AMD Vega 10, Vega 20, and Vega 11 GPUs Detailed

[RejZoR]

I certainly hope Vega will be a success. We need to get product share of both closer to 50%. For obvious reasons. 2017 will certainly be interesting, for both, consumers and AMD.

EDIT:
People dramatizing like crazy again, not realizing even Fury X runs all current games at 4K over 30fps. I'd say that's pretty damn good considering it's last year's card. Whatever Vega will be, you can be assured it'll run things well even if it's not absolute king of the hill.

 

[Captain_Tom]

Even worse, it's a 225W part. GP102 is 250W, but that's without HBM and Vega 10 is supposed to be "between GP104 and GP102". Meaning weaker than GP102.
AMD is skating to where the puck is, not where it will be

Also, first half of 2017 can easily turn into a July or back-to-school launch. Oh well, I wasn't planning on buying any of these anyway.

Can we drop the whole HBM vs GDDR5 thing?


It doesn't matter if they make the card with potatos or Jet packs. What matters is the final product.

Furthermore it kinda sounded like Nvidia was looking at GDDR6 over HBM2. GDDR6 will offer HBM1 - Cheap HBM2 like performance, but with 3x the power usage.

 

[Captain_Tom]

I certainly hope Vega will be a success. We need to get product share of both closer to 50%. For obvious reasons. 2017 will certainly be interesting, for both, consumers and AMD.

Nvidia is a company that is 2-4x bigger than Radeon (Keep in mind most of AMD is devoted to the CPU tech). If Radeon capture even 40% again like they used to it would be a big deal imo. Especially as Intel iGPU's continue to slowly eat away marketshare.

 

[ensabrenoir]

.....so from Hype train:

to silence:

to slowly back peddling....

then a "Second place Champions" /great value /
a little something something for our loyal sheep release......

...... yep....typical Amd.....

 

[bug]

Can we drop the whole HBM vs GDDR5 thing?

No, we cannot. If Titan X was HBM2 powered it would be a 225W part with more power than Vega 10.

If we drop "the whole HBM vs GDDR5 thing", all that's left is that AMD will have a part that's as powerful as the GTX 1080, only a year later. Are you happier now?

 

[Vayra86]

Can we drop the whole HBM vs GDDR5 thing?


It doesn't matter if they make the card with potatos or Jet packs. What matters is the final product.

Furthermore it kinda sounded like Nvidia was looking at GDDR6 over HBM2. GDDR6 will offer HBM1 - Cheap HBM2 like performance, but with 3x the power usage.

GDDR6 doesn't exist. They call it GDDR5X and it's already here.

 

[Captain_Tom]

GDDR6 doesn't exist. They call it GDDR5X and it's already here.

No they already announced GDDR6 buddy. Google is your friend.

 

[Captain_Tom]

No, we cannot. If Titan X was HBM2 powered it would be a 225W part with more power than Vega 10.

But it doesn't. So it doesn't matter what it COULD BE if it isn't.

 

[Vayra86]

No they already announced GDDR6 buddy. Google is your friend.

Actually all they did was vaguely hint at GDDR6, and just a year ago, GDDR5X 'was the new GDDR' and also touted GDDR6. Release for this presumed GDDR6 are somewhere in 2018, in other words, this is completely up in the air at the moment.

Google is my friend - the only confirmation of sorts are 'we will probably'. Seeing as GP102 launched with good ol' GDDR5, I think we shouldn't get ahead of ourselves right now. This falls into the same category as GlobalFoundries being all happy about how they are going to push 7nm very soon. Meanwhile, we've been sitting at 28nm for over two years longer than expected and we saw Intel move away from Tick/Tock.

Perspective

 

[phanbuey]

Let's wait for benchies and pricing.

But yeah bummer...

 

[m1dg3t]

Perspective

If it were up to nVidia we'd prolly still be using GDDR3

 

[LightningJR]

Usually we use single precision to calculate gaming compute right?

If so I am reading 512GB/s and 12 TFLOPs. That would put it at or slightly above the Titan.

