Quantcast
[ 3 / biz / cgl / ck / diy / fa / g / ic / jp / lit / sci / tg / vr / vt ] [ index / top / reports / report a bug ] [ 4plebs / archived.moe / rbt ]

Due to resource constraints, /g/ and /tg/ will no longer be archived or available. Other archivers continue to archive these boards.Become a Patron!

/g/ - Technology


View post   

[ Toggle deleted replies ]
File: 129 KB, 1562x332, lowpower vs high performance node.png [View same] [iqdb] [saucenao] [google] [report]
61759679 No.61759679 [Reply] [Original] [archived.moe] [rbt]

For the first time in history, AMD late next year will have a superior process node for high clocks against Intel.

In some 3 decades AMD will have a fab advantage against Intel.

Scary thought.

>> No.61759708

AMD should have partnered with IBMs fabs much sooner.

>> No.61759711

>>61759679
Yes, it will be hilarious.

>> No.61759721

>>61759708
That's not what happened.

>> No.61759725

JUST WAIT™

>> No.61759753
File: 187 KB, 568x612, the darkness within.png [View same] [iqdb] [saucenao] [google] [report]
61759753

>>61759725
Y-YEE AMDRONE YOU JUST WAIT FOR 10NM NODE.

>> No.61759762

>>61759725
>JUST WAIT™

t. Intel 10nm+++++

>> No.61759789
File: 2 KB, 91x24, samefag.jpg [View same] [iqdb] [saucenao] [google] [report]
61759789

the amdrone samefag is strong ITT

>> No.61759797
File: 9 KB, 250x250, 1500647453955.jpg [View same] [iqdb] [saucenao] [google] [report]
61759797

>>61759679
SHUT IT DOWN

>> No.61759803
File: 295 KB, 410x491, 1459051908151.gif [View same] [iqdb] [saucenao] [google] [report]
61759803

>tfw jewtel will bribe their way to still have >50% marketshare
>amd will be outpaced again
>PC industry will continue to die

>> No.61759814
File: 62 KB, 1072x652, 9fa7538291a8ac4edf0bc9fb67fbca57b83d42c731ec32ff590b50cc39371439.png [View same] [iqdb] [saucenao] [google] [report]
61759814

>>61759789

>> No.61759816

>>61759679
>just wait

>> No.61759840

>next year? What do you need 2 years later? Nobody gets to 5 years anyway. HELP POLICE THIS MAN IS LOOKING 10 YEARS INTO THE FUTURE

>> No.61759843

>>61759803
>Intel
>bribing anyone after the royal fuckup that is Early Purley
Ha ha no. Expect Raven Ridge everywhere, even in cheapest 3rd world-tier prebuilts.

>> No.61759882
File: 36 KB, 221x246, 1451358756801.jpg [View same] [iqdb] [saucenao] [google] [report]
61759882

still nothing from AMD about a Ryzen 1 refresh?

>> No.61759885

>>61759721
glofo got fabrics for free, no way IBM didn't know they are working with AMD

>> No.61759911
File: 108 KB, 1442x719, AMD-Raven-Ridge-APU-Specs-and-AMD-Pinnacle-Ridge-CPU-Specs.png [View same] [iqdb] [saucenao] [google] [report]
61759911

>>61759882
Just a roadmap mention.

>> No.61759925

>>61759911
There's been a lot of talk that Raven Ridge is using Glofo's custom 14nm HP process, we might be seeing quite high clocks from Zen cores in that case.

>> No.61759939

>>61759679
>first time in history
underage confirmed

>> No.61759943

>>61759939
Do tell us when it happened?

>> No.61759958

>>61759882
Zen2 comes too soon, they'll probably skip that.

>> No.61759964

>>61759958
Why would they?
Higher clocking Zen refresh will murder all and every mainstream CFL-S SKUs.

>> No.61759971

>>61759943
Athlon 64

>> No.61759975

>>61759925
It's "14nm+" for RR, which means 14nm HP (which is 14LPP but with 7nm LP-styled transistors, aka much higher fmax).
>>61759971
A64 used inferior node.

>> No.61759976

>>61759971
Are you retarded or can't fucking read?

Possibly both.

>> No.61759987

>>61759975
>(which is 14LPP but with 7nm LP-styled transistors,
No scaling benefits then? Though I doubt AMD needs it that much at this point considering how cheaply they treat silicon

>> No.61760004

>>61759943
different anon, but the athlon thunderbird blew intel out of the water back in 1999.

