6C/12T Ryzen confirmed having 3.3ghz stock - Bad overclocking potential confirmed just like rx 480

You can't make this shit up. [A]nother[M]ajor[D]isappointment on suicide watch.

twitter.com/CPCHardware/status/826829044402552833

Other urls found in this thread:

techpowerup.com/229090/amds-upcoming-ryzen-chips-to-reportedly-overclock-5-ghz-on-air
forbes.com/sites/antonyleather/2017/01/29/no-6-core-amd-ryzen-cpus-4-core-and-8-core-only/#40cec6077f9c
guru3d.com/news_story/amd_might_not_release_a_hexacore_ryzen.html
ark.intel.com/products/75242
twitter.com/NSFWRedditVideo

DELETE

hard to believe considering i managed to undervolt my rx480 and OC (over 200mhz on the mem, over 100mhz on the core)
>still thinking ghz speed matters outside of the same line of cpus
go home shill.

>canard
Fucking leafs

>undervolt my rx480 and OC (over 200mhz on the mem, over 100mhz on the core)

>amd fags literally downgrade their gpus
>ignore that's info from one of most reputable pc magazine who got zen early
>is so delusional that think 3.3ghz zen will defeat 5ghz intel lol
>it really makes me think

AYYYMD fanboys literally on suicide watch

You guys are worse than CTR
go back to Sup Forums with your company wars

>being this MAD

undervolted and overclocked retaaaaard

DELET

>dat jenhsun face in the background
kek

>Rumor = Confirmed
We Sup Forums now.

But what about the price?
t. poorfag

it's not rumor you retard it's confirmed fact

I dont speak leaf, what does its stock speed effect its overclocking potential?

>3.3Ghz base clock
>bad
Try harder, Intelshills

$60 Pentium is better clocked

>i5 7600k - 3.80 GHz STOCK
>i7 7700k - 4.20 GHz STOCK
>i7-6700K - 4.00 GHz STOCK
>I5-6600k - 3.50 GHz STOCK

>3.3ghz base clock for 8c/16ht
>NOT BEING FUCKING BAD

you amd fags are so delusional

they are validating 6 cores.
what's the problem?

So first they confirm overclocking then they go back on that claim?

techpowerup.com/229090/amds-upcoming-ryzen-chips-to-reportedly-overclock-5-ghz-on-air

jesus christ how desperate to have to be for (You)s to samefag this hard?

No. They're basically saying that there are chips which are stock clocked at 3.3 GHz with 6C/12T, and they're assuming that there's no 8C/8T CPUs, since the binned CPUs will be more probably used for 6C chips that for 8C chips without SMT.

they're not assuming, they're claiming

>comparing different dies clock for clock

is it 2001?

There were rumors there wasn't going to be a hexacore in the first place.

Step up your trolling.

They're claiming that no 8C/8T are being validated, you're right.

All rumors....

forbes.com/sites/antonyleather/2017/01/29/no-6-core-amd-ryzen-cpus-4-core-and-8-core-only/#40cec6077f9c

AMD claim they have same TDP as Intel now so it's obvious that you can compare 3.3ghz of zen to intels i5/i7 3.5-4.2ghz

No because that was back when AMD was competitive.

i meant IPC

That's what I've been thinking too. Seems like people care a lot about clock rates again.

i'll be waiting for benchmarks.

Having the same TDP does not mean they're the same architecture.

why 12 thread thought

Literally what is the problem if this is confirmed? It's still a 6C/12T at 3.3 fucking GHz for CHEAP

Since he benchmarked an engineering sample of a 8C/16T, it's not much of a stretch to assume he's seeing some 6C/12T going around, and no 8C/8T at all.

That said yeah, a twitter quote isn't much of a source, as much as I like Canard PC Hardware.

>AMDREDTEAM

Parts with more cores tend to be lower clocked for power reasons. That's why those cheap E5 2670 are just 2.7 GHz.

>comparing quadcores to 6 and 8 cores
Is Intel this desperate?

The 5960X is 3.0GHz. Am I missing something here or is it blatant shilling?

it's a twelve year old from Sup Forums trying to troll

>POOZen barely clocked higher than a years old midrange chip from based Intel
AYYYYMDlmao

>is so delusional that think 3.3ghz zen will defeat 5ghz intel lol

No-one thinks that. The 5GHz Intel is going to cost 4-5 times as much and nobody will bother except professionals that need that kind of power.

>bad overclocking potential

aka optimised hardware

can't pic related be implemented?

3.4 ghz was confirmed to be the minimum, this news is fake

>Bad overclocking potential confirmed just like rx 480
Where do you get this claim from?

in tel aviv

>5280k 6c/12t 3.3ghz

10% is barely higher? Enjoy your 1% gains faggot.

