INTEL IS RIP

impossible
they have too much monie

When @TMFChipFool is taking a piss on Intel you know shit's fucking going south.

Holy fuck, Zen 2 is going to be insane.

yeah when you are just 5% ahead on games and literally some light years behind on productivity to the point that 7980xe is needed to beat a 2700x
you are right amd will never catch up its impossible to develop backwards cpus

S
>inb4 intel goes full netbust again

>up its impossible to develop backwards cpus

>What is Bulldozer
>What is Netburst
>What is IA64

i wouldnt put bulldozer next to houseburst at all
bulldozer at least got fixed with piledriver

oy vey, that's antisemitism.
literally, holohoax.

>risk production 1H'18
holy shit the future IS now

Piledriver should have, at most, been the CPU released back in 2011 instead of Bulldozer while another actually decent arch was in the pipeline for a 2013/2014 release to counter Haswell.

By no means it should have been the last iteration of an underperforming arch that lasted like half a decade.

zentoo for gentoo

kek

Piledriver is to Bulldozer what Cedar Mill is to Prescott.
Prove me wrong.

kek

Attached: 1520011954844.png (606x248, 53K)

What do I press to barf, piss and shit on Intel's grave?

Buy a 1950X

More like 2950X

Attached: DB84B170-B045-4F0C-804B-341D49246308.jpg (237x240, 18K)

>decades
Confirmed underage. Intel's reign didn't even last a decade.

S

Yep they are already making samples.
Ryzen 3000 series is on track for 1H '19 as much as most people assumed it'd be relayed to 2H '19 or late 1H at best.

Now the questions is whether I'm upgrading to a 3700X, or if they might actually make a 2600XG with 6 cores an a more powerful iGPU that's nearly RX570 performance which can be used for async compute while the discrete GPU does rendering...

S

>17m gates/mm^2
>intel 10 nm has over 100m gates/mm^2
AHAHAHAHAHA, INTEL IS AHEAD EVEN ON ITS FAILING PRODUCTS!

newsroom.intel.com/newsroom/wp-content/uploads/sites/11/2017/09/10-nm-icf-fact-sheet.pdf

Attached: 1455357803605.jpg (796x805, 115K)

>Isn't IBM going to use it?
Since when did IBM use Intel's fabs? Since when did anyone but Intel use Intel's fabs?

>40% performance boost over 14nm
So what is their excuse going to be when Zen2 doesn't clock to 6 GHz?

This thread is VERY Anti-Semitic

I demand the administrators of this site to take the thread down as soon as possible

Attached: 1524869769291.jpg (384x384, 14K)

>inb4 it still sucks at games and i am still disapointed

>tfw you realize you were on Jow Forums shitposting when bulldozer was released
I'll never be able to leave, will I?

If 40% is true it would be 5.45 base clock, 6Ghz turbo. Seems unlikely though.

>Seems unlikely though

Attached: this-seems-like-a-shitty-situation-and-the-understatement-of-21316616.png (500x391, 75K)

you'll leave when AMD pulls ahead

Do you know how CPU's work

>tfw 48c/96t TR4 7nm

Attached: 226.jpg (900x900, 266K)

This, built first pc at turn of millennium, AMD and Radeon was very competitive

Nah, probably 24c/48t top TR

48c/96t is Starship

Fix your voltage, Brian. Then get in the oven.

Attached: 7nm Comparisons.jpg (685x599, 69K)

oy vey

>Risk production
>1Q-2017
what a shitshow.

...

This is a louder death knell for digital computing.

The golden age died back in early 2000s and 2020s will be the death of digital computing.

Laws of physics and scale of economics have finally caught up.

If Intel can't scale down beyond 10nm without serious yielding issues then I doubt the rest of the semiconductor industry will fare better.

I used to think this about gasoline engines in the mid 2000 and now suddenly we have tons of hp from naturally asperated engines

It is all from tighter control from computers and better mechanical precision (thanks to CAD) so less is loss from inefficiency.

The basics haven't change though from little a century though.

Yeah jusy strange it took them until seemlingly the last 5 years to really get the ball rolling. I suspect the recession and high gas prices were not helping.

>falling for the marketing slideshows
Jow Forums deserves to be botnetted
I'll be very surprised if actual performance under actual workloads for actual production designs reach half of those numbers.

Wow TSMC and GF are super close in capability there.
But I assume this is from early 2016 and no longer true.

Those Intel 10nm stats are also probably better now, it's just that their yields are terrible.
They're probably especially fighting to get their Vmin down.

Frankly, Intel's should probably be called 9nm and the others 8nm.

>Those Intel 10nm stats are also probably better now,
They're actually worse, Intel has reiterated 2.7x scaling down to 2.4x scaling a few days ago, now they're like 15% less dense.
And they were losing in cache density to TSMC/GF/Samsung already before the change.

