Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
If it is faster at 75W than a 1050 Ti, it would be a buy.

If it's about as fast at 75W as a P2000, it would be amazing.
 
If it is faster at 75W than a 1050 Ti, it would be a buy.

If it's about as fast at 75W as a P2000, it would be amazing.
At 75W? O_O

Notebook version of GTX 1050 Ti is faster than desktop version, because it is clocked over 200 MHz higher. 35W Vega Pro 20 is faster than Notebook version of GTX 1050 Ti by a huge margin.

Desktop GTX 1050 Ti: 1290 - 1392 MHz(Turbo Boost) core clock.
Notebook GTX 1050 Ti: 1493 - 1690 MHz(Turbo Boost) core clock.
 
Cheap, pre-existing, eGPUs or PCs without PCIe power connector.
Have you read the rest of my post? ;)

At 75W TDP we are talking about 1.6 GHz GPU, with 1280 GCN cores, and 2.4 Gbps HBM2 memory stack, with 307.2 GB/s. That would result in GTX 1060 desktop performance levels(10-15% slower).

ALUs are ALUs.
 
True, but what matters is the relation of performance to form factor. If Vega delivers, the MBP will be an uncontested GPU performance leader in its form factor (we are talking about 3+ GFLOPS in a sub 2kg laptop!) . The potential exception would be Razer Blade of course, but that is a machine designed for a very different purpose.
I don't think a qualifier is needed in this particular instance as the GPU will available to other manufacturers. Hopefully Apple can make the most of the GPU software-wise, and maybe -just maybe- find a way to improve the heat dissipation a little bit; they are not exactly world-beaters in that regard.

Regardless, I'll be the fastest and most future-proof dGPU in a Macbook to date that's for certain. If it does nothing other than bring the temps down a bit during regular and intense workloads it'd be worth the upgrade to most users.
 
Have you read the rest of my post? ;)

At 75W TDP we are talking about 1.6 GHz GPU, with 1280 GCN cores, and 2.4 Gbps HBM2 memory stack, with 307.2 GB/s. That would result in GTX 1060 desktop performance levels(10-15% slower).

ALUs are ALUs.
OK, I guess it should be at least as fast as a P2000 then.
 
Those TDP values are my main concern. AMD chip that's faster than Nvdia equivalent AND doing so at half of Nvidia TDP. That's like, never happened before. Looking back when Apple released the iMac Pro they really made sure everybody knew about the 2x scaling in fp16 compute. So while I believe the claim "up to 60% faster" is ok if you add "in fp16 based compute", I still don''t think that the 60% improvement in 3dmark is happening.
 
As I have said before: AMD has such brand perception, that even if they will have better product than their competition - nobody will believe in them. If they will have equally good product as their competition: 70% will buy Nvidia cards. People want AMD to be competitive. Why? Because then they can buy cheaper Nvidia cards.

Cursed Market.

Let me repost that 3dMark benchmark: https://www.3dmark.com/compare/3dm11/12875524/3dm11/12886777#

One particular thing, that shows GPU performance in this bench: Graphics Score: Radeon Pro 560X: 7552, Vega Pro 20: 12405. +64% score for Vega Pro vs Radeon Pro 560X.

And yes, this is still 35W TDP GPU, because that will be its BIOS limit, as all MBP GPUs are ...

This is the GPU, with the same score: https://www.notebookcheck.net/NVIDIA-Quadro-P3000.191075.0.html

1280 CUDA cores, with under 1.3 GHz core clock. ALUs are ALUs. You are getting GP106 GPU, with 1.3 GHz, but from AMD, in much smaller thermal envelope.
 
Let me repost that 3dMark benchmark: https://www.3dmark.com/compare/3dm11/12875524/3dm11/12886777#

One particular thing, that shows GPU performance in this bench: Graphics Score: Radeon Pro 560X: 7552, Vega Pro 20: 12405. +64% score for Vega Pro vs Radeon Pro 560X.