3.4+ was meant for the 8 core.
I see no reason why the 6 core should clock lower, other than a budget SKU.

>just like rx 480
My 480's base clock is at 1.1GHz.
I run it overclocked on 1.45GHz @ 1.18v

NOOOOO

Clock binning and defect binning go hand in hand. These chips would also likely be 65w SKUs rather than 95w.
Su's statement about clock speed was that at launch Ryzen CPUs would have clocks of 3.4ghz or higher. The launch window is the qualifier there.
Lower clocked chips being available later doesn't conflict with that.

Yeah AMD downvolts really well for some reason. I remember some cheapo laptop I got in 2009 that had an absurd stock voltage, 1.425v, wasn't even clocked that high. Took it down to 1.050v and it ran fine.

>Intel released a desktop CPU for $1800 with only 3GHz
AHAHAHAHAHAAHAHAHAAHAHHAH

>5820k 3.3 Ghz
>6800k 3.4 Ghz
>6850k 3.6 Ghz

Am I missing something? Also

>implying clocks matter when comparing different architectures

>a chip with fewer cores has a higher base clock

OMG the g4560 has a higher base clock than the 6950x 1!1!"1!!1!1!1"

This is why I'm interested in the new line of server chips. I've had nothing but trouble with voltage scaling on Intel. The lower you can go the more you can save (which matters when it's running 24/7).

they're red since they acquired ATi

That never happened.

exactly, they killed ATI

ATI made some of the greatest GPUs until AMD bought them out

Time to bankrupt.

AMD's cpu branding is still often green. Radeon branding is mostly red.

nothing mentioned about overclocking

Because this?

guru3d.com/news_story/amd_might_not_release_a_hexacore_ryzen.html

AYYMD IS FINISHED & BANKRUPT

AYYMDPOORFAGS CONFIRMED ON SUICIDE WATCH

Anything more than 4 core will be parts that are going to be considered high end as they all will not have a gpu on chip. 4 core will be considered low tier with apu occupying the mid tier.

I fucking hate you all. Making meme wars out of anything. I will buy ZEN. You can keep your skylake refresh retards.

when do you guys think they'll stop manufacturing skylake? i'm sitting on parts right now and was waiting for ryzen to hopefully cause intel to adjust prices.

please amd for the love of satan dont fuck this up

If you want cheap chips then you'll have to settle for the low clocked SKUs, this is inevitable, you can't expect the Zen F4 stepping 4.0 boost 8 cores to cost the same as the 3.7 boost ones.
That's why overclocking exists, but don't think for a second you'll get good silicon by buying the cheaper ones, they're also more likely to overclock less unless you're real lucky.

...

Intel doesn't cut prices for shit, faglord. Go price a 4790k.

thanks for help. now tell me when they stop manufacturing skylake fagmax.

8/8 or 8/16 will beat out an i7 at 5ghz on parallel tasks.

no idea what this is in context google spits out

"AMD currently validates Ryzen 6C / 12T @ 3.3 GHz base but NOT 8C / 8T. The goal is clearly to recycle the failed 8C dies."

This makes it sound like they sent out a 6/12 for validation and it was 3.3ghz, amd has said the lowest clocked ryzen will be 3.4 so i'm also guessing engineering sample.

Also does this imply they do NOT have 8/8 cpus?

6 cores, 2 threads a core, 12 threads.

I might have to leave Sup Forums due to embarrassment after shilling this so hard. :(

and then they made some of the best gpus they ever put out

i'm taking mother sue at her word, no desktop ryzen will be base below 3.4

"at launch"

if there is a 4 core at launch it is a heavily fucked 8 core with no gpu, they won't have apus till h2

4 core doesn't sound feasible unless yields are real bad, you kinda need real shit yields to have either 2 cores per CCX not working or a full CCX not working on such small dies on a relatively mature process.
Like, 1.7% bad

Well, they did do that for Phenom, Phenom II, and the Construction cores. Pretty sure on release all of those architectures had SKUs utilizing the main CPU die but with the bad cores turned off, including tri and dual cores.

I don't sprache frog, what's he saying?

>stock up 15% today

yeah, really disappointed here

Not the same thing, Zen isn't designed like K* and Intel Core with a monolithic core complex
So we actually don't know how these CCX work, do they need to be symmetrical to even function in a 6C config (3+3 per CCX), or can they just work with 2 cores disabled in a single CCX.

Bulldozer's design made it simple, but Zen's design makes it not so simple at the moment.

