Fucking EXPECT 100500W Fukushima boiling lava SUNFIRES now. For 100500 Masonic tickets, no less.
AHAHAHAHAH, LITERAL "RING OF BINGBUSES"!!!!!111ONEONE
Looks like 1st gen Threadripper.
Are you having a stroke... Your post is incoherent bollocks
>amdpajeet is afraid
>copy big.LITTLE config
>stack up dies so heat won't ever leave the cores
>baseclock will never go over 1.4Ghz cause throttling
yay, what a revolutionary step!
Foveros is by definition for low-power/mobile shit only.
High bump densities don't do well with high, inhomogeneous, or temporally unstable heat dissipation, since silicon actually has a thermal coefficient of expansion, go figure.
>you will witness the first ever 2k W CPU
I'm ready.
While it is true that making sense is optional here on 4channel, I wish you'd tried.
see>Foveros is by definition for low-power/mobile shit only.
Look at those glued together repurposed desktop dies!
OH NONONONONONNO MUH GLUED TOGETHER
>Foveros is by definition for low-power/mobile shit only
NNNNNOOOOOOOOOOOOOO WAIT FOR 10NM+++++++++++++++
MCM, absolutely REVOLUTIONARY
DELETE THIS
So now that Intel is dead can anybody other than AMD make 10nm or 7nm x86 CPUs?
bump
Atleast intel will have far better IPC than poozen 2, imagine a 5ghz processor getting btfo by a 1.4ghz processor because of autism in cache
I'd say VIA, but they're a long running joke at this point.
>pic related
The chinese
:DDDDDD
based and redpilled
>intel gets BTFO by chinks
this is a new low
NEW!
>clockspeed doesn't matter!
Man it sucks that Intel went down the shitter, both literally and figuratively.
Based
>AMDfags can't even make a coherent post
Not surprised in the least.
>p-pls no bully
>Single core in 2019
Is this a joke?
how new?
Yes, but they did release a quad-core (no HT) for LGA2066. Most mobo makers told them to fuck off
>$2000 CPU
>18 cores
HAHAHAHAHAHAAHAHAHAHAHAHAHAAH
HAHAHAHAHAHAHAHAHAHAHAHAHAHAH
AHAHAHAHAHAHAHAHAHAHAHAHAHAHA
ARE THESE KIKES SERIOUS?
The toilet makes perfect sense. If someone uses it, it notifies the government that sends a death squad after that person.
inteljeets are applelfags in denial
>>stack up dies so heat won't ever leave the cores
t. doesn't understand heat transfer
If intel dies then AMD pretty much becomes a monopoly
Interested in where you can actually buy these. I wonder how good the igpu is
People are laughing but 3d stacks are the future. Just like chipper design with amd is basically an intermediate position since you can only glue so many cores together till you have to stack them
you can feel the nervousness of a trembling amd nigger in this post
Makes perfect sense to me... These CPUs are gonna run really really hot and cost many many USD.
>Interested in where you can actually buy these
I believe it's just to reduce their reliance on non-Chinese tech, I'm sure Intel would find a way to bar them from selling it anywhere in the West anyway
>I wonder how good the igpu is
No idea, but the CPUs in pic related have Piledriver-tier performance (might be only because they can't clock higher)
Have you tried speaking English?
Thread theme:
> People are laughing but 3d stacks are the future.
I wonder how you're going to cool that shit. Wouldn't heat pockets be a massive problem?
Single core is all that matters bro AHHAHAHAH
>clockspeed doesn't matter!
clock speed is not an immediate indicator of performance. Clock speed is a factor of a larger equation including cache, core count, and architechtural improvements. Do you know how CPUs even work?
delid sir
Do you know how your mom's pussy works?
They haven't come out publicly with there solution but they should near release.
They'll be lowly clocked though to combat the heat. Like with ryzen since it's only the beginning they'll begin to optimize with later revisions. But they must have plans if they've gone this far into R&D.
Clever arrangement if the stacks can make a big difference in the heat dissipation.
Here is something for everyone in this thread that thing this is a joke.
youtu.be
decent video but he seemed to have no concept of IPC when talking about processor clock rates
FX failed because of its low IPC compared to Intel
Ryzen is successful because it's IPC is so close to current Intel chips at a fraction of the price
>If intel dies then AMD pretty much becomes a monopoly
Th-They will do a better job at spending billions on R&D than intel, right?
>decent video but he seemed to have no concept of IPC when talking about processor clock rates
>FX failed because of its low IPC compared to Intel
I think that's were the bigger cache comes in. He mentions with the design you could get a tb worth of cache in there so he assumes intel will atleast be throwing in a 100mb cache which I'm guessing would make up for the low IPC.
Don't forget intel has alot of money and they've been picking up alot of engineers in the last two years so I'm sure they'll come up with something good.
The managed to outsmart Intel at a fraction of their budget, imagine what they could do with a massive R&D budget
Chink shit is just licensed and shit.
