[ 3 / biz / cgl / ck / diy / fa / g / ic / jp / lit / sci / tg / vr / vt ] [ index / top / reports / report a bug ] [ 4plebs / archived.moe / rbt ]

Due to resource constraints, /g/ and /tg/ will no longer be archived or available. Other archivers continue to archive these boards.Become a Patron!

/g/ - Technology

View post   

[ Toggle deleted replies ]
File: 489 KB, 1071x686, Screenshot_20181204-222650_YouTube.png [View same] [iqdb] [saucenao] [google] [report]
68787650 No.68787650 [Reply] [Original] [archived.moe] [rbt]

Is this real? Is Intel truly BTFO?

>> No.68787863

Boost clocks are probably on 1-2 cores.
But it's possible.
4.3ghz 12nn boost to 4.8ghz 7nm or so.
We already know 5.0ghz on 8+ cores is cause to call the fire department.

>> No.68787892
File: 7 KB, 189x267, BrianJUST.jpg [View same] [iqdb] [saucenao] [google] [report]

fake news

>> No.68787906

This. Those speeds are on ln2

>> No.68788728

Kinda below expectations, I think it's fake since the process should reach 5ghz base at low tdp, so boost should be 5.1-6 as we would be talking about single-core boost.

>> No.68788864

Ryzen + only hits 4.1-4.3ghz single core.
7nm won't allow a 5.6ghz boost man.
Tsmc is claiming a 20% drop in power usage over 16nm not a 20% frequency increase.

>> No.68789605

That was for the IBM/GloFo process which was axed because AMD went with TSMC for 7nm.

>> No.68790271

i can confirm this to be real. my dad works at nintendo

>> No.68790302

too much cores

>> No.68790330

Don't worry Intel bros, Keller will make intel a shiny new architecture. Just wait until 2022.

>> No.68790872


>> No.68790933 [DELETED] 
File: 42 KB, 660x450, intelpajeets.jpg [View same] [iqdb] [saucenao] [google] [report]


>> No.68790956

It's not about TDP it's about silicon quality and binning on lower end CPUs

>> No.68790970

4,2 on every
Somewhere above 5, perhaps not 5,6 but 5
they did advertise it as a node for 5Ghz devices

>> No.68790980 [DELETED] 
File: 337 KB, 1500x1164, 1501867726886.jpg [View same] [iqdb] [saucenao] [google] [report]

goy ples delid dis

>> No.68791000

>below expectation
>he expects 5.6 Ghz Ryzen CPU
Are you retarded?

>> No.68791186

4.2Ghz all 16 cores/threads
5Ghz single (more likely dual like Intel) core/thread(s)
Can probably overcock on all cores to 4.5Ghz but letting XFR do its thing would probably be best.

>> No.68792100

If those "leaks"(Rumors) are real dam that they look good.
Looks like its time to replace my I7-4790K.

>> No.68792203

Welp time to sell my 4770K before it's worthless in 2 months

>> No.68792260

Excusing me sir
But possible to delete and not spread this rumor sir
Can be hurtful and untrue which is bad for intel that has excellent coffee lake r CPU for all everyone
Truly regards

>> No.68792267

it's ok. Intel will release 9990K, with 6GHz turbo boost. It'll be delidded, delapped and comes with a chiller out of the box

>> No.68792301

Can any current AM4 board even properly support these 12 core beasts? Isn't' dual channel memory a bit inadequate for that too?

>> No.68792344

New boards will come out that are 6+2 instead 4+2, problem solved.
And im not talking about the fake Gigabyte or ASRock motherboards.

>> No.68792363

Unironically gigabyte release one of the best am4 mobos

>> No.68792389 [DELETED] 
File: 138 KB, 748x900, 1542085404910.jpg [View same] [iqdb] [saucenao] [google] [report]





>> No.68792409

Only their highest end Motherboard is good, And one of the best(if not the best) X470 motherboards, the rest are shit.
MSI makes the best B450-X470 motherboards but they come with a huge failure, lack of offset voltage (Which kills PBO).

>> No.68792634

>mfw we will see 5ghz+ Ryzen ocs out of the box
Oh fuck my dicc

>> No.68792737 [DELETED] 

Please delet this antisemitic video.

>> No.68792751

the "reddit" leak is completely fake, the fact that he sources it discredits any of this information.

>> No.68792753

If only they could pull such miracles in their GPU lines too.

>> No.68792759

if true, it's a 3600X or 3600 for me

>> No.68792777
File: 142 KB, 700x881, 1543200069122.jpg [View same] [iqdb] [saucenao] [google] [report]

christ I don't want a housefire.

>> No.68792851

Yeah the gaymen 7 methinks

>> No.68792864

Sandy i7 2600K was roughly 125W.

>> No.68792866

People don't just go to the internet and tell lies my man

>> No.68792929

Where do you absolute retards get your information from?

>> No.68792938

No they fucking didn't, kid. TSMC is not Global Foundries.

Yes. He is.

>> No.68792942

If Intel weren’t so dishonest about their tdp, the i9 9900k would be around 140-150w. 105w is nothing

>> No.68792962

this would absolutely destroy the market. holy shit if true. b450 works for ryzen 3xxx right?

>> No.68792985

as far as I know, b450 support for zen2 will be for everything up to the 12c/24t parts, so yes and no

>> No.68792996

*up to and including

>> No.68793114

Yes but not all, some shitty B450 motherboards have problems with the 2700 because the shitty VRM and lack of proper cooling (See all Gigabyte, ASRock, MSI PRO-A or VHD, literally any motherboard with no VRM heatsink).
Such problems would be worse with the new CPUs.

Good motherboards like the MSI B450 Tomahawk shouldn't have problems but dont expect a R7 3700 working on them properly.

>> No.68793146

>still slower than 9900K in games

>> No.68793152 [DELETED] 
File: 339 KB, 682x792, 1515221586890.png [View same] [iqdb] [saucenao] [google] [report]

How could this happen to me
I made my mistakes
I’ve got no where to run
The night goes on
As I’m fading away
I’m sick of this life
I just wanna scream
How could this happen to me

>> No.68793153
File: 171 KB, 1443x123, ryzen 9.png [View same] [iqdb] [saucenao] [google] [report]


>> No.68793158

Honestly why is anyone surprised by this? It's perfectly reasonable frequency increase given the node shrink and adding more cores is trivial with zen architecture.

>> No.68793170

I'd bet my bottom dollar when the 3000 series comes out, will it blow the intards away hard.

>> No.68793181

It's almost certainly wrong

I expect ryzen 2 to top out at 4.5ghz

intel always tries to gaslight amd with fake leaks that make amd seem like it's going to be a monster and then when the amd products actually come out they're much worse than the inflated expectations people had and everyone gets mad and blames amd for falsely advertising and hyping themselves up too much

>> No.68793195

nigger zen+ can already do 4.1-4.2

You really think 300mhz or 7% is all they could squeeze from 7nm? That's literally worst case scenario

>> No.68793209


>it's almost certainly wrong

Based on the source that's real in your mind. AdoredTV is well connected, as is Kyle Bennett from HardOCP who has corroborated most of Adored's info.

>> No.68793216

Zen 2 will be the new 2500k

Literally buy it for 8+ years

>> No.68793243

And the refresh would be the I7-4790k.

>> No.68793244

I don't think so.
We are going to see a core explosion.
I'm talking about 128+ cores in 4 years.