True true, but at the same time AMD would be insane to not have the ability to go hex-core or otherwise turn off single cores with this, because the hex-cores and the quads if they're any decent will be the real moneymakers.

And we come full circle again, quad cores aren't gonna be in this launch window and they won't be using summit ridge dies, if they do there's something REALLY wrong with GloFo, they'll be from raven ridge, which I assume is a single 4 core CCX with a vega GPU if there's any sanity left in AMD

Six cores were all but confirmed, we just don't know if they'll be 3+3 or 4+2 cores
Disabling a single core is pointless, there's a reason nobody ever sells 3 cores or 5 cores.

AMD might create a 4 core Summit ridge by fusing off a ccx on purpose just to put something in that market spot.
Which sounds kinda silly and not very profitable to me, this would just screw over their stock of 6 core SKUs.

>Undervolt card
>frequency goes up

Try harder retard.

While you're responding to an idiot it is in fact quite possible for the frequency to go up by undervolting if the card is thermally or TDP throttled, in most cases the latter.

Selling a salvageable die is better than throwing it away.
SKUs don't have to be super high volume. Viable 4 core chips that pass clock binning may be less than 1% of working chips, but that still works out to be a shitload of potential profit. Fab 8 can put out 60k 14nm wafers per month.
If the Zeppelin die is 200mm2 a 300mm 14nm wafer has about 280 candidates per wafer. If yields are 40% then we have 112 good dies per wafer.

If only 1 of them per wafer turns out to be a working 4 core chip they can make 60,000 quad cores per month.

Though this is assuming that Zen based parts are using 100% of Fab 8 14nm capacity which isn't the case, but its serves to demonstrate the point. Even if they were getting 2,500 per month, a $100 chip is potentially a quarter million dollars in revenue.

A quad core summit ridge needs to fail spectacularly binning for both 8 and 6 core to exist, which is really kinda pushing it.
Here's what it needs to fail.

8 core:
fail voltage target, fail clock target, fail SMT(?), fail power target.
If one of these fails and there's a bunch of these failed chips, you have another 8 core SKU, if more than one of these parameters fail, you got your 6 core.
Now that six core has to fail a bunch of targets targets to get binned to a 4 core.

This is a really small chance, especially since these dies aren't really big.
1% chance? I doubt it's even that big, there's more of a chance AMD artificially creates those 4 core SKUs if they even exist.

>Validates
>Rumor

Voltage and power are effectively the same thing. Its clock binning, hitting a target clock within a certain voltage envelope, and that determines how much power it uses, and how hot it runs. Very few chips are thrown away for failing to hit nominal clocks with acceptable voltage. If a working die cant hit 4ghz at 1.2v for example, it might be held in inventory. When inventory builds up they'll revalidate the chips as either higher power draw high clocking SKUs, or lower clocked low power SKUs.
Either way they're not simply cast aside. These low end parts are typically about recouping losses more so than making a profit. Low end chips soften the blow of poor yields on high end parts.

6 core chips will be dies with either 1 or 2 defective cores.
4 core chips will be dies with either 2, 3, or 4 defective cores.

We don't know how the architecture can be disabled yet, we may not be able to disable single cores, only groups of two, if we could disable single cores, where are the rumors of 7 or 5 core skus?

really, a fucked core could be one spot on one core that takes out a second, and another on another core that takes out another with it, 2 fuck ups on a single die, not exactly statically uncommon even on high yield processes.

from a calculator and figures from samsung on yield they are at the point where 75%+ of the dies is perfect if not more.

we would have rumors of 5 or 7 core units if they were disableable as one core, but we dont, there would be more then 3 skus at launch, but there aren't.

This leads us to believe something interconnects between cores and if one is fucked with fucks 2 cores.

this means 1 defect = guaranteed 6 core unit
2 may give you a quad core unit. and i'm guessing units defective enough to be dual cores just don't happen on this mature of a process.

higher voltage means more chips are viable as many were not low voltage contenders, those were binned away for embedded skus.

you undervolt, you now allow for a higher clock at a lower tdp or potentially you were thermal throttled and the lower voltage allows you to go higher that way.

many things can be undervolted and still run perfectly fine.

Intel never felt the need to release odd numbered core chips like that. Their top server SKU has 24.

Well except that weird 15 core one, but that was the entire die actually.

ark.intel.com/products/75242

Seriously what the fuck were they thinking while designing that die.

Honestly what in the fuck.

they took the 10c die and slapped another row of 5 cores onto it, extended the memory controller for some proprietary bullshit that I dont think ever panned out, then sold it.

>It has bad overclocking potential because of low stock speeds

Yeah, these 2500K's sure never even pushed 4GHz as well.

On about 4 (3.5) years.