Actually, Ryzen/Threadripper completely BTFO Intel's HEDT segment in mid 2017 into 2018. Intel never intended on releasing their Skylake-X HCC SKUs into the HEDT market. They are going with Kaby Lake-X-Skylake-X LCC only.
Kaby-Lake-X was meant to the low-end portion of the Socket 2066 family. Ryzen shat on it hard and Intel decided to go with Coffee Lake. Kaby Lake-X died quickly.
I just cant see them producing 10nm and also 3D stacking at the same time
they have struggled to produce anything 10nm for the past 5 years, and with 14nm+++ a 3D stacking fabrication would quite literally be housefire-tier
plus those chips would be so absurdly expensive I cant see them being in any laptop
3d offset bad yields with smaller specialist silicon.
Since yields could go up you can actually get cheaper
3D stacking is only viable for parts that don't produce much heat, unlike a CPU
too bad intel cant get any yields at all with 10nm chips kek
I thought that too, but that says 'compatible with intel virtualisation' and AMD-V is quite different from VT-d to the point where all hypervisors need separate implementations.
Probably just an error in marketing
Mini heat sink tubes going between the package layers?
Lots of glue (Intel Heat Paste™)?
Reminder that Jim Shit-wrecked Keller said Intel's 10nm would NOT have any clockspeed regression compared to 14nm.
Itt: amd is already afraid
Yeah and SHITel also said 10nm is mobile only
You are both lamers.
>Foveros is by definition for low-power/mobile shit only. High bump densities don't do well with high, inhomogeneous, or temporally unstable heat dissipation
Now tell that to Jewtel and it's brainless poojeet slaves.
Foveros goddamn it if you're Greek this sounds so ridiculous.
>Dies are below
>Several stacks of DRAM on top
>LITERAL greenhouse
>can anybody other than AMD make 10nm or 7nm x86 CPUs?
Loongson and Qualcomm both working on something. Russian MCST is moving to 10nm and then 7nm too, soon.
>MCST is moving to 14nm
Fix. Not 10nm, sorry.
Yep.
|
|>
|
|
|
>WAIT FOR 10NM+++++++++++++++
What do YOU think?
>If someone uses it, it notifies the government that sends a death squad after that person
...or MINES buttcoins. Literal SHIT MONNIES farming. Remember Aachi & Sippak?
>3d stacks are the future
Yes, yes they are. But NOT the way RETARDS from Inturd think it's able to be applied. Like, stacking DRAM on TOP of CPU is THE FACTO most AUTISTIC thing ever. A literal-fucking-greenhouse.
>Foveros is by definition for low-power/mobile shit only
AHAHAHHAAHAHHAHAHAHHAAHAHHAHAHAHHAAHAHHAHAHAHHAAHAHH
The power of Indiantel.
Papermaster, in an interview, said that Zen2/EBYN2 have some things to mitigate the issue. if it's just doubled L3$ or something more remains to be seen/revealed.
No, it's actually these three:
youtube.com
~
youtube.com
~
youtube.com
can't wait for 1 big core and 3 ATOM cores for mobile
>autism in cache
I'm ready for 8gb l4 cash, gimme now
reccommend me a cooler for a ryzen 5 no OC
i have a cryorig C7 but it runs hot as fuck
>recommend me a cooler for a ryzen 5 no OC
>5
>no OC
Stock.
is it possible for AMD to put a single stack HBMeme on top of the I/O-die for Zen3? i do hope with next iteration the I/O die will shrink to 7nm so they have alot more space to tinker with
>i have a Cryorig C7 but it runs hot as fuck
It's only meant for mini-ITX emulation machines...or Raspberry Pi/Odroid.
>for AMD to put a single stack HBM on top of the I/O-die
Why do you think everyone is such an Intbecile as Inturd is?
Show bobs and vagene, mam
>ipc
FX cpus were data starved due to shit cash, upping NB clock had a bigger impact on performance than core clock
>dies have a yield rate
>interposers have a yield rate on top of it, so even a perfect die will be thrown out if the interposer fails
>intel wants to do like 3 layers, with multiple dies on each layer
They only need to use their 10nm process on top of it and we'll get to see sub 1% yields for the whole package
Yields don't matter lol just buy it
is everyone at intel a braindead fucking ape?
They could be planning to go full IBM and run water or a heat exchange fluid through the chip.
>run water or a heat exchange fluid through the chip
This is even fucking worse. Now we're putting liquid cooling directly inside the chip? Holy shit they're retarded
>intel will deliver from now on liquid cooled CPUs only
>IBM is retarded, I wouldn't do it like that
so was Centrino user
>intel doing anything from scratch right
OH NO NONONONO
BASED AND CHECK'D
>TSMC’s 5nm Fin Field-Effect Transistor (FinFET) process technology is optimized for both mobile and high performance computing applications. It is scheduled to start risk production in the second half of 2019.