>58240165
Speaking of bubbles then.

Using your own methodology here.
>Stock value tripled in past 12 months
>Rising every day
>Bubbles must burst

NVIDIA FINISHED AND BANKRUPT CONFIRMED HERE FIRST GOYS

>>58236329
SOMEBODY CALL 911

/g/ was at its best when that was playing in the board.

>>58240205
APUs have been a danger to Intel for some time now, there are clear improvements with every generation, and Intel can't do a thing about it without putting an Entire APU's value worth of expensive L4 onto a chip and selling it at insane prices to cover those costs. Intel simply cannot compete with current APUs, and those APUs are getting better, and with DDR4 becoming more popular it's only a matter of time until that memory works its way to an APU, and all but removes the issue of the memory bottleneck.

>>58240233

They have been dangerous to Intel in so much that there is a clear trend once cpu performance catches up Intel physically cannot compete - when such an event would happen though has (and still is) the great unknown.

Plus we know Intel basically gives their chips away to the big OEMs so AMD breaking into the likes of the laptop market will be very difficult. I do wonder though how readily Apple would ditch Intel and go all-in with AMD.

>>58240254
>Plus we know Intel basically gives their chips away to the big OEMs
Intel need to be really careful about how much of a discount they offer OEMs on "Bulk purchases", The EU Commissions have been known to raid Intel offices, just to 'check up' on them. The EU really don't take kindly to monopolies. So Intel have to be careful where they tread, and if they do get caught out, I have a feeling it'll cost them more than $1B over a 6 year period.

>>58240120
Intel isn't licensing AMD's GCN arch. AMD and Nvidia have patents that say basically "this is what a GPU does, and this is how it does it." No one can make a GPU and compete with them unless they licenses these fundamental patents, save for it they found a radically new way to render a triangle.
The licensing agreement is essentially a fee paid to avoid a lawsuit. Intel previously had an arrangement with Nvidia. That deal expired, now they're paying AMD a few billion over a few years.
I think it was either 4 or 5 billion.

>>58240207
The entire stock market is artificially inflated by QE. All that printed money entered the market, and we're cruising towards a recession.
Nothing but huge increases in GDP, paying down debt, and increasing the dollar's value will prevent it.

>>58240233
Coffee Lake is basically intel's way of staying on top of the market for the single chip solution. With 6 cores they'll have the CPU perf advantage by a decent margin. Even if they're behind in pure IGP performance they should still be neck and neck in game benches.

Intel will have a 6core mainstream chip, and 10core HEDT chip. 2 more cores then AMD in each segment.
This means, 100% officially, intel is now the moar coars company. The meme is no longer relevant to AMD.

>>58240294
AMD still has 6-core Zen chip.

>>58240316
The Raven Ridge APUs are only 4 cores.

The only 6 core chip would be pure CPU. A Ryzen SKU with 2 cores in a CCX binned off. Its still unknown if they'll even bin cores like this.

>>58240254
If AMD can provide 90% the serial performance and notably higher graphic performance within the same TDP, Apple will very likely use them.
As this is Apple, they might actually order a APU with at least a single HBM stack, since they aren't afraid of custom designs, which would be sick.

>>58234946
This is news? FX9590 can go to 5GHz with a NH-D14/5. Doesn't speak of actual performance though.

>>58240323
Raven Ridge is 2017, what about the 2018/2019 APU?

>>58240332
Go away you retard.

>>58240333
We don't know shit about Zen+. But just Zen already spooked Intel enough.

>>58240345
Prove what I said is wrong then...wait you can't.

>>58240359
Go bait somewhere else please.

>>58240359
>Early 8c/16t engineering sample at 3.1 directly compete with a 6900k (8c/16t incase you forgot)
>Same chip being at 5Ghz doesn't mean anything

>9590
>Factory OC'd bulldozer chip
>5Ghz
It's a testament to the OC capability of the chip if nothing else. The arch was always flawed, it was never going to see much performance in single threaded tasks, but you put it in a threaded workload and those clocks make a notable difference on those chips.
There, proved wrong, and proved that you're a retarded weeb poster at the same time. 2 bird, half a stone.

>>58240391
>directly compete with a 6900k
You made conclusion based off incomplete information. You deliberately left out gaming benchmarks because it does not support your claim. Not to say, the benchmarks consisting of items like encoding are not exhaustive ways you can use a CPU.
>Factory OC'd bulldozer chip
[Citation needed]. Furthermore, I only commented that 5GHz on air isn't news. Irrelevant comment.
>testament to the OC capability of the chip
5GHz on ryZen is also a "testament to the OC capability of the chip". You try to insinuate otherwise.
You didn't prove anything but your foolhardiness.
>>58240367
>Desperately resorting to Tu Quoque
No argument again.

>>58240545

>You deliberately left out gaming benchmarks because it does not support your claim.

Zen was right where it should be in relation to a 6900k in the gaming benchmarks given the clock difference.

Losing to a 6700k in vidya does not mean a bad chip as otherwise the entire x99 platform is bad.

>>58240583
I never said the platform was bad,nor ryZen is bad at gaming. I only argued that his opinion about ryZen "directly competing" with Broadwell-E holds little water. Competing on a few aspects, yes--for those who wants to save money on other components, but not on the whole, which is grossly inaccurate.

>>58240623

>"directly competing" with Broadwell-E holds little water.

Extrapolate.

>>58240647
>Extrapolate
That is exactly what that anon did. Nice pot calling kettle black.

>>58238161
Well, considering that, up till this point, AMD has been doing going up and down and all around touting that their Zen IPC can match Intel's 6xxx series IPC (give or take a smidge depending on workload) while showing better margins on power consumption means the frequency it operates on DOES matter.
Equal IPC means the higher clocked part outperforms the lower clocked on.