And yes, this is still 35W TDP GPU, because that will be its BIOS limit, as all MBP GPUs are ...

This is the GPU, with the same score: https://www.notebookcheck.net/NVIDIA-Quadro-P3000.191075.0.html

1280 CUDA cores, with under 1.3 GHz core clock. ALUs are ALUs. You are getting GP106 GPU, with 1.3 GHz, but from AMD, in much smaller thermal envelope.
That looks like mobile Polaris against desktop Vega.
 
That looks like mobile Polaris against desktop Vega.
Because it is Radeon Pro 560X compared to Vega Pro 20. 64% increase in Graphics score, perfecrtly in line with Apples 60% increase in performance, over Radeon Pro 560X.
 
Because it is Radeon Pro 560X compared to Vega Pro 20. 64% increase in Graphics score, perfecrtly in line with Apples 60% increase in performance, over Radeon Pro 560X.
Are you saying the Vega in the test is running at 35W on a desktop?
 
Oh yes, of course, it would essentially make the whole Nvidia lineup obsolete.
Up to GTX 1060 3 GB - pretty much yes. Even at around 65-75W TDP Vega 12 will be at best 1.6-1.7 GHz GPU, with 2.4 Gbps HBM2, resulting in 10-15% behind GTX 1060 6 GB desktop GPU.
The point is that you buy a 1030 when you cannot feed the GPU with more than 30W.

I would happily pay up to 150$ for 30-45W Vega 12 on desktop, that would be fanless ;).
 
I would happily pay up to 150$ for 30-45W Vega 12 on desktop, that would be fanless ;).
I would argue that if 16 Polaris CUs cost $99 in 2017, Vega 12 should not cost more.

Or is Vega 12 the chip code for Vega 20?
 
Last edited:
Or is Vega 12 the chip code for Vega 20?

Vega 12 is the chip for Vega Pro 16 and Vega Pro 20. Vega 20 is a not yet released rumored/leaked big Vega chip that is supposed to replace Vega 56 and Vega 64.
 
Vega 12 is the chip for Vega Pro 16 and Vega Pro 20. Vega 20 is a not yet released rumored/leaked big Vega chip that is supposed to replace Vega 56 and Vega 64.
Pro is just Apple branding.
 
Pro is just Apple branding.

Technically, its AMD's branding. But I am sure that Apple's marketing dep. had a say in it :D Still, doesn't change the fact that its a new AMD chip.

I would argue that if 16 Polaris CUs cost $99 in 2017, Vega 12 should not cost more.

Vega's RAM is much more expensive and also the interposer is an added cost.
 
Technically, its AMD's branding. But I am sure that Apple's marketing dep. had a say in it :D Still, doesn't change the fact that its a new AMD chip.



Vega's RAM is much more expensive and also the interposer is an added cost.
Does AMD use the same Apple Pro branding elsewhere?

16 Polaris CUs would be equivalent to 12 Vega CUs. It is AMD's problem if they cannot upgrade the memory while keeping the price.

I would say 16 Vega CUs should cost the same as 16 Polaris CUs, as they come two years later.

If he was talking about 20 Vega CUs, it would be OK for the card to cost more than $99 (there's already a big premium to go from 8 to 11).
 
Does AMD use the same Apple Pro branding elsewhere?

They don't sell the same products to anyone else, if that is what you ask (e.g. there is no laptop using full Polaris GPU with 16 CUs enabled). They do use Pro branding on their professional GPUs.

It is AMD's problem if they cannot upgrade the memory while keeping the price.

It's one way of looking at this. But it has to be commercially viable for them, or why would they sell the product to begin with. There is a reason why Nvidia does not use HBM2 — too expensive. Vega is a premium product in the sense that it uses very expensive tech. They can't really sell it to an average person. Apple is great partnership for them, since customers are ready to pay more for Apple products.

I would say 16 Vega CUs should cost the same as 16 Polaris CUs, as they come two years later.

I'm not sure I share your theory on price formation :)
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.