>> No.68793267

Why cant we have weaker cores in CPUs yet?
Like 8 5GHz cores and 4-8 2GHz cores like ARM does.

>> No.68793306


>> No.68793320

That might be coming.
We have other stuff on the horizon too like persistent memory which we've only barely implemented.
Our lost decade of technological gains is about to be kicked into overdrive. Moores law won't be aggressive enough to describe what's ahead.

>> No.68793339

200watts all cores at 5ghz oc. What do I do with all this cpu power? Stream games at 1080p60 with x264? Time to learn to learn making blender porn I guess. Can AM4 support more than 2 memory channels? You can have all those cores but not enough memory for multiple vms. What would you use 16 5ghz cores for?

>> No.68793413

Its fake. And gay. Some rando on the internet is not a credible source. Meanwhile intel is shitting up ryzen with their new i9

>> No.68793426


$500 6-core/12 thread, [email protected]+ computers?

Right now $500 gets you a 4 core Ryzen 3 2200G and a RX 570 4GB

>> No.68793476

If only AMD was doing as good in their GPU line as their CPU line.

>> No.68793482

Thread theme if true: https://youtu.be/j0lN0w5HVT8?t=37

>> No.68793487


are the toilet and that celeron real?

>> No.68793496

Yes and no.
That HEDT I3 was close though.

>> No.68793500

Learn how a power curve works, pal.
7nm Zen2 might clock 40% better than 14/12nm Zen1 at the low end of its curve, but have power consumption drastically ramp up at its high end making it unfeasible to hit those clocks at a marketable TDP. Every power curve is different, every power curve has different critical points.

Its astounding how people here don't seem to bother reading and learning fucking anything about this topic at all.

>> No.68793538

something something uncle

>> No.68793551
File: 204 KB, 600x354, arm-cpu-article-2013-4.png [View same] [iqdb] [saucenao] [google] [report]

ARM's big.LITTLE is designed to save battery in a smartphone. Its not advantageous to utilize anywhere else. ARM does this because their big cores can't clock low enough with low enough power consumption to match their little cores. They're in entirely different domains of power consumption.
They can create a little core pulling 150mw that runs all the background processes in Android where as their high performance cores can't scale down low enough to compare.

This paradigm only exists because smartphones have incredibly limited battery capacities, and phones have shitloads of background processes running constantly. Its a total waste of time to implement this for anything else. The complexity of the fabric needed to switch threads between the cores on the fly is ridiculous.

>> No.68793576

>His proof is that someone posted on reddit several days ago, then days after that he got an email with similar info and therefore the 2 corroborate each other as proof

Is this guy a total fucking mong? Yeah journalism pro right here, all you need to do is shit post on reddit then wait a week and email him similar info and he makes a video on it.

>> No.68793606

>Not 2600K or 2700K


>> No.68793611

>Stream games at 1080p60 with x264?

You can already do that with Ryzen 1700

>> No.68793625

Or any IGPU that has hardware encoding.

>> No.68793639

That was the explanation years ago when they developed the technology.
My point is that we are going to see a core explosion, in which case high parallelization may see benefits of low power cores and high power cores to maintain an overall CPU power budget. These are things that weren't fully addressed when previous attempts were made with multiple core technologies, like the Cell processor, and Kilocore.

>> No.68793647

that wouldn't be x264 then

>> No.68793650

>AdoredTV is well connected
Did you even watch the video? He has zero connections, he saw a week old post on reddit and its numbers were similar to, but not the same as, an email he got from a nobody several days later and that's his source.

You could go post tomorrow on reddit saying "Ryzen 7 3950, 5.5ghz all cores, $129.99" then wait 4 days and email him "Ryzen 7 3960, 5.4ghz all cores, $139.99" and he'll make this exact video, because that's all that happened in this case.

>> No.68793656

Not enough ice in Antarctica

>> No.68793672

Adored is literally reposting forum rumors from overclock.net, WCCFtech, S|A, and Anandtech.
He isn't connected at all. He has no sources.

>> No.68793681

They're doing good in their CPU line because they've a̶b̶a̶n̶d̶o̶n̶e̶d̶ shelved their GPU line.

>> No.68793682

is that the predator on the left

>> No.68793693
File: 131 KB, 400x208, 1537876744255.gif [View same] [iqdb] [saucenao] [google] [report]

Ryzen 7 3700X
RX 3800 = RTX 2080 performance for half the price
I'm ready

>> No.68793770

Thermal budget isn't an issue when working with a high TDP part, nor is any modern X86 core starved for performance because the OS is scheduling too many background processes on it.
A core can handle low intensity background threads and stay at 1-2% utilization. The big.Little concept has nothing to offer in a desktop, or even a laptop, unless you want to artificially restrict your laptop with a laughably small battery just to fuck over your consumers. Even then you'd be better off spending the silicon on some hardware accelerator to save cycles in a common workload like we did with video decoders.

The concept of big.LITTLE wasn't just some inane excuse ARM made up. Its the sole reason why it exists, and it will always hold true.

>> No.68793779

Don't get your hopes up about Navi prices

>> No.68793792
File: 157 KB, 565x541, 1526047306344.png [View same] [iqdb] [saucenao] [google] [report]

Don't get your hopes up for Nvidia prices

>> No.68793815

Consider getting a Vega 64 now, they're usually on sale for $400. They're overstocked and trying to clear inventory.

>> No.68793847

Ok that may be the case, but big.LITTLE has some other characteristics, mainly the dissimilarities between core types that I think we will see - for one reason or another - in CPUs (not that we haven't seen something like that with APU or grafted on GPUs integrated in CPU silicon)

>> No.68793858

RX 3080 is competing with 1080/2070. We won't see AMD cards competing with 2080/2080 Ti until late 2019/2020.

>> No.68793891

who fucking care it just $250

>> No.68793898

Considering they couldn't make a dent in Nvidia's sales when "ATI" products were superior in almost every objective measure, they'd only send themselves broke trying to out RnD team green. Seriously, who the fuck was buying Tesla when based HD4000 series existed?

>> No.68793902

It's not going to be $250 fren

>> No.68793909

he doesnt source it, he says the reddit info is wrong in the vid...

>> No.68793915

Having big doubts about their GPUs.

>> No.68793918
File: 81 KB, 1791x582, Screenshot (32).png [View same] [iqdb] [saucenao] [google] [report]

>> No.68794058


>> No.68794079

It'll be both. 2600/2700k like performance leap, 2500k price.

>> No.68794080


Suspiciously /g/ suddenly has declared Kyle a nobody who knows absolutely nothing in the last thread.

>> No.68794151

Novidiots still haven't forgiven him for debunking the GPP

>> No.68794207

Nvidia will never cede an inch to AMD, especially now that they're seeing Intel scram up after dominating for a decade.

Intel will eventually get their shit together and they'll make sure to bury AMD like they did to Cyrix.

>> No.68794218

what the fuck fuck you kyle you lying sack of shit

>> No.68794242

Intelaviv is finished and bankrupt

>> No.68794251

AMD confirmed dead and bankrupt this time next year! For real this time eh?

>> No.68794392

good luck but im snagging some april 2019 amd calls

>> No.68794450

Why do I feel bloodlust from AMD. Do they want to destroy Intel that much?

>> No.68794468

>implying Intel will be destroyed
>not realizing how duopolies work

>> No.68794484

That was my. I have bad memory.

