Zen+

Since everyone is still buying Ryzen because of Coffee Lake paper launch massive flop, what are the real expectations of Zen+ since it's going to be an easy upgrade?

matx with more lanes

-15% die surface
+10% power

I hope this comes out soon. My 1155 motherboard just died, so I'm buying a cheap used one as a bandaid fix. I really need a new system.

50% sea
50%weed

Seaweed

>My 1155 motherboard just died
Jesus Christ don't say shit like that
I'm still on the Ivy Bridge

>Coffee Lake paper launch massive flop
That's a little disingenuous
I just hope for higher clocks. A 8c/16t CPU that gets to 4.6ish would be fucking dreamy!

For real though, Id hate if my ivy bridge shit out on me.

I'm running mine at 4.6 and it has been a great CPU. I know it's supposed to have the mayo tim and what not, but I have been able to cool mine just fine.
I just need it to not blow up until I'm ready to shop.

It was an AsRock E3G3 if that makes you feel any better. I had the CPU running at 4.6Ghz for about 3 years.

adored guy is expecting 4.5ish ghz

Zen+ still on same node ?
4.5 max i guess

They claim +10% clock speeds

>tfw my Z77 MPower suffered a VRM failure right when the warranty ended, killing the i7 in it which is not even overclocked
I will not fall for the MSI meme again.

I really don't want to overhype myself too much. I understand it's new tech so a revision could in theory be a huge improvement.

The only thing I'm hyped for is competition in the industry.

12nm process
10-15% better clocks
Better IMC
Better cache latency
Might come out with a motherboard refresh (B450, X470) that has new features, but backwards compatible

It might be even better that what I posted, since AMD themselves said the Ryzen in its current form was the "Worst case scenario". We will probably see Zen+ at CES or January

Same story here. Haven't felt much need to upgrade my personal rig since I've been running it around 4.5ghz while still having it be at nice temps. Started with 16 gigs RAM, have upgraded the gpu twice over 4 years along with expanding storage. Has been cozy and haven't been dying for a new platform despite it acting mostly as a workstation rather than a gaming platform.

As for Zen+, I'm watching for what the new Threadripper chips will look like based on performance of the 2800X. Quite happy with a secondary 1950x rig I just built recently, and wondering if going up to a Zen+ TR would be worth converting my primary rig to. Not super optimistic, but [email protected] stable would really be enough to win me over.

+10% is the process. AMD could get more from stuff like layout, which can improve electrical characteristics

That's why you return it before the warranty ends, even if it's still working.

I got it when they are clearing old Z77 stocks, there were only S.tooth Z77 and MPower left, and I chose the wrong board. Since it is near EOL already the warrsnty will actually be repair of current board/partial refund. No 1-1 replacement. So no point there.

I am hoping for 4500mhz at 1.4v. And better memory support. If not I'll just keep the fucking sandy and push it even harder. Said I wasn't switching to a tim cpu in 2014 and haven't had to yet.

just hold it unti march or buy coffee lake

Are people still having issues with RAM speeds with ryzen or whatever it was

Generally after the wave of motherboard firmware updates not many people have been having issues. I would still recommend purchasing a Samsung B-Die kit just to avoid any potential trouble. I am getting the expected performance with them on both Ryzen and TR builds at the moment.

At this RAM price point, it's more logical to just care about the frequency at the end of the day.

...

Not same node, Zen+ should on 12nm

Supposedly on 12nm, but the difference from 14nm > 12nm is not a huge leap as far as the process is concerned. 4.4-4.5 stable is a reasonable expectation if you don't want to be disappointed.

Layout, wiring optimizations, changes of design rules (lets go for relaxed CPP and HP libs!) and whatever else.

>Coffee Lake
>Massive flop

I dunno I really like mine a lot. Honestly works better than my ryzen system ever did.

> Since everyone is still buying Ryzen because of Coffee Lake paper launch massive flop

why do Sup Forums shills always have to lie? you can play cheerleader for your favorite corporation WITHOUT making shit up

>Bait so obvious even blind fish wouldn't hook and sinker.