You can't compare TFLOPS across GPU manufacturers if you could then the RX480 with 5.8TFLOPS would wreck a GTX 980 with only 5TFLOPS and it doesn't. The card falls somewhere in between the 980 and a 970 (with 4TFLOPS)

The TFLOPS are there but it just doesn't translate in to equalivent performance so 12TFLOPS looks sexy it just don't translate 1:1 with nVidia unfortunately.

 

[Steevo]

If that picture is accurate, it's going to have a lot more than 4096 stream processors. That's how many the 28nm Fury X had and these chips appear to have the same die space (perhaps even more because the HBM chips appear smaller).

Is Global Foundries really that close to 7nm? How can GloFo be making such rapid improvements when Intel appears to be stuck? If this is accurate (I'd gander that Vega 20 is smoke and mirrors), GloFo could pass up Intel in process tech and that's quite unfathomable.

AMD not having an answer to Pascal for another three quarters is dire news. Vega 10 is no doubt beyond Pascal's reach but by the time it launches, it will have to contend with Volta.

If AMD's goal is to Nano (huge chip, low power) their entire product line up, that eats directly into AMD's profit margins. This news post is making me...

7nm is looking like a go as they are already in risk/validation phase with test patterns, X-ray double patterned immersion lithography requiring two processes with polarization to get better fin patterns is the supposed process. http://www.geek.com/chips/x-ray-lithography-continues-to-advance-551433/ It was the thing IBM was working on for a long time they got.

https://en.wikipedia.org/wiki/LIGA

 

[qubit]

The only way Nvidia can neutralize that advantage is if AMD continues to fail to capture marketshare. Early reports show AMD capturing marketshare with polaris, but imo they will need to capture at least 40% for them to stop worrying about Nvidia's numbers advantage.


Otherwise note that this will have 12 TFLOPS vs 11 of the Titan X, and 512 - 1 TB/s of bandwidth beats the 480 of the Titan X. Even if it doesn't soundly beat the Titan X, it will easily trade blows if these specs are true.

P.S. Where is this 512 GB/s coming from? HBM2 comes in 720 GB/s and 1 TB/s flavors so I am calling BS on that spec.

AMD might capture market share if they're priced well, but NVIDIA want the performance crown at all costs and that's what I'm talking about, since they then set the prices and get the prestige that goes with that performance crown.

How Vega fairs against Volta remains to be seen and until I see reviews of the final product on TPU which shows it beating NVIDIA's best (Volta) or at least equalling it, I'm going to remain sceptical. All the usual stuff about power consumption, heat and noise still applies, too. Personally, I'd rather a card was 10% slower, but much quieter. Protecting what little sanity I have left is very important to me!

 

[HD64G]

You can't compare TFLOPS across GPU manufacturers if you could then the RX480 with 5.8TFLOPS would wreck a GTX 980 with only 5TFLOPS and it doesn't. The card falls somewhere in between the 980 and a 970 (with 4TFLOPS)

The TFLOPS are there but it just doesn't translate in to equalivent performance so 12TFLOPS looks sexy it just don't translate 1:1 with nVidia unfortunately.

Wrong. They translate perfectly in true gaming performance in Doom using Vulcan. Divide FPS/TFlops there and check again. Software was keeping CGN behind until Vulcan. DX11 and OpenGL time will be over in 2017 for any new game imho. That's why nVidia pushed Volta in late 2017 afterall.

 

[Vayra86]

Wrong. They translate perfectly in true gaming performance in Doom using Vulcan. Divide FPS/TFlops there and check again. Software was keeping CGN behind until Vulcan. DX11 and OpenGL time will be over in 2017 for any new game imho. That's why nVidia pushed Volta in late 2017 afterall.

Those APIs are still too rare in real world applications to say anything substantial about that.

 

[bug]

If it were up to nVidia we'd prolly still be using GDDR3

Yes, because it's Nvidia that sells rebranded products from 3 years ago.
Neither Nvidia nor AMD will innovate unless pushed to. Baseless extrapolations do not help.

@HD64G Are you really inferring the future of GPUs from one title or am I missing something?

