Ryzen 3000 4.5GHz all core and more

bilibili.com/read/cv2516248/

good news
>ryzen 3000 hitting 4.5GHz all core, ~8% IPC improvement
>40x pcie4 lanes, some shared with SATA interface
>8x 3.1 gen 2 USB3 ports

bad news
>b550 might not have pcie4, only x570
>a320 and b350 boards might not be compatible with ryzen 3000 due to shit VRMS
>memory controller not improved
>b550 launch not until August or September

Attached: AMD-Ryzen-3000-3rd-Gen-Zen-2-Desktop-Processor-2060x749.jpg (2060x749, 116K)

Other urls found in this thread:

bilibili.com/read/cv2516248/
en.wikichip.org/wiki/amd/ryzen_5/2400g
valid.x86.fr/records.html
tomshardware.com/news/amd-bulldozer-overclocking-record-fx-8150,15793.html
twitter.com/NSFWRedditVideo

chink article also says +15% improvement on last gen generally

so how will these stack up to incel's single core?!

>4.5 Ghz on all core
Kek.
You could hit that years ago on aircooling without having the benefit of 7nm.
How embarrassing for AyyMD

>ryzen 3000 hitting 4.5GHz all core
bery nice
>~8% IPC improvement
expected more, still noice, going to be like a 12% improvement over my 1700 in IPC, coupled with 4.5ghz on all core (probably overclockable higher), going to be fucking noice
>40x pcie lanes
what the fuck
>memory controller not improved
FUG

>4.5 Ghz controlled sample rate
Neat. So there's room for high clock selling at higher price and lower clocked hitting 4.5 Ghz target at lower price.

>40x pcie4 lanes

wew lad. quad sli + 4 nvme drives and there is still bandwidth left.

>>a320 and b350 boards might not be compatible with ryzen 3000 due to shit VRMS
Sounds like bullshit. There's no reason why the VRMs would be insufficient unless they're somehow running even hotter and more power hungry than the 2700x which almost hits 140W in a cache heavy test. Even shit-tier motherboards can handle the 2700x at stock and the VRM situation is mostly overblown by retards

yeah, the leaks are surely real and it will not be bulldozer all over again

here you go

Attached: Capture354.png (851x898, 200K)

gg

Attached: 1537744055076fixamd.png (575x641, 196K)

>because all GHz are equal

Even with 15% improvement it will pulverize Intel sales.

where would 8% IPC improvement put it in comparison to Intel? in terms of clock for clock. like 1c/1t same clock speed?
Also curious if the 4.5ghz is gonna be a brickwall like the last two gens of Ryzen, or we can finally achieve something with decent cooling and aggresive voltage.

I bet that 4.5Ghz is a golden chip, the real ones aren't getting past 4.4Ghz, also is gonna suck for gaming still, screencap this SENPAI.

>chink article also says +15% improvement on last gen generally
I think AMD was claiming a 20% uplift in "performance", so IPC+clockspeed.

Only 12%? The 2000 series I thought was around 6% overall, and in gaming tended to be more around 10% clock-for-clock improvement in gaming.
Higher was claimed, but BIOS updates for the 1000 series actually improved IPC a bit so it wasn't quite that high, I know.

You're lying. The 6900k did 3.7Ghz all-core turbo. This is a 800mhz improvement at lower power. Broadwell-X was also lower IPC than even first gen Ryzen is with the BIOS updates.
>Ryzen gets more IPC gains from BIOS updates over a year than Intel gains in 5 generations
4.5 is clearly the all-core turbo, not OC.
It was seemingly hitting 4.3-4.4 all-core during the demo earlier this year.

PBO and BCLK OC on the higher end chips will likely still be better than an old fashioned OC, yes.

It's currently 3-5% behind, so Ryzen 3000 series will be ahead.

>9900k is still king
LOL

9900k has lower IPC. So at 4.6Ghz all-core turbo, it's actually slower.
Plus HT is garbage compared to SMT.

>bilibili.com/read/cv2516248/
40 pcie
This is double. They do want us to do some NVMe raids.

>July
God damn it, I can't wait that fucking long. GIVE IT TO ME IN MAY!

Yeah it's annoying. Especially if they wait longer to release Threadripper compared to last year

I'm probably going to just buy R2 and then get threadripper with the new (I assume) mobos when they launch. There's no fucking way I'm waiting 4 or 5 more months. I'll sell the R2 system or one of my others, idgaf.

