What are the rumors on the next Ryzen lineup?

What are the rumors on the next Ryzen lineup?

I wanted to build an entire rig around it but I waited too long, then the ram, video card and ssd prices went crazy. And if they ever come back down I expect the next Ryzen CPUs to be out by then.

Other urls found in this thread:

pcisig.com/developers/integrators-list?field_il_comp_product_type_value=All&keys=AMD 400
warosu.org/g/?task=search2&search_tripcode=!!1w3SeuwCso0
twitter.com/NSFWRedditVideo

ssd prices arent too bad right now but ram and video cards did go up again.

Rumor is they're supposed to announce zen+ early next year, but I'm not sure how real that is.

It's suppose to be announced in February and some motherboard manufacturers already said they would have a new chipset out in March. So they probably get a March release.
Zen 2 which is the 7nm shrink was supposed to be late 2018 but now is looking more like early 2019.

No one knows anything about either Zen+ or 12LP.
Current AMD actually reminds me of Andy's Intel in that.

>buying poozen
if you want a processor from 2013 just go buy a used haswell.

GPU prices might go down again, since crypto is
>FREEEEEE FALLLIINNGGGG!!!

So we'll see.

pcisig.com/developers/integrators-list?field_il_comp_product_type_value=All&keys=AMD 400
Looks who's here.

Nothing's known about it, beyond GloFo's nebulous claim of 10%+ better performance for the new process that it's using (which could pertain to anything from clock speeds to power consumption). It's coming in Q1 2018 though, so we'll find out soon.

>Had a look
>Dip to $11k today

Jesus fuck what happened?

It's crypto.
It bubbles then bursts, again and again.

>kek amdcucks will havrnto buy nrw mobos again lol after bashing intel dumb fucking poojets

>new poozen needs new mobo
>amd lied about mobo compatibilitys

*breathes in*

Samefagging is bad.

HAHAHAHAHAHAHAHAHAHAAHAHHAHAHAHAHAHAHAHAHAHAAHAH
HAHAHAHAHAHAHAHAHAHAAHAHHAHAHAHAHAHAHAHAHAHAHAHAHAHAAHAHHAHAHAHAHAHAHAHAHAHAAHAHAHAHAHAHAHAAHAH

I heard rumors of Zen+ having 4.5ghz boost clocks.
That could be a game changer, even without IPC gains.

Impossible, only Intel can reach that

>make taller fins!
Wow so hard.

I bet this favela monkey even coded a script which refreshes catalog on Sup Forums and alerts him every time something about ryzen was being posted.

Jelous AMDpoors

Price went up, people sold, price went down. Standard trading

>Nothing's known about it, beyond GloFo's nebulous claim of 10%+ better performance for the new process...

That would actually be bretty gud. It would mean 4.5 ghz on top tier Ryzens, which would shut up most Intel goys and gaymen.

Try reading the bit of the post that you edited out.

Performance implies clockspeed, as the core most likely remains the same. LPP process already is very efficient anyways and process refinements on an existing uarch always brought some extra clock.

Yer both retarded, GloFo compares it to "INDUSTRY 16nm" which TSMC has, not their own 14nmLPP, which is slightly worse.
Also it was 10% perf at 15% lower power but nobody cares about that number, it's meaningless without knowing the fin height(what you need for high clocks) and libraries.

It will suck at games.

We are nerd-wanking mate. In the words of the representative there is always "10% improvement across the board". If intel's babbling about 7% per generation IPC increase was true, Covfefe Lake would be 50% faster then Sandy Bridge clock for clock.

amd btfo

>If intel's babbling about 7% per generation IPC increase was true,
It's true.

In SysMark.

They didn't lie, technically.

Next Poozen lineup will catch up to the 7700K, but still far from my 8700K.

New GPU line will probably catch up with the 1070ti, but only the generation after that will catch up to my 1080ti.

1st plunge of the bubble cycle, now it's in the Bear Trap, then comes the second plunge that makes the first look small.. after that, slow and steady climb / stagnate into oblivion.

You people think it's BEAwesome?

warosu.org/g/?task=search2&search_tripcode=!!1w3SeuwCso0

It doesn't "NEED" a new motherboard though
It's still AM4 and compatible with previous motherboards it just has newer features

Either way both are filtered since they're cancer.

It has always been clear zen2 is coming 2019 if you're irrational dreams are not coming true that's your problem.

It wasn't clear, 7nm HVM was 2H 2018 which AT BEST would be July 2018, and Zen2 tapeouts was in December so more than enough time to fit 2018 even if it's a CoffeeLake paper launch.

that fucking graph kek

Rumors say the R7s are coming out in February and the mid range and low range models should come out the months after.

However Rome seems to launching late 2018 and that's Zen2, probably in Cannonlake-Y quantities.

I've always heard from Lisa and other execs that LATE 2018 (Nov-December) to early 1H 2019 (Jan-May) is the release window. I think that's reasonable, you need a buffer for technology like this and a 7 month buffer is generous. If they manage Q1 2019 that will be very good, and Intel will be on suicide watch.

>7nm shrink
you mean 10nm?

>2x the density of 14nm for GloFo, Intel and TSMC
>GloFo calls it 7nm
>TSMC calls it 7nm
>Intel calls it 10nm

14/2 = 10

t. Intel

Shaving off cache latencies.

updated. was already done for TR and epyc

How come no one can get latency for L1 down below 4 cycles?

Wasn't IBM's previous power able to access L1 at 2 clocks, or was that registers?Can't remember where I read that

In 1 nanosecond light only moves by 30 centimeters max, at 4 Ghz that's 30 centimeters every 4 clock cycles.

The lines in your CPU are anything but straight, if your L1 is 64K that's still reasonably small so you can fit it close to where it's going to be used, but if you add up the latency of starting a memory read, doing the cache lookup across 1K lines, time to read the memory cells, and all the length of the wires both ways there isn't much margin left.

moar lanes pls
mATX or ATX with 3.0x8, 3.0x8 and 3.0x4 (or 2.0x8)

or 3.0x8, 3.0x4 and 3.0x4

>Ryzen

Vega 64 already beats the 1070ti

>tfw you google how far light travels in a nano second
>it is 30cm or roughly 1 foot

user... I thought you were talking out your ass but I was in the wrong. Good on you.

JESUS

Stay mad Incuck.

Will it support muh av1 encoding?

...

>Wasn't IBM's previous power able to access L1 at 2 clocks, or was that registers?Can't remember where I read that
>In 1 nanosecond light only moves by 30 centimeters max, at 4 Ghz that's 30 centimeters every 4 clock cycles.

velocity of propagation is even a bit slower in copper wires, but the name of the game is really counting up all the wire and gate delays in the so-called critical (slowest possible) path.

most L1 delay comes from needing to support different addressing modes (i.e., register base + various offsets, including scaled offsets sometimes) which needs a mini ALU to generate the correct virtual address, then pulling multiple 64B lines out of SRAM, feeding them to a mux, translating the virtual page ID to a physical tag in parallel in the TLB, then getting the selected line from the mux back to the register through an execution read port.

Itanium lowered L1D delay by restricting addressing modes, Power does it by burning space and tons of energy (purportedly multiple Watts per L1 alone).

>#currentyear
>#2017
>buying poozen at all

>Being this much of Intel shill

Pic related, you should try it sometime

that fucking chart, man.

...

...

the current AM4 mATX lineup is very, very lacking and all my hopes go there for B450/X470

...