And AMD hasn't been talking about their frequencies much up till recently, and only then they've only revealed the BASE clock of ONE retail part. ("starting at" 3.4GHz)

This looks like a winner, and if the 8c/16t part is priced in the 500-700 USD bracket, I think it's an excellent buy.

>>58240741

>BASE clock of ONE retail part

Given its the 8c/16t chip which should have lower base clocks to keep TDP compared to 6/12 and 4/8 chips I would not be surprised if the 4c/8t chip breaks 4ghz base.

>>58240741
Su said that at launch their Ryzen CPUs would have base clock speeds of 3.4ghz or higher.
The statement was about the hole product line, not one part.

>>58240824
I'll hang myself by the balls if Zen 8 core is higher than 3.2GHz, it's just not possible with how shitty GloFo's process is.

>>58241291
Stop regurgitating things you know nothing about, kid.
It isn't GloFo's process. They are running Samsung's 14nm LPP.
You know literally nothing about electrostatic characteristics of their process, let alone any other.

>>58241291
The Stilt please.

File: 1297569197359.png (15KB, 762x333px) Image search: [Google]

15KB, 762x333px

>>58238048
>>58238070

>>58241291

>/g/ in charge of accepting reality

>>58236323
nope only win10/any linux

>>58241291
They already demoed an 8 core running at 3.4 GHz, so go ahead and hang yourself by the balls.

>>58234946
The fx8350 can Reach 5ghz just fine

>>58241655
Yeah, those are the same CPU so that's a valid comparison.

Retard.

>>58241668

One previous thread had anon claiming that zen matching broadwell-e IPC didn't matter if it couldn't clock high because clocks are all that matters.

The internet has been in meltdown over zen with some truly impressive shilling to deny what AMD has shown off so far.

" does anyone even take amd seriously anymore " was a year ago

this is just complete irrelevance

>>58239987
>this level of denial
lmao

>>58239156
AMD GPUs always have been more powerful in terms of compute power/$$

File: 1b70c8cd-73d1-4cab-bc44-903e46d0351a.png (105KB, 255x345px) Image search: [Google]

105KB, 255x345px

He actually did it!

>>58241861
Meme engineer

>>58240233
Ddr4 isn't what's gonna be the future for apu's, HBM is. I honestly think amd designed HBM from the get go entirely with future apu's in mind.
Having the only apu on the market that can deliver performance on par with an actual gpu, even a lower end one, would be huge for amd.
It would be the obvious choice for all in one and ultra compact designs, and could single handedly return amd's relevance in the ultrabook market.
The 8 and 4 core zen parts are what's most interesting to us, but the mega core count vartiants and apu's are what's going to secure long term contract and thus steady cashflow for amd.

File: df5b758f-29bb-45ed-9497-0a28d7f9f72f.jpg (29KB, 700x400px) Image search: [Google]

29KB, 700x400px

>>58241861

You actually dared to doubt him?
Of course he did it.
But at what cost?
Look how much of his own life force he poured into the product to make it good.
He looks like an old worn out alcoholic version of Kevin Sorbo nowadays.

File: 1480858901324.jpg (74KB, 627x627px) Image search: [Google]

74KB, 627x627px

>>58241861
>>58241906

>>58241906
Looking all day at verilog does that to people.

>>58241964

I would argue its that Su is a maneater and Jim "always bet on duke" Keller is brains over brawn.

>>58241964

Not to mention the amount of stress involved in that work must be insane.
>We have no idea how to pull this off and it has to be ready in 3 months. Get to work.

>>58241906
Pretty sure he was just younger and fatter in the picture that keeps getting posted. That said:

JIM "TAKE INTEL OUT BACK LIKE OLD YELLER" KELLER

>>58241903
HBM is a long game play. It's more than possible to fit it to a CPU via an interposer, the question is the impact upon cooling. To compensate you would ofcourse lower clock speeds, but to what degree.

An HBM enabled APU would absolutely win AMD a huge portion of the OEM market, Low to midend desktops, ultrabooks, and a nice slice of laptop action, mayhaps even open up a new "Low cost gaming laptop" alternative, which would print money if the mobile APUs performed.

>>58242457
An HBM die stack pulls 3-4w.
With HBM2 signaling at 1000mhz DDR its providing 256GB/s. Absurd bandwidth for the power its drawing.
Its not contributing a significant amount of heat under the heat spreader. The only temp that matters with HBM is the internal temp of the hottest layer in the stack. This is only relevant internally, not externally.

No APU will need this much bandwidth for a long time anyway. By the time a chip comes around with bandwidth demands this high HBM will have been replaced with some other memory standard.

>>58242518
My thinking was more so that an interposer would increase the distance between the cores and the IHS, resulting in less efficient energy transfer.

>>58242543
If the was a difference in height of the dies on the package then the heat spreader would be shaped to accommodate it.
Increasing air gap between logic and IHS would never happen.

This sounds fake as fuck.

remember poolaris and bulldozer? both shit and both hyped to hell? this is the asme, just mroe crap from AMD which makes crap, a crap company

It's so good they're going to delay it another year or two to let Intel catch up.

Meanwhile they keep releasing these super-low power APUs that sell for $15 and have negative profit margins. They do one of these every fucking year!

>>58245488
Polaris is actually quite nice and competitive. Performed as expected.

>>58246581
Wrong

>>58241842
>compute power/$$

That yes, but they have not been the pure raw performance kings for a few years

>>58240294
>we're cruising towards a recession.
>towards a recession
>towards a
>towards
where the fuck have you been? it's been a depression since 2008. and here you are promising it will get worse?
sheeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeit.

>>58247699
Well the world economy could outright crash, so I suppose it could get worse. As in can't find toilet paper in stores and hyperinflation worse.