>> No.68794503

why wouldn't they? more money for them

>> No.68794515
File: 66 KB, 568x612, 1513978445505.jpg [View same] [iqdb] [saucenao] [google] [report]

>we are at an age where amd is possibility making enthusiast class cpus

Seriously though why don't they bin some 8 cores at like 5ghz and undercut Intel instead of doing this retarded 200 core 3.9ghz bullshit (but 5.1ghz on 1 core for .005ms)

>> No.68794522

It will take years before good will and momentum for Intel runs out even if AMD releases CPUs that are twice as powerful at half the price. They'll be relatively fine until they release a good working 10nm even if it takes them 3 more years to do it. It's pretty sad

>> No.68794523

You know it's real when the shills are scrambling to bury this.

>> No.68794544

in all realness though intel really is shitting the bed lately and if they dont get it together theyll be in trouble soon

-still dealing with security flaws
-not meeting demand on processors
-9700k is essentially a soldered 8700k with 2 more cores
-10nm is 2+ yrs late

>> No.68794556

Boost clocks are here to stay. On >4 core CPUs. It doesn't make any sense from a power or binning perspective to clock every core the same. They do the job really well keeping single thread performance high and peak power draw low.

>> No.68794575

No one is burying this. You guys have posted like 5 similar threads throughout the day.

>> No.68794609

If true, I'll fap until I pass out.

>> No.68794623
File: 501 KB, 1024x572, amd-chiplets.png [View same] [iqdb] [saucenao] [google] [report]

Oh, they will for Zen 2+ in late 2019/early 2020. They'll be able to pull it off because the next gen consoles will be using the exact same chiplets and with the economies of scale of the console market, there'll be a lot more silicon to go into the higher grade bins. The consoles are spec'ed out to be 8 cores, so they're going to pair up zen 2+ chiplets with as many as 4 defective cores each and one navi chiplet per package. The interposer will be the same one used for Epyc/Threadripper, with three different memory controllers on the same die to cover all of the HBM2/GDDR6/DDR5 skus. Almost all of the junk silicon can be recovered.

It's fucking genius I tell you. They've been planning for this since 2011/2012 as they started to ramp up for Xbox One and PS4 and realized if they could use the same silicon for everything, they'd be making bank and out-competing Intel at yields.

>> No.68794629

>It doesn't make any sense from a power or binning perspective to clock every core the same

If your silicon is capable of hitting 5ghz on all cores like Intel there's simply no reason not to. Multipliers exist for a reason.

>> No.68794656
File: 63 KB, 705x602, tapestry,940x-bg,f8f8f8-c,110,110,705,602.2.jpg [View same] [iqdb] [saucenao] [google] [report]

>amd naming their shit chiplets

like pottery

>> No.68794670

The Virgin Monolithic Die vs The Chadlets

>> No.68794778

They are anti semantic, I tell you what.

>> No.68794791

if it looks too good to be true...

...it's a lie from some intel jew

>> No.68794846
File: 77 KB, 320x320, 725.jpg [View same] [iqdb] [saucenao] [google] [report]

t-this can't be happening bros. it's all going wrong

>> No.68794886
File: 12 KB, 408x431, 1543984839956.jpg [View same] [iqdb] [saucenao] [google] [report]

>Intel BAD! AMD Good!

>> No.68794896

Fake as fuck, same leak that show 64/128 threadrippper at 5ghz stock. they remove it because to obvious

>> No.68794917

Please remind me about fiasco with core i9.
I remember it was something to do with huge electricity consumption or overheating
Something about 1k watt, was it consumption or tdp?

>> No.68794927

Your memory is so bad you should probably stop breathing

>> No.68794948

seething tel aviv cuck

>> No.68794949

It's the 28 core 5ghz chip you're thinking about

>> No.68794958

Intel has had shitty business practices that have set back amd like craaazy. Not to mention Intel stagnated the market at quad core cpus for the longest time.

Amd is doing the needful and taking them down a few notches

>> No.68794962
File: 397 KB, 475x481, 1495378144247.png [View same] [iqdb] [saucenao] [google] [report]

>12 cores
>5ghz turbo
>at 7nm
My body is ready

>> No.68794987

Dat tdp tho

>> No.68795050

why did you cut off the price?
that's the killer.

>> No.68795074
File: 1.11 MB, 1684x1080, shoah.png [View same] [iqdb] [saucenao] [google] [report]


>> No.68795205


>> No.68795217

He says right there in the comments he got the info BEFORE the Reddit post.

>> No.68795240

>'Just an FYI for those wondering if the Reddit leak was before my leak - no it wasn't. I got all this information days ago, it just takes a long time to check other sources and obviously make the video. ;)'
Fuck off kike.

>> No.68795243
File: 37 KB, 263x269, 1490112447202.jpg [View same] [iqdb] [saucenao] [google] [report]

>16 cores @ 5.1ghz
>for $500
Oh god I'm cumming everywhere, pls be true

>> No.68795292

Wonder what the IGPU performance would be like. Though it'd still be severely limited by the DDR4.

>> No.68795297
File: 68 KB, 912x554, encodes in vp9.png [View same] [iqdb] [saucenao] [google] [report]

>even worse for av1
Can I finally transcode to vp9 or av1 in real time?

>> No.68795307

2400g is already kinda bandwidth limited. Integrated graphics won't be great until they put HBM on the package or something.

>> No.68795310

>14nm++++++ speeds apply to 7nm

>> No.68795316

how is amd gonna support those 16c in the older gen mobos with less capable vrm's?
doesnt seem possible at all

>> No.68795341

how did they support the 2990wx on the less capable x399 boards

>> No.68795350
File: 1.05 MB, 1196x676, 1535051752782.png [View same] [iqdb] [saucenao] [google] [report]

These aren't even the Threadrippers. Does anyone have that list from a few days ago that includes absolutely everything?

>> No.68795362

tr mobos well most of them have quite the power monster vrm's to feed them so its logical
but the consumer ones arent that good

>> No.68795397

kek. they can't believe that they're getting BTFO'd this hard

>> No.68795399

it's a smaller jump in tdp than the 2990wx
maybe they had enough room in the spec
maybe they will throttle on poverty boards
maybe they will be x570 only

>> No.68795414

they will obviously throttle no questions about it
and i cant even imagine when they OC them lol

>> No.68795433

With their current parts they're limited by process, not temperature or power.
I have no fucking clue. Forced underclock?

>> No.68795456

Not just tdp reqs, mobo manufacturers were bitching they don't have enough space on uefi firmware to fit new CPUs and dropped Bristol ridge.

It's not at all clear who will support what for 3rd gen, absolute poverty boards might skip entire 3rd gen, mid level boards might support upto r7 and high level boards might go up to r9 but at run it at base freqs. X570 would be the recommended way to get full benefits at any case

>> No.68795457

How much does ryzen blow past TDP? 135 watts is in the range of an overclocked 8 core and fine for most x370 or b450 boards.

>> No.68795462

amd went from begging mobo factories to make quality boards
to mobo factories begging amd to stop making quality cpus

>> No.68795464
File: 152 KB, 746x953, Trump approves.jpg [View same] [iqdb] [saucenao] [google] [report]

>mfw intel will lose the MUH GAYMES argument that they're holding on to so desperately
Their products really are going to be dead in the water next year, aren't they? I honestly suspect there's going to be price hikes on the ryzen 3000 series, not only because amd will be clearly in the lead (every market leader ever charges a premium) but there's surely going to be supply problems when literally nobody has a reason to buy intel.

