Nvidia Volta 20 series coming Q3 2017

[five8five]

They may have learned a little from SK.Hynix it seems

Indeed

 

[Chuk_Chuk]

"Early 2018" could mean January or February. If that's the case the cards will be out long before H2 2018.

This^

The question is are Micron erring on the side of caution regarding the announcement. Time will tell......

My vote is Q1 2018.

I think that if micron was confident in getting GDDR6 out by Q1 2018. They would have said Q1 2018 rather than early 2018. You could argue they are being vague as to not reveal to much information to competition, but i'm not sure how that would help them.

 

[five8five]

I think that if micron was confident in getting GDDR6 out by Q1 2018. They would have said Q1 2018 rather than early 2018. You could argue they are being vague as to not reveal to much information to competition, but i'm not sure how that would help them.

Sorry, I was meaning Volta in Q1 2018.

The blog you linked talks about the partnership in glowing terms so I'd suspect they'll be making every effort to meet Nvidia's deadline.
It's just we don't know exactly what that is

 

[silversurfer]

March 2018 is Volta, extrapolating from their presentations and my incredible xray perceptions on corporate rollout vs costs

 

[five8five]

Here's the guys blog regarding GDDR5X from Feb. 09 2016 - equally vague timescales

There have been a variety of rumors in the industry about GDDR5X availability timeframes, so I’d like to clear that up from the beginning: Micron’s GDDR5X program is in full swing and first components have already completed manufacturing. We plan to hit mass production this summer.

Micron is currently ramping GDDR5X to mass production, and will be announcing sample dates later this spring. We plan to be in full volume production this summer.

So they easliy hit their target with the 1080 being released on the 27th May


https://www.micron.com/about/blogs/2016/february/gddr5x-has-arrived

 

[Chuk_Chuk]

Unfortunately past performance is not an indicator of future performance. Just look at Ryzen.
While vague summer starts from June so it was not far off the mark. Also "summer" is far less vague than "early" imo

 

[five8five]

Unfortunately past performance is not an indicator of future performance. Just look at Ryzen.
While vague summer starts from June so it was not far off the mark. Also "summer" is far less vague than "early" imo

I'd agree 100 percent if it was AMD but Nvidia don't seem to have any problems getting new graphics cards out the door

 

[Chuk_Chuk]

I'd agree 100 percent if it was AMD but Nvidia don't seem to have any problems getting new graphics cards out the door

Sounds like they are due a slip up then.

 

[D.P.]

GDRR6 is very similar to GDRR5x, and the speed overlap a lot anyway so it doesn't really make a big difference. I'm still thinking a 1080 sucessor Q4 2017 using 12-14Gbps GDDR5X, the Titan Volta will use higher speed 14-15Gbps GDRR6 around Q2 2018 on 384bit interface.

 

[Simon Belmont]

a 1080 sucessor Q4 2018

Are you having a laugh? I very much doubt Nvidia will wait that long

 

[D.P.]

Are you having a laugh? I very much doubt Nvidia will wait that long

typo, mid-range Volta will be here by the end of the year

 

[Simon Belmont]

typo, mid-range Volta will be here by the end of the year

Oh that's better

 

[five8five]

typo, mid-range Volta will be here by the end of the year

For a moment I was really worried

 

[five8five]

Sounds like they are due a slip up then.

 

[hobib_ali]

Quick sell your 1080ti's fast other wise you will lose massive value on them! the 2080 will be either 10% or 50% faster than the 1080ti

 

[Dicehunter]

Still guessing Q1 2018 for Volta.

 

[JediFragger]

26th March nVidia GDC would be my bet. It fits.

 

[Mauller]

GDRR6 is very similar to GDRR5x, and the speed overlap a lot anyway so it doesn't really make a big difference. I'm still thinking a 1080 sucessor Q4 2017 using 12-14Gbps GDDR5X, the Titan Volta will use higher speed 14-15Gbps GDRR6 around Q2 2018 on 384bit interface.

GDDR6 appears to break each package into 2 separate memory channels, that should reduce overall memory latency with GDDR6. The bandwidth per package is likely to be the aggregate bandwidth of the two channels, with each channel running at 5-10Gbps etc.