Well, I guess that's it for Coffin Lake's time in the spotlight

Well, I guess that's it for Coffin Lake's time in the spotlight

digitimes.com/news/a20170927PD212.html

>AMD to launch 12nm Ryzen in February 2018, says mobo makers

Other urls found in this thread:

videocardz.com/newz/amd-to-launch-pinnacle-cpus-in-february-2018
twitter.com/SFWRedditVideos

cool

now i can buy i9s cheaper maybe

>2nm reduction
why even bother?

Nvidia started it

>Planning on upgrading next year
>Have to choose between this and cofveve lake

Think I'm going team red, guys.

videocardz.com/newz/amd-to-launch-pinnacle-cpus-in-february-2018

Because it comes with what AMD needs most right now, clockspeed.

both are idiots desu, intels cpus and igpu's are enough for me (i dont spend much time gaming) i'd never waste a dime on those companies

AMD should lie about their TDP with Pinnacle and clock it higher, seems like nobody cares when Intel does it.

>Ryzen+

>upgrade version of Ryzen
Probably better leakage control, which means higher clocks. I'd guess it would probably be 4.5Ghz

>digitimes
literally more pathetic than wccf
for one of their articles they literally sourced wccf and the funny this is the wccf articles was sourced from digitimes
they're just throwing shit at a wall until something sticks

Cecily best fucking girl.

Poor Intel

Don't worry, I'm sure some glue can fix that.

Have you even read agners programming guide? It's clear as day that ryzen is miles ahead of Intel. You'll just have to wait for the software to adapt. And most of it is in compiler optimizations.
Seriously they've gotten single core threading to be next to costless. They're pulling 4+ instructions per clock in the trivial cases. Their instruction cache and OOE is way better than previous iterations.
It's entirely on Intel to get off their ass and make some tech again.

10-15% more clock? 4.6ghz confirmed.

Wrong, Zen is at its limits already, it can't go over 4.2GHz

Reminder that Zen has a fuckton of integrated voltage regulators that were deliberately disabled on desktop chips for some reason. Current Zen products are not running as efficiently as they could.

Zen has peaked, there's no more performance there.
Meanwhile Intel can even get 18 cores to 6.3GHz thanks to its forward looking architecture and superb fabs.

Its a 15% area reduction and 10% higher clocks.

The clock wall with Ryzen is from the 14nm LPP process, and it isn't a hard wall either. It isn't thermals, it isn't power, its voltage scaling due to the process.

The Zen core arch has absolutely no problem scaling in clocks up to 5ghz and beyond. Overclocking records at launch well proved that.
Dumb kid.

̣>ln2
>relevant

The reason was to avoid cannibalising real server sales as much as possible

-t. Brian

Inb4 nitrolake (or some other shit) 22 cores 250W TDP "answer" from Intel

Freezing a chip only effects thermals and leakage characteristics, it doesn't alter clock synchronization within all areas of core and uncore. If a chip can reach 6ghz under LN2 then the arch can scale that high. Thats all there is to it.

...

Higher clocks, because of LP (Leading Performance) compared to LPP (Low Power Profile) used in Ryzen. Also more denser

Go home, Brian. Your infantile shitposting is outdated and tiresome.

>again
When was the last time Intel actually made a big improvement in tech? The Pentium III? Pentium Pro? It's certainly not the Pentium 4 or Core 1/2/i lines, at least not meaningfully.

at least Glofo's 12nm has an actual area reduction compared to their 14nm, TSMC's 12nm has none

7nm pushed back to 2019, confirmed?

zen can't go above 4.2GHz because of the silicon, not because of the architecture
just change the process and it's good to go

7nm Epyc parts are slated for Q4 2018, desktop chips following 2019.

I think I remember an incremental improvement before 7nm from the Zen roadmap.

hbv'er lake

Fusionlake

Firelake

>from Fiery Bridge to Firelake
New architecture confirmed Fire Mountain

I can't wait to see mobile CPUs from AMD. When will it be released ?

They don't even have to name it that. It'll just transform into it once you start using AVX.