>muh games

>comparing differing foundries process nodes as if they are comparable.

fucking mongs

intels 10nm, matches or exceeds others 7nm in all but voltage.

im not amd fag, but honestly the performance differnce is like what 5-10% IF the GPU isnt bottlenecked. This is literally nothing.
And amd CPU cost 100$+ for same core count.
Now spend that 100$+ usd for better GPU and sudenly you have advantage in performance in games..

No it doesn't you filthy kike.

>OHLY FUCK MY DENSITY

feature sizes are comparable between 10nm and 7nm, at least with intel 10nm. Nothing changes that.

>higher voltage
>less dense
>less xtor performance until 10++(lol 2020)

Intel has nothing, besides mouthbreathing kikes like you defending it of course.

It's performance, not clock. Can be anything.

>2020
>implying they'll release 10nm and 10nm+ in 2019

>If Intel can't scale down beyond 10nm without serious yielding issues then I doubt the rest of the semiconductor industry will fare better.
Keep telling yourself that, everyone will keep moving forward, Intel will either spin off their fabs or be left behind.

more like a 3950X

>ITT Waaaah WAAAH there's no way Intel can be behind the other fabs the shitshow on early 14nm and delaying 10nm for 3 years are just tricks!

Disgusting, filthy kikes.

Stop lying, satan

S

let's do some math then, the 1800X has:
>3.6GHz Base
>4.1GHz max XFR boost
3.6 + 40% = 5.04GHz base
4.1 + 40% = 5.74GHz max XFR boost

the 1950X has:
>3.4GHz base
>4.2GHz max XFR boost
3.4 + 40% = 4.76GHz base
4.2 + 40% = 5.88GHz max XFR boost

for the the 1650X:
>3.5GHz base
>4.3GHz max XFR boost
3.4 + 45% = 4.92GHz base
4.1 + 35% = 5.72GHz max XFR boost

forgot all core boost, both 1800X and 1950X have 3.7GHz
3.7 + 40% = 5.18GHz all core

what the fuck is a 1650X? did you get dropped on your head?

are you retarded? thats total performance boost, not only clock boost, it could be something like 20%IPC and 20% clock boost.
And max XFR could be somewhere around 5ghz

a smaller node doesn't give IPC by itself, only clockspeed

Threadripper would be a fucking monster

but they are claiming zen 2 will have good IPC boost. So assume 40% performance all comes from clocks is retarded.

that's a slideshow from GloFo, not AMD
that's only talking about the node
globalfoundries.com/sites/default/files/product-briefs/pb-7lp.pdf

10% more IPC, 30% more clockspeed, screencap this

100% they are talking about this with Zen2 in mind

that makes sense, considering the low clocks on zen and zen+

this would be false marketing

Not according to any numbers that have been published.

The numbers are wrong unless they come from CTS labs.

that's comparing to their own 14nm tech (ryzen 1)
on the one hand thats fair enough, since ryzen+ was never meant to be ryzen 2, on the other it's of course blatant marketing because it looks bigger and makes consumers think they're comparing to intel 14nm+++ (soon to be ++++++ when they the 10nm move a few more times)

>slower than 14nm++ until 10nm++
>first gen 7nm is already faster than 14nm++
>10nm++ is coming at best in late 2020

anyone who cared about that presentation would know better than comparing it to intel's 14nm instead of glofo's

kekked

>up to 17 layers

Attached: mqdefault.jpg (320x180, 10K)

finally i'll be able to build my shrek AI chip

the global foundry process is targeting 5ghz as its optimal clock, like their 14nm targeted 3ghz

not sure about you, but 5ghz as a minimum will be what I expect

number 1 was getting infinity fabric 1:1 with memory speed, not 1/2

they only change when government forces them to change, look up old mpg, a president telling them to get fucked, and then magically they got better mpg by a large amount, then a cocksucker repealed the need, and they stagnented again.

they are only talking about the node here, and the node is targeting 5ghz, their prior node was targeting 3ghz
that 40% is pure clock speed boost and its possibly an underestimation.

zen wise, they are likely adding some fpu bullshit along with more cache, likely lower cache latencies, a planned 1:1 infinity fabric speed, along with other tweaks.

yeah, in the paper it's literally stating at LEAST 40% higher performance
if AMD manages 10~15% IPC for zen2 they'll go far ahead intel with that node
it'll also be a great node for RR seeing that they can get something like a 1600 into 25W TDP

hmm, let's see
it's almost 6:00 AM in brazil

???

intel macacos are going to be here soon

>a planned 1:1 infinity fabric speed
can i get a /quickrundown/ on that?

Attached: 1523816018374.png (1440x1557, 738K)

Current Ryzen Infinity Fabric runs at half the speed your RAM is running at, so 1:0.5.
Zen2 is planned to run at full memory speed, or 1:1.

>Zen2 is planned to run at full memory speed, or 1:1.
thanks

Attached: 1523029116592.jpg (1000x800, 82K)