AMD released the first 7nm GPU.
But the yields are still bad, so the price is very high.
But still it won't perform better than nvidia.
Nvidia will only release their 7nm GPU after the process is well matured.
Thus this makes AMDfags the 7nm beta testers.
This is intcel cope
COPE
>yields are still bad
source: my ass
AMD's 7nm has better yields than Intel's 10nm
are you new to how nvidia and amd do things?
I don't know of 7nm's yields being particularly bad
yields are never amazing early on though
This but unironically
So, like what massive bloatfilled OS and super unoptimaized video game do I have to install to make something like this worth it? Do I need to fold proteins in the background too?
Didnt ARM announce they're making a 3nm chip soon?
This is getting close to the limit of silicon.
5nm and they have to move to another material.
Odds are that's fake news.
Unless the chip isn't silicon based.
See
>r*ddit posting
seething little faggot
that was some pseudoscience quantum tunneling moore's law bullshit that got proven wrong in 2006 when 3nm transistors were first successfully printed.
people thought the same thing in 1997 when mercury vapor lamps couldn't output a wavelength small enough to etch smaller than 250nm but they just overlapped 2 images and made higher density smaller transistors with a lower production cost
>yields are so bad
as they have been, for years. That's why nvidia made RTX, so they can make ASICs on their GPU and pretend to gain performance while in reality, GPU performance upgrades are dead.
what other materials can we use
what about wood?
wood screws & graphene
idk about arm but imec and cdns have sucessfully made 3nm test chips using 193nm euv immersion lithography and samsung announced earlier this year that they've got that shit planned
Not as bad as being beta testers for hardware backdoors for 11 years.
>op talk about gpu
>fags talk about cpu
>Posting Intelshit in a GPU thread
>yields are bad
no, they are not, they are expensive, go see a price of a silicon disc for 7nm, go check the size of that silicon it is huge. 7nm is a thing for a long time alreadly.
7nm is going to be busted and outdated in no time, I don't get the hype.
Its like someone freaking out about a Pentium processor. You know number 2 through 4 are coming.
I'm waiting on the .07nm personally.
We already had 1nm transistors 12 years ago. It just won't be cost effective to print that shit for consumer level gear for a long long time. Especially, when it means the very dead end of the technology. Which is nearly there as you can plainly see (adding more cores instead of speeding them up trend.)
I will beta test 2x the performance of a 9900 for half the price any day, intcel HAHAHAHAHAHAHAAAAAAA
>.07nm
Not possible. A silicon atom is 0.11nm in diameter. The spacing between atoms in a silicon crystal is 0.21nm (center to center, not edge to edge).
>beta testers
yeah cause you know MI 50 cards didnt exist prior to radeon VII
/fucking idiot
fanatsytuic
Typical ARM core has significantly lower transistor count compared to the high power architectures, especially if we assume it's going to be just a technology demo with little extra features (single core, no igpu, no fancy coprocessors) it's easily doable for them
>beta testers
For the whole process? No, Apple's A12 Bionic was before them.
>7nm doesn't matter!
>process shrink doesn't matter!
The yields are not bad you idiot.
Radeon VII is literally server Insctinct MI50 rejects that don't make the cut, with features restricted and sold as gaming GPUs.
The only thing that makes RVII an expensive GPU comparatively is the cost of the 16GB of HBM2.
zen 2 got 78% yields.
>Thus this makes AMDfags the 7nm beta testers.
Hopefully this is the thing that Navi actually beta tests.
AMDfags are the 7nm fluffers.
Why are people posting Ryzen on a GPU thread? Do I sense insecurities?
you know there are sizes smaller than a nanometer right. it wouldn't be too big a deal to make a 950pm node by the time 1nm is a big thing.