>> No.61760007

>>61759987
It's intended to compete against 12nm FFN, which also provides no density improvements.
It's all about perf right now for AMD, their biggest die is only 486mm^2, being Vega10, which is tiny compared to 815mm^2 silicon behemoth of V100.

>> No.61760019

>>61760004
I suggest you read the OP carefully again

>> No.61760023

/g/ pls stop being retarded.
>>61760004
We're talking process nodes here, anon.

>> No.61760075

>>61759958
There's no reason whatsoever for AMD to skip out an entire year of new products, be it refresh, slight tweaks at lower power.

You're suggesting AMD not release anything for the desktop until late 2018 or early 2019? You must not run a business.

>> No.61760078

>>61760019
>>61760023
Yep, I didn't actually read the OP...

>> No.61760102

>>61759679
I think this hurts Intel far more than getting their shit slapped by glued together Zeppelin died does.

60%+ of their R&D money are fabs, they depend on fabs to push their CPU technology forward.

>> No.61760109
File: 36 KB, 571x556, Screenshot from 06-07-2017 19_12_35.png [View same] [iqdb] [saucenao] [google] [report]
61760109

>>61760102
Their 10nm node is a fucking abortion.
Just look at that fucking mask count.

>> No.61760121

>>61760109
This needs editing

>> No.61760149

>>61759679
>Scary thought.
Anuddah shoah!

>> No.61760175

>>61760149
Oy to the vey, brother.

>> No.61760204
File: 48 KB, 571x698, 1502017668508.png [View same] [iqdb] [saucenao] [google] [report]
61760204

>>61760121
Fixed, this puts things into perspective.

>> No.61760221

>>61760204
Do GPUs not count in high frequency IC designs?

>> No.61760233

>>61760221
They do, but that's a different discussion, GPUs won't be going over 2GHz for a while.

>> No.61760316

zen+ -> 14nm -> 2018?
zen2 -> 7nm -> 2019?

>> No.61760326

>>61760316
Zen+ is Q1 2018, Zen2 is probably Q4 2018 for mobile parts, Q1 2019 for the rest.

>> No.61760336

>>61760326
>Zen+ is Q1 2018, Zen2 is probably Q4 2018 for mobile parts, Q1 2019 for the rest.
Someone finally understand how AMD is releasing this.

>> No.61760369

>>61760336
You always launch your mobile products before winter holidays so normies will buy laptops.

>> No.61760422

>>61760233
Some Pascal chips can hit 2100mhz.

>> No.61760432

>>61760422
Good bins are good bins. Anyway pipelining the GPU comes at a very steep cost.
Don't expect higher clocking GPUs for a while.
And if someone actually pulls MCM GPUs, these would certainly clock lower than both Vega and Paxwell.

>> No.61761051

>>61760369
well, the Ryzen APUs should come next, after Threadripper and Epyc. Perhaps they will be delayed by the recent bugs found, but it will still be before the holidays.

>> No.61761081

>>61759975
Stop pulling things out of your ass, you're spreading complete disinfo.

The only 14nm HP process Global Foundries had is IBM's SOI process. Raven Ridge is not utilizing it, and it is not a transitional process with a 7nm FEOL.

>> No.61761103

>>61761081
https://forums.anandtech.com/threads/globalfoundries-7lp-7nm-leading-performance-finfet-process-and-fx-7-asic-platform.2508687/page-4#post-38975888
https://forums.anandtech.com/threads/globalfoundries-7lp-7nm-leading-performance-finfet-process-and-fx-7-asic-platform.2508687/page-3#post-38973556

>> No.61761112

>>61759679
Oy gevalt yes goy, you better buy intel pre-emptivey now so that the AMD monopoly distopia doesn't happen, hehehehe..

>> No.61761125

>>61761103
You're literally reposting the the ramblings of a schizo, you tech illiterate retard. Jesus Christ. Seronx is known all around the web for his bullshit.

Holy fuck you need your head kicked in for being this dumb. This is the same schizo who stated there would be an Excavator++ chip on a 28nm FD-SOI process.

>> No.61761131

>>61761125
>-- Working on Different GF technologies like 65LPe, 55LPe, 40LP, 28 (HPP, LPQ, SLP, LPS), 22FDX, 14(LPP, HPP, HP)

>> No.61761138

>>61759789
>it's okay when SOPA DE MACACO is doing it

kys yourself reddit

>> No.61761149

>>61761131
>look at me continually regurgitating irrelevant information I don't even understand
Do us all a favor and stop posting here, kid. We don't need any more WCCFtech tier shitposters.

