Just a reminder to what kind of monster process node AMD is fabbing

And I think I speak for everyone when I say IBM Nazi science is the best science.

Also reminder that IBM has shipped a 5.5GHz CPU using similar nodes custom made for their POWER architecture

>>60910476
>Just a reminder to what kind of monster process node AMD is fabbing Zen2 at.

Just tell me about after it's released op.

JUST WAIT

Please DELET

>>60910476
6 core CCX x2, 12 core at 4.5GHz and 8/10 core at 5.0GHz?

Monster.

File: 62f.jpg (36KB, 630x630px) Image search: [Google]

36KB, 630x630px

>>60910476
OY VEY ANUDDA SHOA! DELID THIS RIGHT NOW

File: delid.png (109KB, 2250x2324px) Image search: [Google]

109KB, 2250x2324px

>>60910476
>>60910786

>>60910476
Uh oh, what is Intel gonna do when Zen runs at the same clockspeed as Kabylake?

>+10-15% IPC increase
>15%+ higher clocks

Intel will be mutilated, Skylake v3 will still be selling at that point in time, if Icelake isn't magical Intel will have a hard time.

>>60910803
>Uh oh, what is Intel gonna do when Zen runs at the same clockspeed as Kabylake?

hmmmm add another Bingbus?
i actually hope AMD will increase IPC by a HYUGE margin again.

>>60910476
If only Ryzen was launched on a HP node.
Well I guess that would be nice for desktop, but their server and mobile line would suffer.

>>60910817
It probably will once they fix certain weaknesses like AVX2 instruction

>>60910476
>AMD is fabbing

AMD has fabs?

>>60910837
AMD only fabs lies for marketing.

Poojeet is here to defend this.

>>60910476
>17 metal layers

Oh shit they're not fucking around with this node.

How long do we have to Wait™ for this

>>60910882
>Design kits are available now, and the first customer products based on 7LP are expected to launch in the first half of 2018, with volume production ramping in the second half of 2018.

Reminder we had a 300+ post thread about this 2 days ago full of Intel tears

https://warosu.org/g/thread/S60886302

>>60910476
https://www.globalfoundries.com/sites/default/files/product-briefs/7lp-product-brief.pdf

>>60910862
>upto
Those 17 layers would likely be for IBM's POWER chips.
Zen uses 12, where Skylake uses 13.
Zen2 may not need to utilise any further layers.

So more layers = better?

I thought more layers is just more expensive?

>>60910476
so what your saying is that when this might be released, I can view this penis-puppetry forum .0179 nanoseconds faster?
WHERE CAN I INVEST ALLL MUH GOVT ALLOCATED SHEKELS?

>>60910943
It's better if you want to make a gigantic and power hungry chip, ala IBM's POWER
Which IBM will make on this exact node, hence the 17 metal layer availability

>>60910803
Get completely fucked because clockspeed is their only real advantage at 4 cores and Ryzen already rekts their shit in multicore.

>>60910476
>5GHz operation
>0.75V DRIVE VOLTAGE!??!?!?

Oh shit Intel is gonna get gutted, this is public execution.
This process density is on par with TSMC and Intel, but its performance is staggering, IBM has done it again

They should rename this from 7nm Leading Performance to 7nm Rape Train

>>60910977
IBM already builds monster chips, density is the least of their worries.
But it's also nice that the node is competitive with TSMC in density and slightly behind Intel's, but the real kick is the performance.

Also from what I know all these companies never use high density libraries but high perf ones, so average process density matters even less.

waiting™ for Zen2
feels good to be free from intel's slavery

>>60910812
>zen2 8 cores at stock is 40% faster than zen 8 cores
I can imagine a lot of people being mad expecting incremental 10% performance increases.

Who gives a fuck, CPUs are finally interesting again!

>>60910977
IBM powering the second shoah

fucking mint

>>60911286
IBM revenge. Intel fucked IBM in 386 era.

>Tfw waiting for Zen 2 and Vega
>Going to build a red beast

I hope they deliever next year

My body is ready

>>60911894
You'll have to wait for Navi too.

>>60910476
why are my shares in AMD dropping?

>>60913558
Because Goldman sachs are manipulating the market by forcing a sell rating.

>>60913558
buy the dip you idiot

>Zen 2 and Navi parts are 7nm
>Zen 3 and brand new GPU arch will be 7nm EUV
>AMD has its future arch and available high perf process nodes sorted out for the next 3 years already
>GloFo also has a 5nm GAA nanowire process in the works after that
>intel hasn't finished construction on their 7nm facility and will be using rehashes of 10nm until 2020-2021~

AMD, Samsung, Global Foundries, and IBM really make for one hell of a working partnership.

>>60914348
for how long will 7nm and 5nm be milked until graphmeme becomes viable?

>>60914439
>graphmeme becomes viabl
Never.

