Vega/Navi Rumors (Updated)

[swilli89]

Enough about nVidia, this is a thread about Vega

So if an unpolished Vega performed ~10% faster than a 1080, I expect we may see a Vega card with mature drivers hitting +20% of a 1080. A card like this that would be nipping at Titan-P's heels would sell like crazy for $600.

 

[SinOfLiberty]

Unfortunately it is dead before arrival. Trades blows with 1080, in most cases 10% behind.,

Trolling is not allowed
Markfw
Anandtech Moderator

 

[Borealis7]

Enough about nVidia, this is a thread about Vega

So if an unpolished Vega performed ~10% faster than a 1080, I expect we may see a Vega card with mature drivers hitting +20% of a 1080. A card like this that would be nipping at Titan-P's heels would sell like crazy for $600.

but that card might require it's own PSU to feed it. AMD is way behind on perf/W and they need to step it up on the power/thermals side.

This is hyperbolic and unnecessarily inflammatory. Please look at overall system power requirements realistically, or cite sources to support your claims.
-- stahlhart

 

[zinfamous]

Unfortunately it is dead before arrival. Trades blows with 1080, in most cases 10% behind.,

Can you please link those benchmarks that no one else has seen?

 

[antihelten]

but that card might require it's own PSU to feed it. AMD is way behind on perf/W and they need to step it up on the power/thermals side.

If the 4x efficiency gains holds true*, and assuming it is a reference to FP16 performance and in comparison to Fury X, then the FP32 efficiency (which is the relevant part for gaming), would be about 2x higher than Fury X, which would mean that Vega has comparable efficiency to a 1080. For comparison Polaris 10 has between 15% and 25% better efficiency than Fury X.

*In relation to this, it's worth noting that the claims of 2.5x efficiency gains for Polaris didn't hold true, unless you count the alleged revised versions.

 

[flopper]

So if an unpolished Vega performed ~10% faster than a 1080, I expect we may see a Vega card with mature drivers hitting +20% of a 1080. A card like this that would be nipping at Titan-P's heels would sell like crazy for $600.

Vega Crushes anything on the market currently and whips the behinds of whats coming.
Its an amazing dominATIon Zenengineering feat from AMD, never say never.

This is a forum for technical discussion, not viral marketing. Cite sources with verifiable data to support your claims, and stop looking ridiculous.
-- stahlhart

 

[zinfamous]

Vega Crushes anything on the market currently and whips the behinds of whats coming.
Its an amazing dominATIon Zenengineering feat from AMD, never say never.

How can you possibly say this without any Vega product on the market for testing?

 

[stahlhart]

Reopening thread. If this gets out of hand again, I'm locking it permanently.
-- stahlhart

 

[guskline]

Hopefully, the CES presentation will put the Vega into perspective relative to other gpus. I think the good news is that a release is likely before mid year. Isn't this the same track that RTG took with Polaris last year?

 

[tential]

Hopefully, the CES presentation will put the Vega into perspective relative to other gpus. I think the good news is that a release is likely before mid year. Isn't this the same track that RTG took with Polaris last year?

approximate release date(Quarter? Can't keep saying 1H 2017... that's just unacceptable) and pricing. (Are you undercutting or matching the competition...?)

I need this by the end of the month for me to be a happy customer. I'm tired of not knowing and waiting for something that may be another 5 months away. I know it's not in AMDs best interest to let us know, but I'm tired of being gamed by GPU companies.

 

[96Firebird]

approximate release date(Quarter? Can't keep saying 1H 2017... that's just unacceptable) and pricing. (Are you undercutting or matching the competition...?)

I need this by the end of the month for me to be a happy customer. I'm tired of not knowing and waiting for something that may be another 5 months away. I know it's not in AMDs best interest to let us know, but I'm tired of being gamed by GPU companies.

I disagree, I think if a release is imminent (say, next 4 months), giving a release date would benefit AMD more than anything. People might be inclined to wait for Vega instead of going for Pascal if they know how long the wait really is. Right now, AMD is not competing in the segment these cards are supposedly aimed at. Their closest card would be Fury (X), which I'm sure they don't want to keep selling for $300.

The only way I see giving a release date would hurt them is if they couldn't actually make that date.

 

[JDG1980]

We've already learned from 9700/9800Pro, X800/850XT PE, X1900/1950XTX, HD5870, HD7970Ghz, R9 290X/295X2 eras that AMD's performance crown means nothing to the very segment of the population that ALWAYS seems to complain about AMD's products. It's good to see AMD finally stopping to care for this group of consumers. Maybe someone at AMD finally looked up the definition of loyalty in the dictionary. Most of PC gamers would much rather want the days of HD4850/4870/5850/5870/6950/7950/R9 290 back than today's market of $700 mid-range cards and $1200 flagships.