>> No.68795477

The TDP for zen2 isn't that far off from zen+ tho
Hell, maybe AMD had a vague idea how much power zen2 would draw and required board manufacturers to adapt even before zen1 came

>> No.68795481

took intel 8 years to finally kill amd even with their shady shit
and it will take amd 2 years to finally kill intel fair and square

>> No.68795497
File: 633 B, 245x19, vp9 encoding.png [View same] [iqdb] [saucenao] [google] [report]

It doesn't matter. I was poking fun at modern webm codecs that aren't well multithreaded to take advantage of cpus with 8+ cores. They can't even encode efficiently on one single core.

>> No.68795514
File: 83 KB, 653x726, 1526519088130.jpg [View same] [iqdb] [saucenao] [google] [report]

>chiplet cores with even more lag than ryzen ccx's
>memory controller off die, adding even more latency
>up to twelve cores on only dual channel memory (LMAO)

oh sure but this piece of shit glued together grade school diorama of a cpu will totally btfo intel for real this time, honest, just you wait (tm). keep dreaming streetshitters.

>> No.68795527
File: 70 KB, 241x241, 1542724578389.png [View same] [iqdb] [saucenao] [google] [report]

Ryzen has always oc'd at least to single core boost, usually going a bit further. Idk why people don't think this will be the same. base clock of 5.0-5.1ghz all cores is a conservative estimate for what we can get out of overclocking and realistically 5.2-5.3.

Intel is finished.

>> No.68795537
File: 69 KB, 1024x1023, 1543791157003.jpg [View same] [iqdb] [saucenao] [google] [report]

>Buy $99 6c/12t cpu
>50w TDP
>OC to 5ghz
>btfo $400 8700k with an entry level aftermarket cooler

the absolute state of intel

>> No.68795545
File: 113 KB, 446x636, 1468987706036.jpg [View same] [iqdb] [saucenao] [google] [report]

>every market leader ever charges a premium
Market leaders have to be very established as 'da bes' to get away with this, generally speaking. We've also seen people choose the inferior competition over amd's products before (fermi) so I'm not too worried about supply. There's still plenty of retards to buy whatever they fuck intel decide to shit out next year, if anything.
Anyway, we're getting way ahead of ourselves. All this shit is technically still a rumour. If the announcements at CES next month prove these leaks to be true, that's when we'll know that the 3000 series is a game changer. I hope it is, it'd be a very much needed shakeup and I would like to buy an eternal cpu to last me the lion's share of a decade.

>> No.68795555

who's this? I don't frequent any other forum besides 4chan. Help this boomer out lads.

>> No.68795600

> be expensive i9 9900K
> get reduced to Ryzen 5 levels of performance

>> No.68795605

you should know if you're a boomer

>> No.68795635

i'm a boomer from yuropistan unfortunately. I only ever watched ncix, tigerdirect and carey (((holzman))). I'd guess he's some kind of tech show host as well.

>> No.68795678

> 4.5
The 2950X already does 4.4GHz m8, and that's on the 12nm.

Also, remember the clock bump from 14nm to 12nm?
> Ryzen 1: 4.1Ghz
> Ryzen+: 4.4GHz

Since the shrink from 12nm to 7nm will be much bigger than 14nm to 12nm, plus architecture improvements, I assume 5GHz is perfectly possible. 4.7GHz is the worst case scenario in my opinion.

>> No.68795702

AMD is using TSMC 7nm HPC process, not it's LPP process.

14nm was on the LPP process, thus the upper clock limits being hit.

>> No.68795709

big.LITTLE would make a lot of sense on laptops tho....

>> No.68795711

also, HPC vs LPP means massively higher clock ceiling and overall lower power usage.

>> No.68795734

Are those terms standardized? I remember lp meant low power for one manufacturer and ""leading performance"""" to another

>> No.68795746

He only sites them as references, but as far as rumours go, when there are multiple sites reporting on the same rumour, he finds it suspicious.So if SemiAccurate posts a rumour and then Videocards posts that same rumour independently, that sounds awfully strange. Also, Swedish Haggis man reported chiplets when everyone was arguing against it and proven right. Even if CanardPC did it a full year before he did. So he is connected to a degree.

>> No.68795780
File: 162 KB, 555x490, 1520396382702.jpg [View same] [iqdb] [saucenao] [google] [report]

this is too much power guys

>> No.68795797

>Ryzen 5 3600G 8C / 16T / 20CU
What kind of graphic performance can we expect from this if these numbers are real? Would it double the performance of the Vega 8 that comes with 2200G now, so around a GT 1050 Ti? Or is DDR4 RAM too much of a bottleneck for integrated graphics?

>> No.68795799

>6 cores for Ryzen 3
>Intel eternally BTFO
How will they recover guys and gals?

>> No.68795825
File: 93 KB, 575x1024, 1486101563459.jpg [View same] [iqdb] [saucenao] [google] [report]

jesus, what the fuck are the threadrippers going to be

>> No.68795844
File: 90 KB, 321x228, IAmTheEggman.png [View same] [iqdb] [saucenao] [google] [report]

>Athlon 2c/4t Zen APUs recently introduced for $55
>6c/12t about to bump quad-cores out of the $100 entry level
>AMD is eventually going to give us Kaby Lake i7-tier APUs with better graphics for like $60

>> No.68796024

>lack of offset
is that like LLC? or what do you mean?

>> No.68796029

I used dual channel DDR4 with a 12/24 Xeon for months and the performance difference between quad-channel was negligible.

>> No.68796038

>didn't watch the video
if only we didn't have you in this thread to show us the way
clearly all the people in this thread were blind and needed your insightful comments

>> No.68796062


>> No.68796082

Fak and gey.
If it's true though I'll throw my current pc out the fucking window and buy their stock.

>> No.68796091

u r so smrt

>> No.68796102

Without the memory restriction 20CU would be the same performance as GTX 1050 (not Ti).
But DDR4 is going to make it significantly slower.

>> No.68796143

>CPU performance doubling every 2 years
Stop it. This isn't the 90s anymore

>> No.68796150
File: 246 KB, 427x347, 2016-12-11 15_05_04-Facebook.png [View same] [iqdb] [saucenao] [google] [report]

Will these be able to handle 128GB of RAM?

>> No.68796154


>> No.68796164
File: 137 KB, 1280x720, 1529499860077.jpg [View same] [iqdb] [saucenao] [google] [report]


>> No.68796171

>this bait

>> No.68796177
File: 36 KB, 480x300, 1528516809826.jpg [View same] [iqdb] [saucenao] [google] [report]


>> No.68796181 [DELETED] 
File: 995 KB, 1543x1615, the cute menace.jpg [View same] [iqdb] [saucenao] [google] [report]

Double the cores for double the fun and even double the competition can't compete.

>> No.68796250

If we go by this video speculations.
20 Navi CU is around 1/2 of Navi 12, which is speculated to be around RX580. Soo... 1050(ti?), which sounds pretty damn good.

>> No.68796307

3600G for me if true

>> No.68796326

>peasant with no graphics card

>> No.68796333

I want Amada back

>> No.68796375

>Buy stuff you don't need, goy
We're not all gaymurz

>> No.68796392

Thank you based AMD for making CPU releases interesting again. I look forward to see if this is true or not, and what meme will come out of Intel trying to respond to it.

>> No.68796394


>> No.68796549

3300G for me desu. 65W is ideal.