>>60914348
>tfw AMD only has to work on designing a single powerful quad core complex and engineer it to work with IF
Their future is very bright.
I don't see them letting IF go to anyone.

>>60914459
then what will replace silicon?

>>60914480
I wonder what will Navi be all about.

>>60914439
Zen 3 is slated for 2019 or 2020 at the moment, that'd give the 7nm node a 2-3 year life span. It depends on how long it takes GloFo to get the 5nm process up to volume levels.
We'll likely still be utilizing silicon substrates for years to come after that. A new substrate would be earth shattering news in the industry because of the financial investment it would require. Tens of billions if not a hundred billion dollars. Hell it cost TSMC something like 3 billion to bring their 28nm HKMG process online years ago.

>>60914491
Likely something compatible with current infrastructure. Some other material like silicene, molybdenum disulfide, or else similar.

>>60914513
Applying success of ryzen to GPU market.

>>60914597
Would be absolutely hilarious to see Huang get pajeeted for the 4th time. With another housefire to answer.

File: 1496878992876s.jpg (2KB, 125x125px) Image search: [Google]

2KB, 125x125px

>we need more cores and threads!
When will this meme die?

>>60914715
Servers do want more cores and thread. And more memory. And more I/O. AMD gives exactly that.

AMD fucking played dead for like 3 years, Intel appears to have fired it's whole R&D and engineering department as a result, then AMD just stands back up. Wow. I don't think there has been a business comeback as great as this for the past 20 years. Except Apple from a business standpoint.

>>60910476
Remind us on release date op. Kind of really tired amd's wait, and see game that's been around for over a decade.

>>60910476
>wait for threadripper nearly over
>wait for zen2 starts

>>60914765
>AMD fucking played dead for like 3 years

What did they have 4 years ago?

>>60915267
shitty bulldozer/excavator APUs

>>60911894
Same anon
It's gonna be glorious

>>60910812
Haswell v5

>>60914348

This is really the key here. The stars aligned and some how, some way, we have an entire generation of AMD superiority.

So my Ryzen will be depreciated? fuck AMD. Another reason you should never buy from them.

>>60915695
>oh no, why does technology advances????
t. caveman

Psh. Ryzen is trash, it delivers clocks less reliably than British rail delivers its passengers.

>>60915735
Source? Oh wait, you have none

>>60914715
w're reaching the limits of silicon soon. The only logical process is shrink it and add moar coars.

File: lisa_su_smug.png (123KB, 240x286px) Image search: [Google]

123KB, 240x286px

>>60915735
LOL

>>60911004
Looks like I'll wait™ too, Zen2 will be really interesting.

>>60914226
THIS..

I'm buying when I get home.

>>60916017
>>60911004

But will it fit into my X399 motherboard

>>60914515
>molybdenum disulfide
Fuck you nigger I need that to lube my bushings and gears

>>60917169
The threadripper version will.

>>60916901
Buy now or right when it starts going up?

>>60917242
Right now.
You're not going to get many, if any chances.
EPYC NDA lifts on the 20th, so full details of the lineup and performance.
Sure, if you buy now, it may drop another 40-50cents, but that's not a risk worth taking.

IBM's process will bury Intel.
Intel is nothing but a regular chipmaker without its fab advantage.

>>60917277
Just bought 300 :)

>>60917406
Intel makes good SRAM.

>>60917504
Too bad for Intel that good caches are not all it takes to make good CPUs

>>60917557
Hey, at least they are decent at something.

File: mentally-challenged-indian.jpg (25KB, 250x250px) Image search: [Google]

25KB, 250x250px

>>60910476
>reminder to what

>>60910900
So, q1 2019.
Which is great actually, they already stated the next iteration of Zen would have ipc gains of ~15%, so next year a big tock and the year later a big tick.
If Intel falls behind samsung/gf on the nodes, it will be reeeally difficult for them to stay at the top.

>>60910476
Ya intel is doing nothing at all, just going to keep using same old architecture

Look sweetie, there is a reason amd stock is shit and they can't make a profit any quarter.

>>60917277
>Thinking a cpu that will only be purchased by 1% of market will do anything to break amd out of their deficit

>>60917735
Hooo boy, denial.

File: fnially-worthwhile-amd-upgrade.png (70KB, 921x408px) Image search: [Google]

70KB, 921x408px

>>60910812
10-15% higher IPC wont' be enough to make me care right now, attached picture is my latest upgrades.

>>60910835
What do you mean by this? Are they weak on AVX2 instructions? AVX2 isn't that much faster than SSE2 for RAID anyway but you made me curious.

>>60913558
Because the sellers are more aggressive than the buyers.

