AMD accelerates 7nm process adoption

[kalniel]

GF 7NM is cancelled:

https://forums.hexus.net/pc-hardware...tion-tsmc.html

Ah, so hence this news. Definitely sounds more positive to say you've accelerated your process than to say you've had a major partner pull out. At least some AMD PR is on the ball

[CAT-THE-FIFTH]

Ah, so hence this news. Definitely sounds more positive to say you've accelerated your process than to say you've had a major partner pull out. At least some AMD PR is on the ball

From what I gather 7nm Zen actually taped out on TSMC 7NM.

This makes it the fourth straight node GF have cancelled - 20nm,14nm,10nm and now 7nm. GF 28nm was delayed and 14nm was licensed from Samsung since their own one didn't work. Looks like WSA kept them alive - much of the AMD losses were down to WSA charges.

[philehidiot]

Why?
The vast proportion of PC gamers do so at 1080p and the largest sales group is RX580 / GTX 1060 and lower. If AMD wants to go after sales this market would be better, the rest is just bragging rights.

If they want to go after sales, yes this is the market and so why don't AMD just stay here where the investment is lower and so on? Because the margins are way higher on the high end stuff.... or they should be... Also, if you're developing professional cards you might as well do the extra R&D and make a high end gaming card whilst you're at it.

[darcotech]

And if they slap to chips on one card and use Infiny Fabric to make it look like one to OS, so no Crossfire troubles, bud almost double performance? Their new Pro v340 is dual chip.

If it is true that they made them half power use but 1.35 faster, what that makes for 2x Vega(7nm) 64 ?
Faster than Titan V for sure...

[watercooled]

From what I gather 7nm Zen actually taped out on TSMC 7NM.

This makes it the fourth straight node GF have cancelled - 20nm,14nm,10nm and now 7nm. GF 28nm was delayed and 14nm was licensed from Samsung since their own one didn't work. Looks like WSA kept them alive - much of the AMD losses were down to WSA charges.

I'm not too concerned about capacity at TSMC given their size. Losing an option for dual-sourcing is a potential concern, also for competition/pricing reasons, hence my comment wondering what Samsung's response will be, in the other thread.

[Fury559]

End of the year...they better not be lying again. This is close enough to make people wait 2/3 months after Turing, but if we get to the end of December and stock isn't on the shelves the people giving AMD a chance are going to go ballistic. They better keep their word this time around since they seem so confident in their assertions.

Last time I wasn't happy and ended up getting my 1070 in August 2016 after my R9 290 died, however this time I'm happy waiting because I don't need an upgrade and more than happy to wait for 1080ti performance for £300 with Freesync

[ik9000]

Last time I wasn't happy and ended up getting my 1070 in August 2016 after my R9 290 died, however this time I'm happy waiting because I don't need an upgrade and more than happy to wait for 1080ti performance for £300 with Freesync

So, which benevolent company are you waiting for to supply these, and how long do you think it will take for this utopian dream to come to pass?

[badass]

So, which benevolent company are you waiting for to supply these, and how long do you think it will take for this utopian dream to come to pass?

At a guess, that'll happen with the 2nd or 3rd generation of products released on the 5nm TSMC node. So 2022/2023 probably.

[watercooled]

Intel support Freesync too. Just sayin'

[Safetytrousers]

'the next-generation Radeon Vega Instinct GPU...with 32GB HBM2'

No date on consumer products.

[Tabbykatze]

'the next-generation Radeon Vega Instinct GPU...with 32GB HBM2'

No date on consumer products.

It never starts with consumer products, dates will come

[Millennium]

Removed

[watercooled]

AMD are hilarious. They're just announcing the 7nm vega part to secure the console markets (because they don't like kids of course lol). I wonder if they care to give us an update on 7nm Ryzen or are they waiting for memory to actually function?

That whole post makes no sense, unless I'm missing the point? Why would Microsoft/Sony remotely care about what AMD announce in the compute GPU market when they're making decisions on what to implement in their consoles? Also what memory are they waiting for? Zen2 uses the same socket so must therefore be compatible with DDR4. If the 7nm EPYC parts are out on schedule then it's reasonable to expect them to stick to the yearly release cadence for their desktop parts, so around spring 2019.

[Tabbykatze]

AMD are hilarious. They're just announcing the 7nm vega part to secure the console markets (because they don't like kids of course lol). I wonder if they care to give us an update on 7nm Ryzen or are they waiting for memory to actually function?

Everything you just said made very little sense. It frustrates me how some people just can't tell the difference between a compute card and a graphics card...

What "waiting for memory to actually function" nonsense are you on about?

Both HBM2 and GDDR5x/6 works perfectly fine, it just costs a bomb.

[Millennium]

Removed

[Tabbykatze]

Well my post made sense to me. It was made in a rush I will submit but I would appreciate it if you would be more polite thank you.

Taking time to write posts in an understandable manner will prevent incredulous responses.