AMD's new Ryzen 7 2800X teased: 12C/24T at up to 5.1GHz

Discuss

>TweakTown
Eh.
>teased
Where.
AMD's own press-release might please.

Are Incel, dare I say it, finished and bankrupt?

..but can it run Crysis?

>benchmarks as well as a 2012 laptop i5

NOOOOOOOOOOOOOOOOOOOOOO

I'm coming at just the thought.

I FUCKING HATE AMD SO MUCH

Yes Brian we know.
Also stop defending TMG please, they need a firing squad for 10nm mess.

12 cores, 24 threads = as powerful as a 7700k

(lol

Meanwhile Intel pumps out an 8-core, 16-thread 9700k that will wipe out their entire Skylake-X platform

...

Pls don't be another FX 9590

delid dis

5.1GHz in a 12 core package seems like a fucking lot

Didn't IBM made a 10 core 5GHz pocessor with triple-threading last year?

Even with its current IPC, a 1.4ghz bump will fix most of its issues

Say his name Sup Forums. You know you want to.

Also, why isn't he working for the government to release IP-free CPUs?

The 9590 was ahead of its time. Just look at Intel's HEDT lineup today.

NOOOOOOOOOOO THIS CAN'T BE HAPPENINGGGG

If this turns out to be the case Intel is fucking done. Coffee lake can't really compete properly in overall value or quality against ryzen as it is its only marginally better performance wise plus you have all the Intel baggage like the TIM and other shit.

Intel is fucking finished. They have no answer to this stuff. Nvidia and AMD will be the future.

...

It would mind blowing of GloFo could actually pull this off. I can't even imagine what they'll be able to do once they get EUV implemented full swing. I actually thought this was impossible until late 2018 to early 2019.

If they managed this without increasing TDP at all, if they had a true 95w~ part with 12c at 5ghz, it would be world changing. The implications for enterprise with power per core being that low are astronomical.

I hope this drives the price of Skylake into the mud so I can upgrade to an i7 for cheap.

THANK YOU BASED MOMMY

What kind of software actually and properly uses 12 cores?

mining shitcoins

>mining on CPUs
What the fuck.

>>>/ltt/

Science shit like modeling and numerical shit in numerous fields, servers and crap obviously, VMs, rendering, image/data/whatever processing, and so on. What kind of shit doesn't use every core or processor you can throw at it? Games, I guess?

This is literally based on a fake slide made by Sup Forums

Run 3 desktops off one box?

>The new 12LP technology provides as much as a 15 percent improvement in circuit density and more than a 10 percent improvement in performance over 16/14nm FinFET solutions on the market today.

this is literally fake news. ryzen isn't breaking the 5ghz barrier until 7nm unless they go full netburst and completely disregard power efficiency.

Prove to me that they aren't just reporting based on an image that someone linked to them that they got from Sup Forums.

Depends on the price. Is it more expensive than current KIKERIPPER 1950X?

>he doesn't know

Anything that you don't want stuttering.

...

same am4 socket?

hmmm

running like 2 or more decent VMS? (4cores/4/8 threads each)

smooth gaming and streaming in 1080p. i have a ryzen 1700 and can't do it seamlessly. a 12 core would allow me to.

wait, i thought Zen+ was next, then Zen 2 was after?

Liza Su, if you'll do this, I will be ready to kiss the sand upon which you were walking.

This would be fucking awesome, but the zeppelin core complex is only 8 core. That would mean they'd either have to completely redesign it (unlikely) or add another zeppelin complex (not possible on AM4)

I wanna see this as much as the next guy, but this is a bunch of shit.

...

Will it run at such high frequencies on older boards? Crosshair 6 here and i hope it does

Motherboards have only limit on power delivery. Crosshair 6 Hero/Extreme has plenty of that. Otherwise its just bios update and slot in.

same slide of kitguru with no facts to backed it up

its fake we know since day one

...

Can one of you worthless fuck CPU makers start producing 2 to 4 core chips worth a fuck for about $20 instead of yet-another fucking super computer processor? Goddamnit man, there's a huge market out there for affordable computing.

>inb4 beaten by 8400 in all gaming benchmarks and emulation

So intel was retard for putting more cores, but amd is like fucking savior for doing the same thing?
Or did i misunderstood the threads on Sup Forums?

in the end, single core or multi core performance for what you are doing is all that matters
historically, with an Intel CPU with a specific clock rate, cache size, core amount, etc. compared with an AMD CPU with the same clock, cache, cores, etc., the Intel would outperform the AMD in that generation and for years/generations after
usually, core number and clock rate don't correlate well with benchmarks

>TweakTown saw edited image and bought it as real

Funny, I can game and stream on my 4790k just fine and it only has four cores and eight threads. And it's been running 5Ghz under 75c since the day I bought it almost four years ago, static, every single time I turn on my computer.