>>60914715
Silly little boy, more performance is always desirable. And it doesn't matter if you get higher total performance by better-performing cores or more less-performing cores as long as the total performance goes up. This is true for everything but a handful of poorly written Windows games. For everything else it makes no difference. Just try yourself with ffmpeg or anything else that's demanding, you'll see all cores and threads fire up and work until the job's done.

>>60914765
To be fair, AMD's been dead longer than 3 years and Intel's been dead about 3 years. That 1-2% increase in performance per CPU generation from Intel has been pathetic, they basically got the upper hand and since AMD just wasn't there they kept on releasing the same shit in new wrappings.

>>60918068
>pissmark
The rest of your post is unimportant.

File: jewshown.png (138KB, 354x504px) Image search: [Google]

138KB, 354x504px

>>60910476

>Just Wait guys!

>>60918086
It works just fine as a general ballpark metric.

It's good enough to see if a new CPU is about twice as fast as the one I have or not.

File: 1450143264931.png (283KB, 623x469px) Image search: [Google]

283KB, 623x469px

>>60910476
>"Just wait 2 and a half years"
Ok bro>>60910476

>>60918097
You're not funny.

>>60918123
We're all eagerly waiting for Intel to deliver their 10nm node which is fucking basically never.

File: 1494959711971.png (316KB, 538x510px) Image search: [Google]

316KB, 538x510px

>>60917621
reminder that your a fucking intel shill


can somebody please post the screencap?

File: 1461262274864.jpg (41KB, 688x456px) Image search: [Google]

41KB, 688x456px

>>60918206
good try kike, enjoy the new JEWRIPPER processor causing anuda shoah.

>>60918244
You're not funny.

>>60918271
Bot.

>>60918229
The pro AMD shitpost threads all contain street shitting grammar because the abject poor are the only people that fall for AMD marketing, it isn't shilling its just stupid people being stupid.

>>60918341
Don't worry I think you'll find a few like-minded tripshits that aren't giggling when looking at your shameless posts.

>>60910476
So I'll keep my 4790k until then

Shit hasnt advanced enough to change up

>>60918418
At least not to another Intel

>>60918425
I'll see how things are at that point. Maybe intel unfucks their shit by then.

I got the 4790k for $250 when it was new so I'm chillin"

File: jewtel.jpg (112KB, 691x771px) Image search: [Google]

112KB, 691x771px

>>60910476
DELID THIS RIGHT NOW

>>60918377
You poor deluded wretch.
Save up your good boy points for your shiny ryzen and project your angst, seems to be working for ya.

>>60918543
I'm sorry I'm still on a shitty

vendor_id       : GenuineIntel
cpu family      : 6
model           : 60
model name      : Intel(R) Core(TM) i7-4720HQ CPU @ 2.60GHz

But buying a good desktop next year sounds really fun :)

>>60918575
I bet you used a college loan to buy a gaming laptop and dropped out.

>>60918582
>college loan
I don't live in Burger.

>>60910476
Intel is done, said increasingly nervous man for 2 decades straight

>>60918493
You're not funny.

>>60918582
>projecting

File: 1489103005792.png (162KB, 633x900px) Image search: [Google]

162KB, 633x900px

They're doing it again
https://en.wikipedia.org/wiki/IBM_and_the_Holocaust
How do we stop them?

File: 1497356671341.jpg (49KB, 552x587px) Image search: [Google]

49KB, 552x587px

>>60910476
OH
MY
GOOOD

>>60917573
They also make fucking baller SSDs and incredible Ethernet chips, I won't deny them that

I plan to upgrade to Zen2 from my i5! Hope it delivers.

>>60918985
I'm giving it a very conservative 4.5-4.7GHz single core and +10% IPC improvement over Zen(Zen+ is a mystery, could be even higher)
So at least you're looking at minimum 20% higher performance per core.

>>60919053
Sweatingcat.jpg

>>60910476
>14nm
>3ghz optimal operation
>Zen runs most efficient around 3GHz

Well would you look at that!

>meanwhile 5GHz for 7nm
That's fucking 70% HIGHER!
And it's drive voltage is actually lower.
And also it seems it has actual high performance 6T SRAM instead of dense-over-perf SRAM it has now.

Zen2 is gonna be something else, the current 14nm process makes Zen the best server and mobile CPU, 7nm will make it the best high performance one.

File: LRM_EXPORT_20170411_145818.jpg (1MB, 1293x1730px) Image search: [Google]

1MB, 1293x1730px

>>60917621
Kill yourself you perma-BTFO'd tripfaggot

>>60918168
pissmark is absolutely shit for that though, single threaded performance from 1600x is actually 50~70% higher than that APU

>>60919785
So what site do you recommend, then?

For example, http://cpu.userbenchmark.com/ could have been useful. But it's not. There's no search box where you can just type in some CPU, add it to compare and repeat.

http://cpuboss.com/ could also have been useful except that it doesn't actually have any benchmarks, it just shows you which CPU has what amount of cache and so on.