>> No.61761178

>>61761149
Literal project manager of GF says he works on several 14nm processes and you respond with shitposting, how cute.


https://sg.linkedin.com/in/gyana-sahoo-33294015


AMD's own roadmap says that Zen and Vega will have "14nm+" variants, so there's some process distinction from the current "14nm" designs in the same roadmap, you denying this is as cute as it is sad.

>> No.61761214

>>61761178
>tech illiterate underage kid keeps posting even more inane and irrelevant nonesense

Global Foundries having a different refined VT of a process does not have anything at all to do with Seronx's schizophrenic delusions. Stay in /v/ with the rest of the cretins. Literally everything you posted was ass pulled speculation from a confirmed mentally ill brony who has been spamming forms with his crazy predictions for a fucking decade.

>> No.61761226

>>61761214
It's okay anon, just take the meds.

>> No.61761266

>>61761226
You're a bigger retard than Juanrga.

>> No.61761274

>>61761266
It's fine, it's fine, nobody is judging you.

>> No.61761290

>>61761266
>Juanrga
The faggot is still alive?

>> No.61761301

>>61761274
Keep on shitposting, little kid.
Its definitely going to make you look like less of an abject retard for mindlessly regurgitating the techno babble of a schizophrenic.

>>61761290
Banned on S|A, yet again, but still ruining better places like RWT with his unwanted presence.

>> No.61761326
File: 1.60 MB, 2560x1440, AMD-Zen_3.png [View same] [iqdb] [saucenao] [google] [report]
61761326

>>61761214
What's 14nm+, anon?

>> No.61761349

>>61761326
The exact same thing that Intel's 14+ was, you moron.
Its a refined VT of the exact same process. It it not a whole new FEOL.

>> No.61761351

>>61761326
Nothing, it's just more smokes and mirrors from AMD since they're wholly uncompetitive, trying to make people think there are actual improvements in a maxed out design like Zen when there are none.

>> No.61761359

>>61761326
No one knows...
...yet.
>>61761301
Why can't Kanter just fucking ban him ffs.
>>61761351
t. JUSTnich

>> No.61761362

>>61761351
SOPA

>> No.61761388

>>61761349
Yeah, nah, Glofo engineer linkedin proves that there are several different FinFET processes on 14nm, you harping that AMD won't use any of them and stick to Samsung licensed LPP is just a hopeful fantasy for whatever reason.

>> No.61761459

>>61761359
>Why can't Kanter just fucking ban him ffs.
Same reason why Weitken doesn't get banned from shitting up comments on EETimes articles. Nobody cares enough to.

>>61761388
Yes, I know your IQ is only 70. You don't need to keep driving the point home, little kid.
Global Foundries absorbed IBM's foundry business, and they're IBM's dedicated fab now. They have IBM's 14nm SOI process. That is not at all related to the 14nm LPP they're using for Zen, and designs cannot be directly ported between them.

Raven Ridge is not an SOI part. Its bulk 14nm LPP exactly like the rest of the product stack for Zen. Processes have different VTs, and VTs are refined over time. There is an immense world of difference between completely run of the mil improvements, and pulling an entirely new process out of thin air. You're desperately trying to validate a delusional schizophrenic brony. Thats what level you're on. Because you're too much of a tech illiterate retard to recognize his bullshit.

>> No.61761506 [DELETED] 
File: 154 KB, 1920x1002, Screenshot from 06-08-2017 15:19:30.png [View same] [iqdb] [saucenao] [google] [report]
61761506

>>61761214
LITERALLY
DENYING
REALITY

>> No.61761513

>>61761506
You've got to be Seronx himself to be this desperate and stupid.

>> No.61761518
File: 154 KB, 1920x1002, Screenshot from 06-08-2017 15:19:30.png [View same] [iqdb] [saucenao] [google] [report]
61761518

>>61761459
LITERALLY
DENYING
REALITY

>> No.61761530

>>61761506
>>61761518
Seronx confirmed.
Butthurt that you're infamous around the whole internet for being so pathetic?

>> No.61761549

>>61761530
>he still is trying to ignore what's said on project manager's of GLOBALFOUNDRIES linkedin

Seronx didn't put it there, but for sure he put the foot up your ass from how you're acting.