AMD doesn't have the efficiency per clock cycle or per core that Intel does. They compensate by having more cores and cranking clockspeeds. That's valid, to a point, but falls flat when Intel can do exactly the same thing and still maintain higher efficiency.

Cause the world needs to watch you play video games, right?

this guy comes up to you and slaps your delidded cpu on the ass

what do you do?

The actual material cost of something like a 4790k/6700k/7700k is about 25-35$. The rest of the cost is paying for the millions of man-hours of research that went into developing that product.

In a sense, they've already done exactly what you're asking for (and it's the reason why Intel has their Tuning Protection Plan and why you can get a replacement 5960x no questions asked for 35$ if you return your old one and it's not literally fucking scorched).

Yes, they charge too much for their products. But it's because nobody has really come close to having something competitive to what they have in a long time, and the level of semiconductor engineering and chip fabrication they've achieved is literally fucking magic even to people that understand that kind of shit.

...

Seal the IHS back down to the PCB with the same silicone sealant they use at the factory, send it in, and get a new one for 25$ (that'll probably clock higher at the same or lower voltage anyway).

I've done it four times with Intel CPU's. Once with a G3258, once with a 4690k, and twice with 4790k's. Each time they never asked me a single question and sent me a replacement within two weeks.

With the tools, techniques, and cooling compounds available nowadays I don't know why every single enthusiast doesn't do it. Liquid metal TIM is within 1-5c of what soldered TIM can do and doesn't damage the core of the chip through repeated heating and cooling cycles, something der8auer proved can happen even using far better and more exotic mixes of solder.

>4790k
>And it's been running 5Ghz
and that's how i know you are lieing, keep justifying yous stuttering housefire

Even if they did what architecture did it have?

Nice one
t. Owner of a 2012 laptop i5 and a ryzen 7

POWER9

The difference is how they are doing it.
AMD is doing nice and efficient modular design.
Intel on the other hand... pic related

>Goddamnit man, there's a huge market out there for affordable computing.
Not really since normies are using ARM devices now. Even if there was the market is not profitable.

>historically
You havent been around that long I see.
Also not true for ryzen.

anything that's not playing games

>You havent been around that long I see.
been on Sup Forums 11 years

Historically wheels were made from wood so they can't be used in automobiles. Cars cannot exist.

...

Pretty much, but a tweaked ryzen at 5.1GHz is going to have better single threaded performance than intel chips.

If these leaks are even remotely accurate, intel doesn't have a competitive processor. This is athlon64 all over again

>this is literally fake news
feel free to sauce me up, bitch

litography faggot

clock to clock ryzen has a better ipc we know that since day 1
problem is we wont see any sort of 5.1 ghz on a freaking LP node
what we knew couple of months ago as ryzen 2 (the x2 the cores bla bla 7nm) is now called matisse or mattisse or whatever
what we are going to get now as ryzen 2 is just a small refresh with a small bump on the speed mayyyyyybe 4.6-4-8ghz tops but nothing else

You caught me. On my current 4790k I only run it at 4.8Ghz/1.4v because it wasn't as good as my earlier ones. It's still a static overclock, it runs at that speed and voltage from the time I boot it up to the time I shut it down, and it stays under 75c during Handbrake encodes (a way to measure heat produced by the chip with a maximum real-world load) for 48 hours straight. Prime95 v26.6 will get it up to 82c on the hottest core after about four hours, but the design of the chip means that even with perfect thermal interface material application max coretemp will vary 3-5c between different cores since the iGPU acts as a heatsink when it's disabled.

It ranges from 71c-75c depending on whether you ask Core #0 or Core #3 during Handbrake, 78c-82c during Prime95, and typically runs in the high 50's to mid 60's while gaming.

But if you get lucky and get a 4790k that can do 5Ghz at or below 1.4v Vcore, if you delid it and apply Liquid Metal TIM on the core you can run it within safe temp parameters even on air cooling. My NH-D15 has served me well.

In benchmark runs when I've removed a stick of RAM for clearance, added a third fan, and placed a 3d-printed cowl over the heatsink and fans themselves that extends all the way to my exhaust fan at the top back of the case, I've gotten it to hit 5.2Ghz at 1.478v Vcore Handbrake and Prime95 stable while staying under 81c/88c respectively. However, the whitepapers for this specific processor state that degradation will occur over 1.4v Vcore and/or 77c coretemp.