>> No.68796625

The "muh gaemes" argument is already a fucking meme. Only delusional faggots believe it. On a 50 game average, intel is only 1.25 fps ahead of ryzen at sub-similar price bracket.

>> No.68796651

Unlike intelaviv's (((igpu))), even a 50 bucks 200ge can give better performance in certain gaemes than a 1030, with the proper setup.

>> No.68796656

Even gen1 Zen can already do it.

>> No.68796679

>What kind of graphic performance can we expect
>20 Navi CUs (Navi can do 40 tops in discrete)
RX 590 on a chip, essentially.

>> No.68796693

source? I'm itching to go back to amd and zen 2 is what I've been waiting for, but I was under the impression that in more cpu bound scenarios (like high refresh rate) that intel still has a significant lead, especially R7 vs i7

>> No.68796698

>intel will lose the MUH GAYMES argument
They already did when 2700X came out, since it literally matches (+/- 2%, a margin of error) 8700K in gaymen and utterly OBLITERATING it in serious productivity/multitasking WHILE being much cheaper and way more efficient at the same time.

>> No.68796775

Trips of truth

>> No.68796798

It's actually worse than that (for Intel). Since all their performance gains were at the cost of security (spectre, meltdown, and the 10 other CPU bugs/exploits most of which are Intel exclusive), if you were to patch the CPU bugs you'd get less performance than AMD. AMD lost only 2-4% performance with security patches, Intel lost 6-16% performance.

>> No.68796829

This just makes no fucking sense. How can AMD charge so little for so much? Don't they even care about their profit margins? I don't fucking understand.

>> No.68796840

Majority of games are still optimised for intel. So if you watch only one or two benchmarks of the same games that were made at day 1, you won't get the whole picture. Some anons posted links from anandtech(or arstechnica, I forgot). Some intelkikes also posted their own benchmarks on the same thread. Someone made an average of all of those7 lists and found that from a 50 game average, intel would only give 1.25 more fps on average. If you want to find out for yourself, you can search for whatever benchmarks you want with focus on the quantity and variety of games, devs, etc. Then make an average of it yourself. Don't trust only one source. Be it from an intel shill or amd fangay.
I'm sure intelkikes will make botched benchmarks of these new chips at week 1 and repost it everywhere.

>> No.68796842

>AMD lost only 2-4% performance with security patches
Barely ~1.5%
>Intel lost
At least 15% in best case, up to 57% loss in worst (enterprise/MySQL)

>> No.68796858

>This just makes no fucking sense
It actually does, because
>How can AMD charge so little for so much
Over 80% of the PERFECT bins in the yields ALREADY, with 12nm Zen+. TSMC's Zen 2 7nm has roughly 89~90% right now...of PERFECT ones.

>> No.68796887

I think the reality is that they have to be very aggressive with their prices because they're not perceived by the common buyer as being a premium brand. Also presumably the chiplets allow them to be so competitive with their prices. As >>68795545 has pointed out, they're not in a position (and still won't be next year, even with the performance lead) to charge a premium for their offerings. It just won't work.
AMD's strategy is simply getting people, who've maybe used intel their entire lives, to use their stuff and realise it's fine. They have to increase their mindshare.

>> No.68796890
File: 58 KB, 2000x1543, 1535526563072.png [View same] [iqdb] [saucenao] [google] [report]

>the 10 other CPU bugs/exploits
>he thinks it's only 10

>> No.68796910
File: 256 KB, 483x581, 1539990217776.png [View same] [iqdb] [saucenao] [google] [report]

Post yfw AMD fucking delivers

>> No.68796927
File: 386 KB, 1000x1000, lisa_su_pepe.png [View same] [iqdb] [saucenao] [google] [report]

still won't believe this 'leak'
too good to be true. TSMC must have amazing yields and good binning process from AMD
>6c/12t entry level chip for 99burgers
no way in hell

>> No.68796929
File: 34 KB, 184x184, 1543532926643.jpg [View same] [iqdb] [saucenao] [google] [report]


>> No.68796944

>TSMC must have amazing yields and good binning
See >>68796858 and also https://www.youtube.com/watch?v=I6AG-woscDU

>> No.68796981

Actually possible. They can make the same chiplets for consoles, mainstream and server. This means they have at least 90% yieldy since the defective ones can still be used. High clocks are possible this way. I'm still not seeing 16 cores with 5 GHz turbo with 125W on x570.

I think the Ryzen 9 will actually have 12 cores . This would also fit the expected benefits from 7nm better.

>> No.68796992
File: 571 KB, 1024x572, amd-chiplets.png [View same] [iqdb] [saucenao] [google] [report]

See pic related. They can utilize more of the silicon on a wafer, defects aren't nearly as costly, all of their different market segments allow for consolidated scaling and binning.

And it's going to get even better for Zen 3 in early 2020. This is the true genius of AMD's long term strategy.

AMD knew one day that new game consoles would be launched and that they could do it at near cost as before for each console manufacturer. But this time, they were going to leverage the economies of scale of the console market for all of their other products. The PS "Five Guys" and Xbox "Zero Ultra" are both speced out as 8c/16t CPUs + 12CU Navi on the same SiP. But they're actually using two 8c Zen 3 chiplets with as many as 4 defective cores each. The interposer is the same one used for Epyc/Threadripper, which has redundant memory controllers for HBM2/GDDR6/DDR4 on the same die. PS5 is using HBM2, next XBox is using GDDR6. The Navi chip is the same 20CU chiplet used for Zen 2 / 3 APUs and discrete GPUs. All of their junk silicon is going into the consoles, and this is going to massively increase the amount of silicon for higher-tier bins. Zen 3 will have enough volume for 5.6GHz or higher bins on the desktop (another 10%-15% clock improvement in a single year) plus one other big surprise. Even thought Intel has had a much larger market share, and thus has traditionally been able to bin their CPUs higher, AMD is going to completely destroy them when it comes to yields and binning.

This is why AMD is completely confident on TSMC's 7nm, while Intel is royally fucked.

>> No.68796993

Yeah, I'm looking at the compile times benchmarks which show the numbers I wrote. AMD will have a good decade. They got the Ryzen series just in time when CPU bugs were disclosed, which is a huge deal as AMD CPUs are now genuinely not only better at price/performance, but better at performance in general. This makes them more valuable than Intel, which, paired with their low cost makes them an only option for any use-case and is why a lot of big companies are moving to AMD now. Anyone buying Intel at this point is retarded.

AMD is vulnerable to Meltdown-BR, a recently discovered attack. They're vulnerable to Spectre 1 and 2 attacks. I hope you know the image you posted is a joke. But yes, Intel is vulnerable to 5x as many exploits.

>> No.68797000

Is Noctua NH-D15 sufficient for 105W tdp?

>> No.68797004

good chance you'd buy the bottom of the bin chips on non-X models now unlike Ryzen 1xxx series if we take that leak for granted. don't expect a 99$ part to OC higher then 4.6/4.8 max

>> No.68797005
File: 276 KB, 584x562, 1534697438305.png [View same] [iqdb] [saucenao] [google] [report]

I wonder if the 3600/X could get to 5.0 single

>> No.68797008
File: 374 KB, 576x578, [FFF] Love Live! - 09 [BD][1080p-FLAC][9FFEBEBC].mkv 00_01_24.459.png [View same] [iqdb] [saucenao] [google] [report]


>> No.68797013

it's rated for like 250W

>> No.68797016

>They can make the same chiplets for consoles
Most funniest fucking thing is that PS 5 going to have ONLY bad/faulty Zen 2 bins...meaning mere 10% off of Zen 2's entire yields. Everything else is PERFECT and goes straight to the PC MASTER RACE (down to mid tier offerings, at the very least).