>being this deluded

AMDrones are intense.

It's not bait, Sup Forums AMD shills actually think this. Go read any AMD thread. They literally can't praise an AMD product without trash talking Intel. It's pathological.

But CFL-S is not selling that well.
It's a rushed lineup.

Availability for high end coffee lake cpu's is actually worse than even Threadripper at the moment. It's hard to get one and not pay an inflated price or end up waiting for a backorder. The problem is somewhat bad in the states, but entirely catastrophic in Europe and elsewhere. Mindfactory directly reports their cpu sales so it's confirmed this has been the situation, rather than simple speculation.

>threadripper availability is bad
What?

Yes, not as many places have decent numbers of Threadrippers in stock, or carry them at all compared to 1600/1700. It's fine if you're trying to buy one, not if you're trying to buy enough to resell. There are more 1950x available to purchase at the moment than the mainstream coffee lake chips. This is demonstrating a poor availability of a supposedly mainstream chip.

He simply said coffee lake was a paper launched flop, which is true.

Shut up goy

First thing is zen with minor tweaks on a better process node. Expect higher clocks probably up to 4.4-4.5 but maybe higher. There was a rumor of a zen chip hitting 5ghz on all cores but i strongly doubt that. 4.4ghz is enough to basically eliminate the permormance gap between stock CannonLake and Zen. So unless you are doing high end Z mobo with unlocked cannon lake there wont be any point in going intel anymore because the mobos and corecount will be better at lower pricepoints on AMD.

Zen2 is going to be a complete refresh with probably 10-15% IPC gains and an improved Infinity fabric. Fabric is already great but reducing latency between CCX modules will massively boost efficiency. Even if they cant improve the fabric itself they can improve it by improving compatibility with faster RAM.

On top of IPC boost zen2 will be on 7nm so more power efficiency and higher clocks. the gaming performance gap between intel and AMD will be completely gone and Zen might even be a bit faster than intel when zen2 comes out. Also core and thread count goes way up. Expect 50% more cores in each price bracket. Assassins Creed: Origins set the new standard for properly utilizing CPU threads and scaling them. If more cores/threads game engines will be able to add more visual effects and increase the ammount and complexity of AI/NPC in games. Big fucking deal if you an open world RPG fanatic.

The trick is to ensure GPU power keeps up with CPUs and the only way for that to happen is if AMD's NAVI architecture in an MCM (multichip module) basically you thow out the monolithic die design and create smaller ultra efficient 1-2,000 shaders dies and then connect them with infinity fabric type of design to scale up to the huge 12-16,000shader GPU that you want. AMD is limited in their R&D budget so they will likely use the vega20/10 architecture and just scale it to mesh for this first MCM design.

First Navi chips will likely be high end flagships. The second tier will be 4x2,000 shader dies on an interposer. Look at the current Volta GPU and just divide the die into 4 sections with the four HBM stacks on each side. They will use the lower binned HBM2 that vega currently has.

First tier flagship GPU will likely be 4x4,000 shaders. The dies wont be that big on 7nm and this yields a titanic 16,000 shader behemoth that can crush 8K and beyond. Use the HBM2+ modules for this with higher clocks.

The power draw of these cards is dependent on what the core clock is and as we all know power consumption is not linear is very close to exponential so all you need to do to fit thoe 16000 shaders into a 350watt or even a 250 watt TDP is simply to lower the voltage to that point and see what the highest stable boost clocks are at that voltage.


This design can work and destroy nvidia but it hinges entirely on AMD being on top fo their Drivers for these GPU's they will only be naturally saturated on Vulcan API's and DX12 APIs where the game devs properly leverage Asynchronous compute. AMD needs to either have game release ready drivers for every major title to ensure no bubbles in the pipeline OR they need a command engine that takes DX11/DX12 code and breaks it down and threads into all the idling shaders properly. All that horsepower wont mean shit if it cant be leveraged.

YOu see the same problem with the new 5,000 shader Voltag GPU. In benchmarks its performance the same as a 1080ti with 2/3rds of the shaders because its not being fully saturated.

Explain.

...