For even more unnecessary referrence, a God-Tier 4790k would do 5Ghz at 1.265v Vcore. So, no, you're wrong. But I've spent literally thousands of dollars and hundreds of hours tinkering with processors to be able to emphatically and beyond a shadow of a doubt say that you're wrong, so... you're not that wrong.

This. Ryzen IPC is fine, it just clocks rather low. If you could make Ryzen run at 4.5 to 5, that'd be insane. I don't believe for a second there'll be a 12C at 5.1 though.

Not really, they just wouldn't be able to afford putting in shit TIM and pricing their products how they do now. Imagine if 8700K was soldered and because of that clocked slightly higher and ran cooler, it would be a great product. Right now that's their only real issue on the desktop market which supposedly could be fixed somewhat easily.

When you're running multiple programs.

fucking WHEN i see high variety of notebooks with mobile ryzen?

I'm not reading this

is 1.4v even safe for haswell? my 4.4 ghz 4690k is running at 1.28v right now.

this sounds like total bullshit, if 16 fucking threads aren't enough for you to stream and vidya at the same time, you're doing something wrong

Imagine if you bought an 1800X for 500 dollarydoos

black friday wise a 1800x was as low as 330
a 1700 was seein on 230
a 1600x on 180

i mean their yields are so damn high they can practicly give them for free and still make profit

>I don't know why every single enthusiast doesn't do it
Because we can get a soldered CPU that performs great for a lot less, and don't have to spend money on liquid metal, delidding kits, massive watercoolers or protection plans. We don't need to send our CPUs or wait for replacements.

>Liquid metal TIM is within 1-5c of what soldered TIM can do and doesn't damage the core of the chip through repeated heating and cooling cycles, something der8auer proved can happen even using far better and more exotic mixes of solder.
This is so stupid I'm going to have to leave out a few points. They proved that the solder will crack and reduce thermal conductivity when cycled between -55 °C to 125 °C and on small dies ONLY. As you should know CPUs normally cycle between 20-75c. Quite a bit of difference.

There's no evidence of small soldered CPUs failing faster than others in real life use. Unless you go full retard on voltage or delid(lol) CPUs almost always outlive both their usefulness and supporting components.

This clearly isnt the reason intel doesnt solder anymore. They clearly just want to save a buck, because they could still be soldering their giant housefire CPUs even if der8auer is right.

If you can keep it below 77c coretemp, yes. The 4690k's that I've overclocked and benched with didn't start seriously degrading until I pushed them into the 1.55v+ range, although that was with phase-change cooling or liquid helium.

Reducing the operating temperature of a semiconductor by 10c will double its operating lifespan at a given set of parameters. Past 1.4v electromigration can occur in Haswell CPU's regardless how cool you can keep them; past 77c coretemp electrical resistance can change enough to make electromigration happen regardless of voltage (although realistically speaking even at 100c it's not going to happen to an appreciable amount if you're running stock Vcore or lower but that risk increases exponentially the more voltage you add).

Electromigration is the process by which the electrons flowing through your CPU literally tunnel through the "walls" of the semiconductor gates and damages them over time which will eventually damage your CPU, either requiring a higher voltage to maintain previously stable clockspeed/reducing the clockspeed that your CPU can run at regardless of voltage applied/outright killing the CPU by making it unable to run at any appreciable clockspeed regardless of voltage or making it require an insane or impossible voltage to maintain a usable clockspeed.

You can push those limits a little bit, briefly, but doing so on a regular basis with even a small amount of even for a fraction of a second with a large amount can permanently damage or break your CPU.

What kind of cooling do you have and would you be willing to delid your chip? Is that overclock one you setup yourself? Can it maintain 4.4Ghz in benchmarks/games with lower than 1.28v? Taking a total shot in the dark if it takes all of 1.28v Vcore to maintain 4.4Ghz under max load conditions then you'd probably be lucky to get 4.7-4.8Ghz out of your chip (without base-clock speed tuning) at 1.4v and 1.4v requires some serious cooling even delidded.

I run 4.8Ghz/1.4v Vcore (1.415v input to overcome LLC) under an NH-D15. I delidded all of my CPU's besides my current 4790k with a razor blade and a steady hand, and this one I used a 25$ 3d-printed tool that I had made at the local highschool for 33$ to cover material cost and labor. The total cost of all of this including the protection plan was 173$ and my time. The NH-D15 was 90$, the protection plan was 25$, the Coolaboratory Liquid Ultra was 20$, the delid tool was 33$, and lets say 5$ for razor blades. In exchange I got performance that couldn't be matched with any stock component until the release of the 7xxx series at safe temperatures with minimal noise and no noticeable degradation and (what's admittedly my guess) minimally reduced lifespan if any at all compared to the claimed lifespan of 10 years at stock settings.