 

[Captain_Tom]

You can't compare TFLOPS across GPU manufacturers if you could then the RX480 with 5.8TFLOPS would wreck a GTX 980 with only 5TFLOPS and it doesn't. The card falls somewhere in between the 980 and a 970 (with 4TFLOPS)

The TFLOPS are there but it just doesn't translate in to equalivent performance so 12TFLOPS looks sexy it just don't translate 1:1 with nVidia unfortunately.

It does in Vulkan lol.

http://i.imgur.com/OITaDBd.jpg

I never claimed that it directly translates though. But if you look at the Fury's TFLOPS it will be a 30-50% increase over that, and actually Polaris offers more perf/TFLOPS than Fiji did - so this is quite promising.

 

[Captain_Tom]

Yes, because it's Nvidia that sells rebranded products from 3 years ago.
Neither Nvidia nor AMD will innovate unless pushed to. Baseless extrapolations do not help.

@HD64G Are you really inferring the future of GPUs from one title or am I missing something?

Hey when a higher binned 7970 can beat a 780 why release something new?

To be fair I think AMD is done with its near-constant rebranding. However they wouldn't have rebranded cards 3 times if Nvidia could make an arch that ages better than fruit.

 

[bug]

Hey when a higher binned 7970 can beat a 780 why release something new?

To be fair I think AMD is done with its near-constant rebranding. However they wouldn't have rebranded cards 3 times if Nvidia could make an arch that ages better than fruit.

Right, when can it it beat the 780? https://www.techpowerup.com/reviews/ASUS/GTX_780_STRIX_6_GB/25.html

 

[Captain_Tom]

Those APIs are still too rare in real world applications to say anything substantial about that.

Really rare? Mantle was "rare", but this is different.

Tomb Raider, Hitman, DOOM, DEUS EX, and now BF1.

It's pretty clear that this will be the new standard of 2017.

 

[Fluffmeister]

Even worse, it's a 225W part. GP102 is 250W, but that's without HBM and Vega 10 is supposed to be "between GP104 and GP102". Meaning weaker than GP102.
AMD is skating to where the puck is, not where it will be

Also, first half of 2017 can easily turn into a July or back-to-school launch. Oh well, I wasn't planning on buying any of these anyway.

Seems they rounded down to get that 150W Polaris too.

Still 5-6+ months away anyway, can't get excited over this yet.

 

[thesmokingman]

If that picture is accurate, it's going to have a lot more than 4096 stream processors. That's how many the 28nm Fury X had and these chips appear to have the same die space (perhaps even more because the HBM chips appear smaller).

Is Global Foundries really that close to 7nm? How can GloFo be making such rapid improvements when Intel appears to be stuck? If this is accurate (I'd gander that Vega 20 is smoke and mirrors), GloFo could pass up Intel in process tech and that's quite unfathomable.

AMD not having an answer to Pascal for another three quarters is dire news. Vega 10 is no doubt beyond Pascal's reach but by the time it launches, it will have to contend with Volta.

If AMD's goal is to Nano (huge chip, low power) their entire product line up, that eats directly into AMD's profit margins. This news post is making me...

That pic is from the Fiji release. Big grain of salt...

What? No wood screws?

lol, did you go to the link? Any pics there?

 

[cdawall]

Yes, because it's Nvidia that sells rebranded products from 3 years ago.
Neither Nvidia nor AMD will innovate unless pushed to. Baseless extrapolations do not help

8800GTS 512, 9800GTX, 9800GTX+, GTS 250. Know what those 4 different graphics cards have in common? They are all the same damn card. Don't forget nvidia has rebranded and carried over cards for generations just as often as AMD.

 

[the54thvoid]

Hey when a higher binned 7970 can beat a 780 why release something new?

To be fair I think AMD is done with its near-constant rebranding. However they wouldn't have rebranded cards 3 times if Nvidia could make an arch that ages better than fruit.

And on the flip side Nvidia haven't had to innovate too much because of a lack of serious competition. Anything AMD released, Nvidia matched or beat within weeks, they're always holding back. Until Vega, Nvidia have the high ground and will (ab)use that position. If Vega doesn't dislodge Titan X as the undisputed champion (especially in Vulkan Doom) all hope is lost.