>> No.68797023

>24T @5GHz
Fuck me... Guess it is time to make a new desktop.

>> No.68797031

>AMD is vulnerable to Meltdown
And that, my friends, why you should NEVER respond to buttmad Intbeciles any to begin with.

>> No.68797037

if they really use 2 chiplets, yes. But even if they use 1 chiplet it would be the low clocking ones, that wouldn't even be usable in pc.

>> No.68797039
File: 58 KB, 645x729, 80c.png [View same] [iqdb] [saucenao] [google] [report]

I still don't understand the concept of a chiplet, can someone explain pls?

>> No.68797047

>I'm still not seeing 16 cores with 5 GHz turbo with 125W on x570.
propably around 135-150W with 1 core turbo to 5.1Ghz while other cores are at base or near it.
STILL absolutely wrecks anything on the market

>> No.68797059

You don't need that, though.

4.2GHz stable on all 16 cores or 4.4GHz stable on all 8 cores 24/7/365 >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> "MUH 5GHz on just first two cores while everything else gets downclocked massively" (c) Inturd.

>> No.68797082

>which has redundant memory controllers for HBM2/GDDR6/DDR4 on the same die
That's not how it works.

>> No.68797090

>a small chip
>I don't understand

>> No.68797097

still too optimistic I think. But even 12c/24t at 5 GHz boost (150W) with better latency and therefore ipc for $349 would wreck anything. and if they use chiplets in desktop CPUs that's definitely possible.

>> No.68797098

>if they use 1 chiplet it would be the low clocking ones, that wouldn't even be usable in pc
All 1-chiplet designs have Navi APUs with at least 4GB GDRAM on board instead of dummied out empty space, massively increasing the value.
Also, you get 6 core AT THE VERY LEAST.
If you think this is "unusable", you're a fucking idiot.

>> No.68797114

read the posts before, I'm talking about how and will increase yield and improve silicon quality by using the worst silicon in next gen consoles.

>> No.68797121

I/O is separate and much larger, chiplets are smaller chips which are wired/connected to the much larger I/O. Insane yields, better efficiency, great security, amazing scalability.

>> No.68797131


ps5 and probably the next xbox will not need super high clocks since they're targeting 60 fps at best. so a chiplet with each core clocking up to ~2.5 GHz will already be a huge improvement compared to jaguar but not really useful for pc.

>> No.68797136

Thanks anon.

>> No.68797152

Sucks the 3850x needs a beefier mobo as my 470 is limited to a 3700x not enough of a upgrade form my 2700x
I wonder if we will get mcm gpus? 15%+ perf over a 56/64/1080 isn't worth a upgrade but the navi gpus sound interesting I'll wait for arcturus

>> No.68797158

20 24 28 without a chiller

>> No.68797165

>1 core turbo to 5.1Ghz
Stop projecting, coping Intbecile in heavy denial of harsh reality.
Only Inturd does this shit like this, AMD doesn't "underclock everything to shit to get slightly better perf on first two cores". They underclock SOME notnetheless, yes, but only at extremely high core counts AND EVEN THEN they do NOT underclock much WHILE having solid high clock on more than 2 cores (usually 4 at the very least). Sure, 16 core 5GHz stable at all times is not achievable right now on air, but you're surely to get at least 4.4GHz stable on all 8 with Zen 2 (and that will actually be a DOWNCLOCK from default Turbo specs, lul), which is INSANE and utterly BTFOs Inturd the flying fucks out.

>> No.68797172

What are you talking about?

>> No.68797180

>since they're targeting 60 fps at best
At 4K, though.

>> No.68797191

Zen is absolutely invulnerable to Meltdown by the very architecture to begin with.

>> No.68797194

3.7-3.8 ghz they will get from these chiplets easily.

>> No.68797206

It's 560 tier actually

>> No.68797210


It won't even outperform the 8700k in Photoshop. Only in vms and blender/premier rendering.

>> No.68797215

>l-lol amd can't even do 16 cores @ 6ghz, intel wins a-again baby

>> No.68797221

Do you think it's worth waiting for these? I'll probably do a CPU upgrade in the near future and I was thinking about grabbing an r5 2600. I'm just concerned about that bullshit where I'll have to wait until compatible but affordable motherboards to avoid having compatibility issues and having to do that bullshit where you have to ask AMD for a kit so that you can update the BIOS beforehand.

>> No.68797225

Mind share anon

>> No.68797245

Which would still be incredible for an iGPU though. I'm still amazed at what the 2200g and 2400g can do.

>> No.68797249
File: 94 KB, 950x288, new-meltdown-attacks.png [View same] [iqdb] [saucenao] [google] [report]

AMD is vulnerable to MeltdownBR
>The researchers demonstrated their attack on an Intel Skylake i5-6200U CPU with MPX support, an AMD 2013 E2-2000 and an AMD 2017 Ryzen Threadripper 1920X. They note this is the first time a Meltdown-style transient execution attack has been shown to be able to take advantage of delayed exception handling on AMD hardware.
>Meltdown-BR - exploits an x86 bound instruction on Intel and AMD


>> No.68797257

560 tier is already here, in that Apple one, I forgot how it's called.

>> No.68797278

>*on air it can't, that is

>> No.68797281

AMD is beating Intel to a pulp, they got stronk to the point they're really making this motherfucker bleed

I remember years ago buying a cheap AMD build and thinking "damn, I bet this piece of shit would be better with Intel". Meet Haswell, unimpressed and now this bullshit. AMD is forcing their cock down Intel's throat with no shame.

>> No.68797303

>Ryzen Threadripper 1920X
1. SEV blocks any and all attacking attempts = inapplicable in real life.
2. Zen 2 has Meltdown/Spectre hardening at hardware level.
But nice try anyway, Intbecile.

>> No.68797316

>1. SEV blocks any and all attacking attempts = inapplicable in real life.
That's just memory encryption.
Totally unrelated to timing side-channel attacks.

>> No.68797317

My guess is

3920WX - 24c/48t
3950WX - 32c/64t
3970WX - 40c/80t
3990WX - 48c/96t

All capable of 5GHz single-core turbo

>> No.68797333

big, if true

>> No.68797337

>That's just memory encryption.
It's not, you uneducated fuck:

>> No.68797339

I didn't talk about gpu performance at all.

>> No.68797343

If you don't need it yet it's worth waiting cause either you get one of the new ones once they are available or a cheaper 2600/2700 in a few months, no?

>> No.68797349
File: 227 KB, 1280x960, 1489825231844.jpg [View same] [iqdb] [saucenao] [google] [report]

can't wait for ryzen 9 to come out

>> No.68797353

It is just a memory encryption.
Either full memory (SME) or keyed per-VM (SEV).
You've been spamming for a while already, and you're still dumb as fuck.

>> No.68797354

which is a renamed old Spectre vulnerability - it doesn't cross the current privilege level, so it's not on the same level as the old Meltdown vulnerabilities.

