• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD VEGA confirmed for 2017 H1

Status
Not open for further replies.
The only way the big Vega is not more powerful than a 1080 is if they screw up like ATI did with 2900xt, even a shrinked furyX on 14nm would beat or close to a 1080 with a few bottle necks removed.
 
TNA said:
The type of evidence you are talking about typically only come from reviews before release. AMD or Nvidia do not release the info you are after until the card is out or about to come out within a week or something.

Loadsamoney thinks it is a 1070 competitor also Griffildur :p They say this, but they never say how they arrived at that conclusion. Just a gut feeling or something? I explained all the way back in September why I expect it to be Titan XP performance. At worst however they will easily beat a 1080. They have had tons of time to get the drivers right also. People who talk about the 4K Doom and Battlefield demo; they will have had 5-6 months to optimise drivers before launch since we saw that so I expect much better performance than what we was shown.
Click to expand...

It'll be great if they do.
 
TNA said:
I am expecting 2 small vega cards and 2 big vega cards. Think 1070/80 and Titan XP/1080Ti. Or for those who cannot fathom that, think Fury and Fury X for big Vega.
Click to expand...

I am one of those people who "cannot fathom that".

The whole referencing xp/ti/fury/fury-x is too confusing between brands, let alone a specific card.

What 'name' should I expect as being the equivalent of my 290 in the upcoming releases?

If it has an 'x' after it (e.g. 290x or 590x) I may be able to fathom that i.e. it's a bit faster. But the whole naming system is just too confusing
 
kration said:
I am one of those people who "cannot fathom that".

The whole referencing xp/ti/fury/fury-x is too confusing between brands, let alone a specific card.

What 'name' should I expect as being the equivalent of my 290 in the upcoming releases?

If it has an 'x' after it (e.g. 290x or 590x) I may be able to fathom that i.e. it's a bit faster. But the whole naming system is just too confusing
Click to expand...

But AMD did it with Ryzen. R7, R5 etc.... done so the average consumer can relate to the i7, i5 etc...

Depends on the marketing needs.
 
AMDs naming scheme is a lot easier to grasp than Nvidias Imo. First number is the series its from, second is what tier level it is, third only used for refreshes and is always just a 0 or a 5. Lastly stick an X on the end if its the go fast version. Easy.
 
Lokken86 said:
AMDs naming scheme is a lot easier to grasp than Nvidias Imo. First number is the series its from, second is what tier level it is, third only used for refreshes and is always just a 0 or a 5. Lastly stick an X on the end if its the go fast version. Easy.
Click to expand...
how is that different from Nvidia, they both have the same naming scheme ?
AMD used to have more complicated naming before polaris with X variant, which they do not do anymore.
 
kration said:
Which also depends on my buying decisions, and I'm too confused before I'm even in a position to buy anything.

I'm being a bit too dramatic ;)

I'd just like to know what to expect in the medium term.
Click to expand...

The way I do it, is based on need and benchmarks.

For example, when Vega launches I am going to get a 1440p monitor so will buy what ever card gives me the performance I need at 1440p based on price.
 
Last edited:
Lokken86 said:
Because of all the Titans. I'm beginning to lose track of the TitanX now. Or is it TitanXP. Wait TitanXp?
Click to expand...
i think Nvidia learned the lesson, next Titan naming will be much more clear, like Titan Ultra, Titan Extra, etc...:D
the Titan XP or Xp is probably due to poor planning, imo TitanX pascal was not supposed to be the cut down version of GP102, it was supposed to be the full chip like everytime, and the Ti the cutdown, the only problem is that AMD did a weird launch, 1st year line up only with low-mid range, so Nvidia decided to make the cut down version the Titan since they do not need the Ti ( and they need to milk ), then launch the Ti with full chip by the end of 2017, the problem again is that AMD's high end took so much time, that nvidia was forced to launch the Ti before the time windows between the 2 lineups becomes a problem, ppl doesn't like when they buy a high end to see it dethroned couple months later by another card from the same vendor, and nvidia had too much built up stock since last summer, with no competition again nvidia decided to make the Ti the cutdown version instead of the full chip, and shift the Titan to full chip, got milk ?
 
Last edited:
AlamoX said:
it's like AMD s trying a new market segmentation, low-mid-high without many sku's inbetween 480 is about double the performance of the 460, and vega about double the performance of the 480, this would make sense for navi, and the way it scales, 1-2-4 GPUs
Click to expand...

You will still be able to get tight segmentation with Navi. There will be two 'base' chips, let's say X and Y, where Y has 1.5 times the CUs of X.

You thus get 1/2/4-chip configurations across scales that don't overlap (1*1.5/2*1.5/4*1.5 = 1.5/3.0/6.0) meaning your range runs across 1/1.5/2/3/4/6 multiplier range, which could translate into x60/x70/x80/x90/Fury/FuryX.

Imagine for example X=16 CUs and Y=24 CUs:
  • Using 1/2/4 for X will give you 16/32/64 configurations.
  • Using 1/2/4 for Y will give you 24/48/96 configurations.
  • Between the two, you get a full range 16/24/32/48/64/96 configurations.
Sure, the lower-end spectrum ends up being more 'packed' at steps of 8 CUs, whereas the high-end runs at steps of 16/32 CUs, but they might even bother to mix X+Y chips (depending on how well this works of course).

EDIT: OR you might get a Z chip with 20CUs to fill that space.
 
Last edited:
Status
Not open for further replies.
Back
Top Bottom