>> No.61761582

>>61761549
>Seronx so desperate to defend himself that he can't even comprehend a single post.
It is pitiable that you're so ill. If we lived in a better world people in your condition would simply be euthanized.
You don't even know what you're saying anymore.

>> No.61761624

>>61761582
Kek, I don't know what grounds you have to stand on acting like that, basically ignoring what's said by a GF manager, and trying to spin the direction to some forum faggot.


What have you got from this? Besides looking like a tool of course, I doubt that's your goal.
What's your next plan? The old rulers of Europe were niggers?

>> No.61761670

>>61761624
>Seronx persists in his attempts to justify his schizo babble delusions

You're genuinely a sad person, being so mentally ill, doing nothing but shitposting on forums with your drivel, year after year.
Nothing on that Linkdin profile reflects your completely baseless guesswork. There is no 14nm process listed that is new. Nothing that indicates an entirely new transitional process with a 7nm FEOL. You're just a mentally disturbed cretin who doesn't know when to stop posting.

So detached from reality you think you're proving some point when really you're just babbling more and more.

>> No.61761700

>>61761670
>Nothing on that Linkdin profile reflects your completely baseless guesswork.

IT LITERALLY SAYS 14nm LPP, HPP AND HP HOLY HAHAHAHA LOL

The first one is known where it's used, the other two are unknown, stop trying to deny reality.
Fucking hell.

>> No.61761715

I love foundry threads, they always end up in a massive technical argument that you rarely see on /g/ nowadays

>> No.61761743

>>61761700
>Seronx is so deep in his unmedicated schizophrenia that he can't even remember his own posts

14 HP is IBM's SOI process. Welcome to 2 years ago.
14nm HPP is what was previously referred to as 14nm SOC. It is just a variant of the 14nm LPP.
As I said, there is nothing new here. Nothing at all to justify your defective brain ramblings.

>> No.61761758

>>61761743
>14 HP is IBM's SOI process.
No, that's 12FDX.
You're both fucking retards.

>> No.61761776

>>61761758
No u

>> No.61761780

>>61761715
These threads are filled with little kids who parrot words they see on Anandtech without knowing a thing about them.

>>61761758
>being this much of a clueless fuck up
The FDX line is licensed STMicro IP. It is the successor of their 22nm FDX line. It is primarily aimed at RF applications, and has literally nothing to do with IBM.

Don't jump into a conversation you know nothing about.

>> No.61761784

>>61759814
>>61761138
Why is this a response to intel shills now?

>> No.61761801

>>61761784
Favela cannibal Intel shill that has been very active since Ryzen launched accidentally revealed himself to post on hue FUD blogs

>> No.61762453
File: 123 KB, 547x261, 10nm by 2015.png [View same] [iqdb] [saucenao] [google] [report]
61762453

>>61759725
ahahahahahaha

>> No.61762672

>>61759939
Athlon 64 was way netter than Pentium 4but intel still had the frequency war. OP specificly noted frequency not superiority of the product.

>> No.61763082
File: 232 KB, 800x449, intel ceo yells at clouds.png [View same] [iqdb] [saucenao] [google] [report]
61763082

>>61761351
Get back to work, Brian.

>> No.61764771

AMD is shit tho

>> No.61764913

>>61764771
>>61763082

>> No.61765298

>>61759679
More AMDelusions

Keep shilling your stock

>> No.61765314

>>61765298
Brian, stop shitposting.

>> No.61765397

>>61765314
Schizophrenic AMDtards calling everyone Brian

I swear AMD fanboyism is a clinical disease

>> No.61765430

>>61765298
>>61765397
>>61763082

>> No.61765443

>>61765397
UMA DELICIA

>> No.61765455
File: 72 KB, 1659x278, oy2.png [View same] [iqdb] [saucenao] [google] [report]
61765455

>>61765397
SOPA?

>> No.61766156

can we expect 4.5ghz from zen+?

>> No.61766231

>>61766156
I don't know

>> No.61766254

>>61766156
Assuming we're getting a Zen+, no. Maybe slightly higher clockspeeds. Zen 2 is when you'll see much higher clockspeeds.

>> No.61766884

>>61759679
>14 nm: 8.5M gates/mm2
>7 nm: 17M gates/mm2
One wonders why they don't just call it 10 nm.

>> No.61766932

>>61766884
Because it's half of 14nm

>> No.61766938

>>61759679
>For the first time in history, AMD late next year will have a superior process node for high clocks against Intel.
I also want to believe, but on what basis are you assuming it's superior?