>> No.68797370

Are fanboys this fucking delusional and insecure? If you actually read my post and thought "this guy is an intel shill" then you're fucking retarded. I clearly stated that anyone buying Intel now is an idiot and that AMD is the only x86 brand worth buying.
Fucking kill yourself, you behave like a crybaby feminist. As soon as ONE FUCKING STATEMENT goes against your believes you immediately start crying and calling people shills.
I'm not an Intel shill, and I'm especially not an AMD shill now that I know I might be associated with retards like you. I'm just saying what's officially accepted by researchers as true. If you think it isn't then go ahead and get a fucking job in security research and prove everyone wrong.

>> No.68797379

That's just our local spamming retard.

>> No.68797389

>t. shill

>> No.68797391
File: 127 KB, 926x624, Pottery.jpg [View same] [iqdb] [saucenao] [google] [report]


>> No.68797410

Nice try, CTS employee.

>> No.68797414

g versions of 3 and 5 are just weaker versions of x? why bother? what's the use case for them? tdp is the same.

>> No.68797418

Calm down, schizo-kun.

>> No.68797422

>why bother? what's the use case for them?
You're also brain damaged besides just being blind, huh?

U seem MA

>> No.68797424

>he doesn't know

>> No.68797428

igpu, you blind fuck

>> No.68797451

New versions of 2200/2400G with integrated graphics.

>> No.68797464

nowhere in that picture it said that g stands for igpu, faggots. now i remember why i don't come to these shitty threads.

>> No.68797473

Just wait©
Another massive deception
I bet it will be 5ghz on 1 core only.
All cores will be shitty as always on base frequency.
Also it's a 'leak' of poojetv notorious amd shill

>> No.68797474


Why so many cores?

>> No.68797482

fuck off you retard.

>> No.68797499

kek, well said.

>> No.68797511

>t. intfail bellend

>> No.68797518

xtreme sniper precision dorito boosted dew dispenser? fuck off with your abbreviations

>> No.68797525
File: 470 KB, 1536x2048, 1541538140567.jpg [View same] [iqdb] [saucenao] [google] [report]

8 7nm Zen chiplets, 1 big fat I/O die

>> No.68797532

>Tsmc is claiming a 20% drop in power usage over 16nm
I don't even know if I should laugh my ass off at your idiocy, or be very sad and feel sorry for you.

Here's a tip:
>"TSMC 7nm is two modes: either 100% of Zen+ performance at 50% power consumption, or 125% of Zen+'s performance (IPC/PPC-wise) at 100% of Zen+'s power consumption."
(c) AMD

>> No.68797542


>> No.68797546

X is XFR enabled, you dumb fuck.

>> No.68797558

Every Zen chip supports XFR.

>> No.68797563

amd is the chinkphone of the processor world

>make cheap product that ALMOST competes with the top brand
>push it so that it can compete
>have to raise prices to cover the new R&D
>/g/ poor faggots are too blind to see that you're now the villain

we need a new cpu producer to undercut amd and intel, competition is stagnant

>> No.68797568

sorry for hurting your feelings, you blind fuck

>> No.68797587

I almost feel sorry for you

>> No.68797597
File: 27 KB, 326x294, 2gpygg.jpg [View same] [iqdb] [saucenao] [google] [report]


>> No.68797635

no it means extreme, meaning it has higher clocks and tdp from non-x counterpart. every one of them has xfr.

>> No.68797645

>have to raise prices to cover the new R&D
If these prices are real that wouldn't be true yet. Maybe by Ryzen 3-4 they have enough market share to implement abusive pricing, but so far Ryzen is the sweet spot for consumers.

>> No.68797671
File: 31 KB, 640x480, 1499021837602.jpg [View same] [iqdb] [saucenao] [google] [report]

doesn't this design adds latency?
I don't want AMD to let me down

>> No.68797672

Turbo =/= XFR, you dumb fuck.

>> No.68797686

Every Zen chip supports XFR.
It's just some ~50MHz for non-X models.

>> No.68797723

>every one of them has XFR
Turbo is NOT XFR, you dumb shit.

>> No.68797761
File: 56 KB, 688x610, Screenshot_2018-12-05 What is %%title%% - Custom PC Review.png [View same] [iqdb] [saucenao] [google] [report]

are you an idiot? nobody said anything about turbo.

>> No.68797834

>undercutting intel by less than 5%

>> No.68797846


The point the other guy making that seems to be lost on you, is that big.LITTLE is a complete fucking waste of engineering resources x86 side. The uArch of mainstream PC doesn't have power caps, and by the time we scale down to 3nm EUV, the power usage per core will be basically 16nm ARM tier, in which case we can stick that into a phone and suddenly ARM is completely BTFO.

A single Zen+ core at idle uses 3W and under load does ~7W @ 14nm. 7nm (no change in uArch) it would do 1.5W idle and 3.5W load. But uArch redesign also means better perf and efficiency for the same power drop.

Since 7nm is a full node shrink from 14nm, the 0.5x drop in power rated by TSMC holds true. 3nm is a full node shrink down from 7nm, therefore a 0.5x power rating over Zen2, will also hold true.

This means that a Zen5 core @ 3.3GHz would use 750mW at idle and 1.75W underload. Each Zen2 core has two true SMT threads, unlike Intel's half-assed garbage. So on paper and in system, Zen2 is 1c/2t. The power rating above applies to 2 threads at 100% usage for load and 2 threads at 1-2% usage for idle. If you lower the voltage on the core and drop it to 2.8GHz for example, you might be able to get idle power as low as 500mW and 1.25-1.5W underload. Double that and you get 1W for idle and 1.5-2W load spec for a 2c/4t Zen5 chip @ 2.8GHz for example.

Can you stick this in a phone? No, too expensive power wise. What about a tablet? A high end one definitely, low-mid-range unlikely; not without further clock drops. That said though, the IPC difference between x86 and ARM is pretty massive, even though ARM is RISC (which is more fixed purpose, and therefore has many gains to improve the gap as a result). But ultra-thins? Yeah. You can have very high performance ultra-thins now as a result.

Additionally, Zen2 has 4x256-bit AVX instruction pipelines. By Zen5, assuming standard growth patterns, it would be entirely fair to expect 4x512-bit AVX pipelines. Which'd be insane perf overall.

>> No.68797887
File: 53 KB, 764x464, 3x91bih2bg221.png [View same] [iqdb] [saucenao] [google] [report]

Sorry, I see now that OP cut the prices from the leak. The difference is more like half I think, unless Intel reacts.

>> No.68797910

>only costs $9990

>> No.68797931

>only 9990kelvin heat output

>> No.68797940


To bring you up to speed, he's been in the industry longer than any of those people you mentioned doing reviews and interviews/deep dives into tech (CPU & GPU predominantly) for well over a decade. Is pretty respected, and considered to be a very reliable when it comes to leaked information as a result.

For him to confirm what Jim said in his 5GHz Zen2 & Navi video (part of OP's post), to be say 90% true, means that we're in for a really wild ride next year. The kind of ride that makes Intel start pulling their anti-trust violation tactics that got their offices raided by multiple countries, multiple times, and fined twice for over a billion dollars, by multiple nationalities to compete with AMD (as they won't have a proper answer till Q2 2020). Of course, this time around the entire world has their eyes on Intel--because they know the kind of shady shit Intel has pulled in the past and the lengths it will go to stiffle competition and maintain its insane profit margins.

If they so much as breathe wrong this time around, regulators, especially in EU & Asia will bring the hammer down on them and throw the fucking book at their face with anti-trust violation charges and fines; won't happen in America though, cause the justice system at the very top is corrupt as fuck. That said though, it won't matter.