AMD doesn't have to beat the best. They just have to shake up the price/performance across key high-end segments as plenty of PC gamers have no interest paying $800-1200 for 2016-2017 high-end cards that will be easily matched by a $450 2018 mid-range. The opportunity for market share gains in the $299-549 space is absolutely enormous right now.

That's fine. But if AMD is trying to beat the competition in perf/$ at mid-range pricing rather than win the performance crown outright, then HBM2 seems a questionable design choice. Everything we know so far indicates that stacked-memory-on-interposer arrangements are more expensive to produce than standard GPUs with GDDR. Wouldn't it have made more sense for AMD to use GDDR5X on a GTX 1080 competitor? I can't see HBM2 bringing in a reasonable rate of return for AMD unless they have a card good enough to command at least $699, and that means they need to decisively beat GP104 and come close to GP102 - close enough that Nvidia can't undercut them with GTX 1080 Ti the same way they undercut Fury X with GTX 980 Ti.

 

[tential]

I disagree, I think if a release is imminent (say, next 4 months), giving a release date would benefit AMD more than anything. People might be inclined to wait for Vega instead of going for Pascal if they know how long the wait really is. Right now, AMD is not competing in the segment these cards are supposedly aimed at. Their closest card would be Fury (X), which I'm sure they don't want to keep selling for $300.

The only way I see giving a release date would hurt them is if they couldn't actually make that date.

Good point.... But maybe if it's too far away I buy pascal? I'm going to probably buy Vega no matter what. This really just goes to how I recommend products for other people.

Amd needs to be more competitive at the high end especially when they have such an amazing monitor advantage in terms of options. I love my 4k monitor and can't wait to get a 120hz one.

I just want to know how far my 1400 budget will take me so I can start getting excited.

 

[Sweepr]

 

[guskline]

Thanks for the video link Sweepr. Perhaps in a little over 1 day we will get more specifics about the Vega gpu in that rig.

Perhaps 2017 will tempt me to build a WC Zen/Vega rig.

 

[Glo.]

https://translate.google.com/translate?sl=auto&tl=en&js=y&prev=_t&hl=ru&ie=UTF-8&u=http://www.4gamer.net/games/300/G030061/20170104006/&edit-text=

Here are even screens with settings. Its 4K Ultra.

Pretty impressive. As a point of view, I also posted benchmarks from last gen. GPUs.

 

[AtenRa]

I believe it had V-sync enabled at 60hz.

 

[ShintaiDK]

https://translate.google.com/translate?sl=auto&tl=en&js=y&prev=_t&hl=ru&ie=UTF-8&u=http://www.4gamer.net/games/300/G030061/20170104006/&edit-text=

Here are even screens with settings. Its 4K Ultra.

Pretty impressive. As a point of view, I also posted benchmarks from last gen. GPUs.

Depends what site you check.

EDIT: Now I found your numbers, old beta numbers from 2015.
http://www.guru3d.com/articles_page...ta_vga_graphics_performance_benchmarks,6.html

 

[Krteq]

THX for vid link

Quite impressive for pre-production sample, but will be nice to see FPS with v-sync off.

//EDIT: More info on VC - AMD demos Star Wars Battlefront on RYZEN and VEGA PC at CES2017

 

[biostud]

Why do they run with v-sync on?

 

[Krteq]

Well, to not show all at once?

 

[antihelten]

Depends what site you check.

EDIT: Now I found your numbers, old beta numbers from 2015.
http://www.guru3d.com/articles_page...ta_vga_graphics_performance_benchmarks,6.html

Based on those numbers, and the fact that the Vega chip was running with vsync on and never dropped below 60 FPS (as far as I could tell*), it would seem that it is at least as fast as a 1080 (which has minimums between 56 and 63).

This would also make it about 25% faster than Fury X, which would make sense if Vega is a 4096 SP, 1.3 GHz chip (i.e. same number of SPs as Fiji, but 25% higher clock).

*Just saw that Krteq's link mentions it dropping to 57 FPS occasionally, but that would still be in line with it matching a 1080.

 

[Thedarkchild]

Guys you cannot compare different benchmarks, on different levels, with different CPUs (this one running Ryzen). No body knows how fast it is, but it appears to be faster then GTX1080 (judging by same level compared in yt).

 

[Glo.]

It appears that Vega is close to Titan X. Interesting how mature are the drivers.

 

[dacostafilipe]

You should try to filter the benches that use the Endor map. Because the min/avg/max fps changes a lot between maps.