On top of that, I got my 4790k for 250$ from Microcenter. Compared to the launch MSRP of 350$, spending 423$ total on delidding and properly cooling one 4790k to get a level of performance that couldn't be matched with any off-the-shelf component for literally years is a fucking steal. That's not taking into account that I reused the same tube of CLU TIM, the same NH-D15, and the same razor blades across four separate processors.

Yes, it was more expensive. But not really all that much more. Yeah, it sucks having to wait two weeks to get a replacement CPU when I've killed them, but it sucks a whole lot fucking less than having to pay 100% of the cost of a new one for a replacement compared to 10%.

The cracking of the solder isn't the issue. The damage it does to the die (ANY die size small or large although, yes, smaller dies were less affected) when it cracks (and it will still crack only cycling between ambient and max safe load temp with enough heating and cooling cycles) is the issue. It's pulling on the surface of the die. Over time that will cause micro-fractures.

1/2

>Coffee lake can't really compete properly in overall value
Do you have any idea what Intel's profit margins are?

>The 9590 was ahead of its time
Are you just joking, or are you genuinely an retard?

2/2

Are all these concerns very niche and do they only really affect overclockers at the fairly extreme end of the spectrum? Yes. But objectively speaking it still makes more sense to use a conventional TIM application between the die and IHS if that risk is present at all and you're marketing a product specifically to be run out-of-spec. They carried it over to the rest of their product line to cut costs and only soldered the larger-die chips and server chips because those typically aren't overclocked and if they are overclocked no sane individual would do so without getting what basically amounts to a 25-35$ insurance policy on a 700-1600$ processor. Furthermore, there is evidence of cracked dies on failed and failing 5960x and other large-died processors that were overclocked or otherwise subjected to a high number of restart cycles or higher-than-expected temperature fluctuations.

Solder with a melting point of 150c+ cannot be expected to not crack as it begins to partially melt and soften when exposed to even just half of that temperature for extended periods of time or multiple times. Using a liquid metal TIM makes a fuckload more sense since it goes from a solid to a liquid at ambient temp, meaning it will always be a liquid during operation and will never pose a threat of cracking the die, ESPECIALLY if you're planning on overclocking.

And the regular fucking TIM application was more than good enough for everyone that didn't ever overclock at all or didn't overclock at a serious level anyway. With an NH-D15 you could still push a non-delidded 4790k to 1.3v+ and keep it under 80c, under 77c coretemp if you kicked the fans up to max. If you wanted to get that extra tenth of a volt and extra 200-400Mhz operating speed you had to tinker with it.

Well no shit sherlock, if you start slapping turbos and NOS kits on a stock crate engine past a certain point you're gonna have to replace the pistons and other components with forged internals.

This is why I'm not getting my hopes up yet. Next year seems like a good year for consumers, so I am glad that I waited for a new build. After benchmarks, and prices all that's left for me to wait on is what the situation is like with TEE. Hopefully there is no need to worry about PSP, TrustZone, and ME next year.

Well, if he's talking about the 8700k and i9's then it's partially right, both need a water loop just to not burn down your entire house like the 9590.

Why do people make posts like these?
They're not funny and provide no contributions to the discussions.

>12C
>5.1ghz
even amd said that ryzen 2000 will still have 8 cores
and every possible source said that somewhere around 4.2-4.5GHz will be the top frequency

Because literal shills. It's not just a meme, with the amount of shilling I see on Sup Forums it wouldn't surprise me if they were actually getting paid for it.

you complain about these posts having no contribution to the discussion yet you make a post which also has no contribution to the discussion. Shut up faggot

HAHAHAHA

you just did the same thing

you can't read at all

>NOOOOOOOO!!! STOP BULLYING INTEL!!!

>you can't read at all
You're just dumb. Please go the fuck away fagget.

are you implying i'm the person who wrote the first post? because i'm not and i will explain to you what that person meant
>9590 was ahead of its time
>Just look at Intel's HEDT lineup today.
he means that intel's HEDT cpus are just overheating pieces of garbage, just like the 9590 was 3 years ago
you're the dumb person, and you're the faggot here
you can fuck off now.

The i9's, and 8700k brought an increase in performance along with them, and the fx9590 single threaded worse than a core 2 duo t9900.
user was just being an idiot, so please don't encourage him.

>are you implying i'm the person who wrote the first post? because i'm not and i will explain to you what that person meant
I don't care if you were that idiot, or not, but people like you need to leave Sup Forums.