When Zen came out, many reviewers initially didn't do a good job in doing comparisons because Zen was the first new AMD uArch in over a decade, whilst the last decade Intel spent time putting out the same Arch w/ 5% YoY improvs. Which lead to insane compiler optimizations. As a result, Zen1 was like 90% of Kabylake IPC. But AMD competed via core/thread performance & power/perf/value greater than Intel.

Intel doubled down and charged more, while token slashing prices. Zen2 is AMD matching or exceeding Intel IPC, with better perf/power/watt AND superior fab process AND core/thread # in consumer space. Basically disembowling Intel. Expect shady shit soon.

>> No.68797946

>crashes premiere

>> No.68797965


>> No.68797972
File: 1.48 MB, 480x320, 1456316821018.gif [View same] [iqdb] [saucenao] [google] [report]


>> No.68797985

Must be interesting to be so good at your job that only person you compete with is yourself

>> No.68798006

>2024? Why do you need a completely new uArch on absolutely new materials in 2028, /g/oy?

>> No.68798080


>> No.68798093

>Keller will make intel a shiny new architecture
he's doing SoC's there, would like to know what exactly. i mean he could do it without breaking a sweat in his spare time but will he?

>> No.68798094

>implying any of that matters

>> No.68798107

Daily reminder that your computer is a container, and software will grow to fill it like a liquid. The more powerful the hardware gets, the slower the software gets.

>> No.68798228
File: 53 KB, 1024x256, amd-next-horizon-40-1024.jpg [View same] [iqdb] [saucenao] [google] [report]


>R3 3300X
>R5 2600 +10-15% IPC improvment
>for $129.99


Assuming a 10% conservative increase in IPC, we're looking at: 193.6 whose closest equivalent in consumer grade CPU performance is the Core i7-7820X | According to the amazon link from the source above, that costs $749.99.

Essentially a toaster CPU can match Intel's HEDT class 8c/16t chip. That's crazy. If we assume that Zen2 brings 15% IPC increase (no change in clocks in this comparison yet), then we're looking at: 202.4 whose closest equivalent consumer grade CPU performance is the Core i7-8700 | According to the amazon link from the source above, the K series costs 369.99, so a non-k you can assume to be around 30-40 dollars less. Let's run with $329.99 to be safe.

So now you have a $129.99 CPU matching Intel's own 6c/12t part for $200 less. That's beyond crazy.

But we KNOW based on this slide from AMD's next-horizon keynote, that Zen2 is bringing 1.25x perf over Zen1, aka 25% performance increase. This includes clock & IPC gains together. So taking that into consideration, and applying it to the R5 2600X, we're looking at: 220 whose closest equivalent in consumer grade CPU performance is the Core i9-9900K.

So now you have a $129.99 CPU, the R3 3300X matching the single-core IPC performance of the Core i9-9900K. This would confirm suggestions from other anons & just some logicization, that the Core i9-9900K is now only capable of competing with the R5 3600X, & that the 3600X will likely be superior to this product while costing ~6-700 dollars LESS, at a lower TDP.

Finally, this would also mean that Intel would not be able to compete with AMD at the 3700/X AND 3800X and 3850X segments at all. Nothing in their entire CPU lineup would be able to match AMD's at perf/watt, IPC, & above all core/thread @ near 5GHz or at 5GHz clocks. All with AMD's offerings between anywhere from 500-1000 dollars less--if leaks prove true.

>> No.68798349
File: 46 KB, 968x681, notbad.jpg [View same] [iqdb] [saucenao] [google] [report]

thank you anon, very cool!

>> No.68798460

Try to keep up with conversations instead of blindly replying.

>> No.68799254

>Advertising lower production costs for them and bigger dies as something good
Why would server owners care about that. Monolithic is more efficient and that's all that matters

>> No.68799273
File: 385 KB, 620x573, 18d9d1f83ff78fe07b502b6c2146199fbdb69d8b74bb0dc4e415b956239ef701.png [View same] [iqdb] [saucenao] [google] [report]

>> No.68799341


This fucking timeline. Also mobo makers can get fucked. They're just upset that AMD has promised socket longevity which eats into their sales. They've gotten used to a new socket every year from Intel to drive sales, and it doesn't surprise me that they make poverty boards and shity BIOSs to try and drive sales instead.

>> No.68799405

So nothing out of the ordinary given they're going 7nm.
Why are people excited? The fact it's not intel?

>> No.68799467


I mean you say that but it isn't always true. Unfortunately some programs cannot be infinitely parallelized by definition to leverage more cores. These programs rely on 1-2 main threads to get the work done. This is why PBO is so good now and is going to be even more amazing on Zen2+. Boost the best core(s) to what you need, while retaining good multi threading. If you need pure multi thread performance just do an all core OC. With PBO you are limited by your best core, with all core boosts/OC you are limited by your worst core.

>> No.68799523


I think what I was more trying to convey was that we might start seeing integration of heterogeneous cores into a single CPU.

It's all speculation anyway

>> No.68799562

I wish i could get hyped for new games so i had a reason to upgrade.
My 2500k is still trucking on for my day to day use, i need a game good enough to ruin my economy and social life to justify the upgrade.

>> No.68799686

>i need a game good enough to ruin my economy and social life to justify the upgrade

>> No.68799706

also Star Citizen

>> No.68799723

Yep, but it's not released yet, lul. You can fully play the MONSTROSITY that is X4 RIGHT NOW.
16K res support, strictly ONLY Vulkan, up to 256 cores multi threading, PBR & shiet...

>> No.68799739
File: 30 KB, 298x279, 1541538409899.png [View same] [iqdb] [saucenao] [google] [report]

>16K res support, strictly ONLY Vulkan, up to 256 cores multi threading, PBR
this is a sfw board now, anon.

>> No.68799757

>They're just upset that AMD has promised socket longevity which eats into their sales
Except AyyMD releases new motherboards every time they release a new CPU.

Asus was the one who revealed that Z270 boards could run Coffee Lake but Intel didn't want to.

>> No.68799770

I already play the X4 games on my Sandy system. Good suggestion though.
I'll consider it once it's actually playable for more than a few hours before the content becomes monotonous and/or pointless.
I still have hope though, praise Christ Roberts.

>> No.68799849

NH-D15 is rated for 250-275W anon.

>> No.68799881

>I already play the X4 games
X4, not 4X, you dumbo. 4X is a genre, X4 is a newest entry in GODLIKE single player (mostly) Space Sim series which saved entire genre from completely obliteration back in late 1990s/early 2000s.

>> No.68799959

>amd releases new cpu
>here's a new motherboard specifically for it, but olders work too
>old intel motherboard would work for new cpu
>intel:'lol no fuck you'
What's your argument here?

>> No.68800096
File: 50 KB, 584x575, it's a miss.jpg [View same] [iqdb] [saucenao] [google] [report]

Looks like you're the dumbo, read my post again carefully doofus.

>> No.68800142

>X4 games
You really should go and buy yourself a brain.

>> No.68800209

>Boost clocks are probably on 1-2 cores.
that's not how ryzen works.

>> No.68800662

and their dynamic lightining is some magnitudes of order better than any rtx shit nvidia has throw so far

>> No.68800691


oh fuck off the last one was a FUCKING MESS in literally every single ASPECT

>> No.68800762


Name (leave empty)
Comment (leave empty)
Password [?]Password used for file deletion.