I can't wait to have nothing but nvme raid 10s in an itx system.

OH NO NO NO NO

Attached: jim.png (1531x902, 842K)

>12 cores at 4.5GHz
Bretty gud, if that is all core turbo then 1-2T turbo should be 4.7-4.8
Also I don't exactly trust the memory controller information since the radically different architecture of Rome requires a new memory controller.

seems like an accurate summary based on those numbers. not as good as many thought but still a solid improvement
i imagine the landscape will be flipped in a few months with AMD chips having 5-10% or so edge vs intel in the desktop market

how good is the 3600G going to be? will it be equivialent of a nvidia 1050 or a 1060?

No. It will still be limited to 720@60 1080 20-30.

seems unlikely itd be as good as a 1060 but im not an expert on AMDs APUs. its also not going to be on the 7nm process like the non igpu chips are going to be. itll be on the current zen+ 12nm node
i read the driver support for the apus was or is very bad compared to radeon or nvidia cards

say it aint so

i heard it would at least compete with the 1050

i just want to play tropico and age of empires 4 and a match of battlefield every now and then. id be happy with battlefield med-high @ 1080p at 60 fps

>40 lanes
fucking please be true

It's not that HT is slower, it's that ryzen has a more beefier back end

If it competes with a 1050 Nvidia is dead.

>5xx motherboards release on July
so should we assume that will be the same release date for the CPU's as well?

Buy a 9900k before its too late

Probably a 1650 at least

you're shitting me

It will likely btfo a 2080ti without breaking a sweat! That's the power of AMD!

>16 cores at 4.5GHz

Attached: Ryzen3000.jpg (500x281, 75K)

4.5GHz 16 cores is pretty crazy.
5GHz turbo basically confirmed.

This is going to completely rape Intel.

>Boost clock and all core is the same
I have bad news user

Attached: 2008_tropic_thunder_-1.jpg (800x469, 230K)

The popular belief is the release date will be 7/7.

Please goy, do the needful

If the midrange isn't very good value I don't think it's going to make more than a moderate dent in intel, even if the high end makes intel look like a joke.

If you dont buy Intel you are anti-Semitic. Just like this thread.

The whole thing really comes down to how they're going to price these things.
AMD is in a very good position where they can take over a good chunk of the market by undercutting the fuck out of the competition's pricing.
Or they can mess this up and go full Hebrew and try jewing extra money for their midrange CPUs and shoot themselves in the leg in the process.

only buy intel aviv goy

low prices are anti-semitic

By law you will not boycott the great products of the State of Israel.

They can do both. Undercut and increase margins.

i really doubt theyre going to raise the prices. intel and nvidia raise the prices because they can. even if rumors are true and zen2 chips are slightly ahead, amd is not that well known. then again i dunno anything about business so maybe im way off

death penalty for those who criticize jews, as per the noahide laws!

Do we have any knowledge of the cheaper shit being announced?

Is it saying the memory controller has improved a bit, but not as much as people rumored or the chip overall? It does say some sort of improvement occurred. Shame if AMD couldn't do much with it, since that's been holding it back the most.

we'll get em next gen AMDbros

>>memory controller not improved
Into the trash.

is it really that bad now? i thought they improved it a bunch from zen to zen+? i know it was not that great on 1st gen zen chips but never heard complaints with zen+

>not buying it after zen2 is announced for better prices
I thought jews liked money?

Attached: 1547325061063.png (1056x762, 274K)

I'm not saying it's bad per say, just that it seems to be the main limiter, particularly in games, which is what the broad consumer market seems to care about.

I thought the 3850x was supposed to hit 5.1ghz turbo?

i dont think intel drops their prices

>7nm will be good and cheap guys

Attached: 8167B1F0988941BBB7592F5479B82131.jpg (970x647, 99K)

>being this much of a patiencelet

Yikes

Go drink some bleach, shit for brains.

Why SOPA? IF fabric runs twice as fast now.

Still means you can't have more than two sticks of cheap RAM and be able to go 3200+ with low latencies

shit, because its just zen+ with vega, APUs are one genertion behind

No. That can't be right. That may be true with the mobile, but not the desktop.

