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: 1007 KB, 2048x1536, kikeripper.jpg [View same] [iqdb] [saucenao] [google] [report]
64840235 No.64840235 [Reply] [Original] [archived.moe] [rbt]

https://www.anandtech.com/show/12438/the-future-of-silicon-an-exclusive-interview-with-dr-gary-patton-cto-of-globalfoundries

>Q15: With the first generation of 7nm, do you expect to be high volume production by the end of the year?
>GP: By the end of the year or most likely in early 2019, with a couple of key partners. Our ASIC customers, of which there are quite a few, are also lead users of our 7nm process.

>Q17: Does the first generation of 7LP target higher frequency clocks than 14LPP?
>GP: Definitely. It is a big performance boost - we quoted around 40%. I don't know how that exactly will translate into frequency, but I would guess that it should be able to get up in the 5GHz range, I would expect.

>> No.64840935
File: 350 KB, 672x794, NAO E UMA DELICIA.png [View same] [iqdb] [saucenao] [google] [report]
64840935

>> No.64841005
File: 114 KB, 633x900, 1515816507397.png [View same] [iqdb] [saucenao] [google] [report]
64841005

>> No.64841010

>>64840235


REEEEEEEEEEEEEEEEEEEEE DELETE THIS

>> No.64841023

>>64840935
>>64841005
?

>> No.64841025
File: 89 KB, 645x729, dunce46.jpg [View same] [iqdb] [saucenao] [google] [report]
64841025

>>64840235
>we quoted around 40%. I don't know how that exactly will translate into frequency

>> No.64841035

oh AMD, don't stop winning.

>> No.64841057

>>64841025
Of course the final clocks depend on the core.
Something like reference A73 core is not going to hit 5GHz.

>> No.64841432

When will Intel be on 7nm. 2021?

>> No.64841492

>>64841025
They quotes a 40% increase in famx vs 14nm LPP. This is with internal testing done on specific test circuits. In this case if I recall they used the FPU from an ARM core.
His statement is that he can't say how it will translate into frequency of specific chips, obviously.

>> No.64842090

>Q9: As far as what we've been able to determine, the difference between 14LPP and 12LP is just a difference in going from 9T to 7.5T design, with that tuning. Are there other changes?
>GP:"Along with the track changes, we also changed the middle and back-end-of-line ground rules. One thing that I think we're noting is that the industry focuses a lot on pitches, but that's like having half the story. So much of it is also all the little subtle secondary ground rules."
-"At least one of our competitors spends a lot of time talking about the pitches and the things like that, but at the end of the day, a lot of it is how you develop all the rules for the back end and how you develop the routing.
-"We will intentionally skip 10nm. I call 10nm more of a ‘half-node’ you know, and there are some people who are very focused on the Christmas season. Therefore they have to get something out and it may not have a great deal of improvement for you."
>Q10: What is your opinion on these half-nodes?
>GP: -"The problem with 10nm, as you've added a lot of masks, you've got some scaling, but you've added a load of mass so when you combine it, the cost improvements are not that great."

HAHAHAHAHA Inlel 10nm BTFO

>> No.64842203

>>64842090
Intel's "10nm" == fab "7nm".
What Patton talks about is foundry "10nm" nodes, which are indeed like half-nodes worth of density.

>> No.64842236
File: 127 KB, 600x600, 1515193273478.gif [View same] [iqdb] [saucenao] [google] [report]
64842236

>>64840235

>> No.64842241

THAT'S GREAT BUT TRY MAKING DDR4

>> No.64842257

>>64842241
The only memory GloFo makes is eMRAM.

>> No.64842282

>>64842241
Whats that? You want DRAM to lower market prices?
Why don't you go fuck yourself instead?

t. Samsung, SK Hynix

>> No.64842314

>>64840235
30.000 7nm Vega units create an ExoFLOP machine

Please USA make another SC program we need robot waifus

>> No.64842321

>>64841025
40% performance boost doesnt necessarily mean 40% higher frequency you fucking mongoloid

>> No.64842338

>>64842321
When a fab talks about performance uplift they mean frequency. Things like gate delay are too small to matter in any practical terms. The fab has nothing to do with architecture, the only practical performance they offer is fmax. SRAM speed is too abstract to work its way into any marketing fluff.

>> No.64842345

>>64842321
It means exactly that in fab context.

>> No.64842718

>>64840235

And yet, it won't reach kaby lake performance

>>64841023

Ignore them, they are salty amdtards trying to defend their spendings

>> No.64842733

>>64842718
Hi BK!
Where's Cannonlake?

>> No.64842800

>>64841025
i bet they will give us a nice 5% boost with new socket that is compatible only with new chips
maybe even some new spectre or melltdown vulnerabilities :^)
oh wait, were talking about amd ? shoot...

>> No.64842872
File: 37 KB, 800x450, brainlettttt.jpg [View same] [iqdb] [saucenao] [google] [report]
64842872

>>64841025

>> No.64842892

>>64842241
That's up to Samsung, SK Hynix and Micron.

>> No.64842911

>>64840235
We already knew about this, it's all in GloFo's website.

>> No.64844801

DELID

>> No.64844859
File: 39 KB, 630x354, Housefire.jpg [View same] [iqdb] [saucenao] [google] [report]
64844859

>>64842718
Yes, Kaby Lake is unreachable in its use as kindling

>> No.64845248
File: 127 KB, 640x427, rms facepalm.jpg [View same] [iqdb] [saucenao] [google] [report]
64845248

I know you guys like to repost news and post crying wojaks but how many of you even own a threadripper? Most of this board literally runs on shitty hardware or some ryzen 3.
My 1070ti probably puts me in the top 5% of this NEET board.
Also for fuck sake please wait until they release their shit, trying to hype something that we don't have a final version yet is stupid else we end with bulldozer disappointment.
ryzen 2 may be really nice but don't expect 40% increased clocks the marketing man said in the interview, jesus christ, just think about it. 1950X already chokes past 4ghz because all that heat and cores. you guys think it will reach 5.5ghz on the next one while having a truckload of cores?
What next, an epyc running at 7ghz with 728 cores?
Not even shills are this retarded because they probably get some sort of guideline to what sort of retarded shit they write on /g/.

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