>>60910476
>5 Ghz Zen2

Intel on suicide watch

>>60921038
I doubt that will have much of an impact on Intel compared to

my old mom's next laptop / notebook will probably have a Qualcomm or other some other ARM CPU in it. Since Qualcomm's a bit expensive relatively speaking (in their main markets) it'll really hurt of Mediatek and the likes start making light notebooks.

Consider the Doogee Mix, just to throw that in your shilling faces. 6GB RAM, 128 GB storage, Helio P25 2.5GHz CPU and a screen and cameras - for _less_ than $200.

It's going to hurt both Intel and AMD when Windows-laptops by cheap Chinese companies like that hit the market.

>>60921026
I heavily suggest on looking at hard reviews instead of using these search websites, and if you have to compare cpus, look at their respective cinebench scores

>>60921091
Nah, they will get a bad rap because they either won't run programs people already have, or they run them piss poorly because x86 emulation blows on ARM.

>>60921261
This is where I strongly disagree with both you and most of /g/

Notice how I said that this would be my old mothers next laptop. She's currently using a Dell with Fedora GNU/Linux on it. She also has a Android phone. She's happy with both.

She doesn't have any programs beyond what came with Fedora and the packages I installed just in case (libreoffice and so on, but she never uses that anyway). Your illusion that most people have a lot of programs they want to run simply isn't true. Yes, it's true for /g/ but it's not for the majority. This is why x86 emulation, poor or not, also doesn't matter, not even a little. Again, a large part of the market won't use it.

All those Facebook-users who just need a Facebook-machine will be more than happy with something that's equal to a flagship phone in terms of power and has a keyboard and a screen and speakers.

Just to illustrate why I think these machines will do just fine: I have a NEET brother who plays XBox all day. He's installed exactly zero programs on his phone. The stock apps are more than enough for his NEET needs. Neighbors who call me to help them figure out how to get their TV set to the right input source for their TV box when they accidentally pressed the "source" button on their remote also generally have zero programs or apps installed on both their computer and their phone. It really is quite common.

>>60921026
http://cpu.userbenchmark.com/Compare/Intel-Core-i5-7600K-vs-AMD-Ryzen-5-1600X/3885vs3920

>>60915695
If you aren't le trolling xD, AMD has said that AM4 will be around until DDR5, so around 4 years. Zen 2 and 3 will be compatible with your current board, so you can just sell your Ryzen and grab a Zen 2.

File: ibmh912-12.jpg (78KB, 763x475px) Image search: [Google]

78KB, 763x475px

>>60911286
>If at first you don't succeed...

File: 1490946390905.png (97KB, 692x444px) Image search: [Google]

97KB, 692x444px

>late 2018

>>60921396
userbenchmark is possibly even worse than pissmark, it's geekbench3 levels.
Just look at the fucking APIs it uses for measurements, and measure weight is far too biased.

Aggregate scores are cancer, find a benchmark fitting your workload, or multiple, and use that.
Putting everything under 1 number is nice for normalfaggots, but it's disgusting for someone who actually knows a few things about hardware and software.

>>60918582
>projecting this hard

>>60910476
>pic
You vs the guy she told you not to worry about

>>60921665
Stop being assblasted. Dont look at their "total" scores if you dont like them.
They have 1 core, 4 core and 32 threads int and floating point test, everything else doesnt matters.

>>60918543
>biggest troll on /g/
>talking about projecting angst

man, i like me some irony and this one was amazing

>>60910476
Process nodes haven't been beneficial to clock speeds in a very long time.

>>60910476
I'm too old for hype shit. I'd believe it when it's on the shelves and benchmarked.

>>60911004
>tfw own 4790K and considered an upgrade to Zen a waste of money
Zen 2 will definitely be worth it if it's overall 40% superior to the former architecture.

>>60914715
Zen 32 cores 64 threads!!

>>60910835
because avx2 has so wide adoptation that amd needs it so badly..

>>60910817
sorry if i sound uninformed but what is a bingbus?

>>60923751
A meme, he meant Ringbus, basically the spürdo version.
Ever seen that "Bipeline :---------D" image?

>>60910837
OP is wrong. Its a partnership between Global Foundries, Samsung and IBM. After the whole semiconductor industry unanimously agreed that 5nm nodes were almost beyond the feasibility of physics on silicon transistor technologies, Samsung IBM and Global foundries they can ramp up 7nm to full volume production by 2h 2018.

AMD will use the node for sure, but Apple always had first pick on new nodes, since about 2011.

>>60924056
7LP is high performance node, not targeted at low power mobile shit.

File: 1487297768743.jpg (125KB, 960x878px) Image search: [Google]

125KB, 960x878px

>>60910476
I HATE AMD SO MUCH

>>60924394
You should hate IBM even more. That 5ghz clocks is entirely their job.