>> No.61766982

>>61766938
>>61760204

>> No.61766989

>>61759725
UMA

>> No.61767005

>>61766932
What is area scaling?

>> No.61767100

>>61760432
>Anyway pipelining the GPU comes at a very steep cost.
Credentials or source, please.

>> No.61767141

>>61767100
Vega itself.

Most of the nrpew xtors went into enabling 1700+, clocks on GCN

>> No.61767162

>>61760204
Interesting. Fabbing quality really was the one thing I didn't think Intel would fumble away. What has happened?

>> No.61767196

>>61767141
>Most of the nrpew xtors went into enabling 1700+, clocks on GCN
I would've thought the node shrink was the main facilitator of the clock improvements.

>> No.61767210

>>61767162
Even Intel can't go against the Alliance.

>> No.61767211

>>61767196
No, Vega clocks much higher than Polaris on average.

>> No.61767231

>>61767196
Not a 70% clock increase for a wide high IPC design like GCN, lol
If AMD could have reached those clocks without spending a lot of xtor budget Vega would be 5000 ALUs

>> No.61767268

>>61766156
For turbo frequencies at least its well within the range of plausible.

>> No.61767377

>>61767231
>If AMD could have reached those clocks without spending a lot of xtor budget Vega would be 5000 ALUs
Unless they spent the gates on the architectural improvements. After all, the difference from Maxwell to Pascal is from 2.6 Mgates/ALU to 2.8 Mgates/ALU, which is far less than the difference from Polaris to Vega, from 2.5 to 3.1.

>> No.61767703
File: 12 KB, 160x160, 🤔.png [View same] [iqdb] [saucenao] [google] [report]
61767703

>>61759679
>In some 3 decades AMD will have a fab advantage against Intel.
>fab advantage
>AMD
>fab

>> No.61767736

>>61767703
They don't have to own the fabs to have a fab advantage, you know.

>> No.61767859

>>61759679
If someone told me 2 years ago that AMD would have a superior node for their CPU than Intel I'd spit in his face.
Now we're actually there.

Some people don't wanna admit it but the x86 market is changing, and fast.

>> No.61767915

>>61759911
>pinnacle ridge
>works with the same chipset as zen

DELID THIS REEEEEEEEEEEe

>> No.61767994

>>61767915
We knew am4 will last for a long time a year ago

>> No.61768045

>>61767915
buy new socket dearest customer

>> No.61768244

>>61759679
5.0ghz won't happen.
I want it to
But it wont
4.4-4.5 is a realistic clock speed for zen2

>> No.61768280

>>61768244
It gonna be over 5Ghz.

Not stock of course.

>> No.61768328

>>61768280
If it's not stock, what's the point?
All Intel CPUs hit 4.4 with boost 2.0
Ryzen needs a clock advantage as their IPC will soon be on par with Intel.

>> No.61768388

>>61760432
My stock strix 1070 reaches 2000Mhz...

>> No.61768410

>>61768328
AMD isn't going into this ignoring power usage, which has a cascade effect on platform, reliability and cooling, as Intel and Bulldozer beautifully demonstrated.
No company should risk their reputation for a handful of faggot gamers chasing clocks.

>> No.61769247

>>61768388
My msi 1080EK maxes out at 2,156mhz
Use the MSI "force voltage" and move the voltage to 100%.
Will up the average clocks a lot and increase max frequency.
Pascal is too stingy with voltage.

>>61768410
Power usage is an issue
But if Zen 2 clocks at 4.5+ I can't see it be any more efficient than Zen at 4.0ghz

>> No.61769296

>>61769247
That depends on process, LPP at 4ghz is far from its sweet spot, 7nm LP's optimal clock is far higher

>> No.61770740

>>61759725
JUST WAIT for 10nm with more than 1.7% yields

>> No.61770798

>>61761326
Probably Raven Ridge.

>> No.61770989

This already has been talked about before

http://wccftech.com/intel-losing-process-lead-analysis-7nm-2022/

http://semiaccurate.com/2016/09/26/globalfoundries-7nm-process-isnt-even-close-name/

>> No.61771094

>>61770989
Is it really 7nm? No. Does it need to be to beat Intel's 10nm? Also, no. >>61760204

>>
Name (leave empty)
Comment (leave empty)
Name
E-mail
Subject
Comment
Password [?]Password used for file deletion.
Captcha
Action