Core count is good
Temps and TDP are good
IPC gain marginal at best
Memory controller is still bad-ish
Clocks are good for the core count, but too low overall.
So another Server chip repurposed for desktop.

Wait for zen 4

I waited and the time has finally come.

user, I truly hope you're in this thread. You ephemeral, insufferable, transactivist niggerfaggot.

Attached: Post this When Zen2 releases.png (1021x707, 46K)

>release date is my birthday
now I just have to get it, dont I?

This is a TSMC node, not the originally planed IBM designed process that global foundries shut down because lack of customers and high cost.

>This processor is based on AMD's Zen microarchitecture and is fabricated on a 14 nm process.
en.wikichip.org/wiki/amd/ryzen_5/2400g
APUs are one generation behind user, whether you like it or not

The I/O is still on the 14/12nm process, now has to contend with multiple chip signalling. While decoupling IO from the cores will give some level of signal integrity improvement, decoupling IO from the cores also increases distance and power needs.

I would imagine that 4133 is going to be possible on Zen2 and not much higher.

It was never going to be on IBM's 7nm, user-kun. That was all superFUD to distract. There's no way they redesigned quick enough to still hit shelves on time target, and there's no way a redesign only cost a few months (I was hoping for a March reveal/April release)

But OPs pust ITT is also nothing but out of the ass speculation. You have to wait longer.

I know one thing - these CPUs are going to OC to 5Ghz and someone will try throwing that in my face.

I know one other thing - It's impossible to double the cores, cache, and FPU and provide a 20% clock increase at the same TDPs.

>beaner from SA
Omfg. The memories, the laughter.
His name is huangra.

> Not 5Ghz

happy I bought 2700x and didnt wait a little longer (was on 2500k for 7 years)

now compare the transistor densities, brainlet

>a little longer compared to 7 years would've been unrealistic
consumers are strange

They do. They just do it by announcing an identical product for cheaper. Look at the current i3.

Smells like using intel's marketing bs and combining cpu and chipset lanes together where most of them are limited to the 4 lanes pch link in most cases. From that leak it seems:
16+4 from am4 cpu; 20
8+4+4 from x570; 36
There are four extra ones, could be anything, hopefully more extra from the cpu itself or four little ones for network/wifi/extra sata lanes etc.

4.5 Ghz is the standard they want their 3600X to hit. I think 4.7 will be hit my 3700X variants, similar to how you get 4.1-4.2 with 2600X but 4.3 for 2700X.

They use the same dies

>It was never going to be on IBM's 7nm, user-kun. That was all superFUD to distract. There's no way they redesigned quick enough to still hit shelves on time target, and there's no way a redesign only cost a few months (I was hoping for a March reveal/April release)

Zen 2 was in planning for over 4 years. They should've have known sometime in the first two years to not miss dates but a switch is totally possible

>memory controller not improved
lies

> 4.5GHz all core
Remember the 8700K reaches 4.3GHz all cores, and has a boost of 4.7GHz, and easily overclocks to 5.0GHz.

If zen2 is getting 4.5GHz all core, I would guess it reaches boosts at least to 4.8GHz.

Intel isn't AMD.
AMD cpus overclock fuck all past their boost clocks. Stop comparing apples and oranges and getting your hopes up

AMD has fastest clock speed world record Intelliot dumbo

valid.x86.fr/records.html

tomshardware.com/news/amd-bulldozer-overclocking-record-fx-8150,15793.html

As an ass man I look forward to ditching my intel and getting some zen 2 booty

>Using intel misleading propaganda on AMD

>baseless speculation
its a new uarch, new process node, you know absolutely nothing about it

So it's fine to compare it to a completely different node and uarch when the conclusions are favorable?

Confirmed hold for PCIE 4.0 5nm and DDR5. Good news for people looking to save some money but still not enough to justify a major system overhaul.

no, I was merely commenting on
>AMD cpus overclock fuck all past their boost clocks.

You people are idiots, none of this is confirmed except for one cpu, definitely 8/16, that's it, the rest is speculation

As long as you feel right

It can still give us a hint of what the rest of the lineup is going to be.

Kinda hard to not be able to speculate in a market that's slowed to a snails pace in improvements and marketing practices

Yeah but if we follow that logic then the smart thing to do would be not talk about it.
And I like talking to you guys even if most of the conversations is just yelling back and forth about how we're all the right ones and everyone else is wrong and a shill