Intel Chips With “Vega Inside” Coming Soon?

Page 10 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

ksec

Senior member
Mar 5, 2010
420
117
116
And funniest part: Pike, from PikerAlpha has said in December 2016(!), that next generation Mac Mini will be bigger than current one. And more expensive ;). He, in the same post on his blog, said that next gen. iPhone will have wireless charging(and was pretty darn correct on this matter).

The whole point of the Mac Mini was an entry level Mac, regardless of power, i.e its Price point is what makes a mini; mini. Not its size.

Sadly we wont be seeing anything more from Pike, I wish him well. But the last bit he also mention all the CPUID points to Coffeelake, not Kabylake for iMac and Mac mini.
 

Glo.

Diamond Member
Apr 25, 2015
5,707
4,551
136
The whole point of the Mac Mini was an entry level Mac, regardless of power, i.e its Price point is what makes a mini; mini. Not its size.

Sadly we wont be seeing anything more from Pike, I wish him well. But the last bit he also mention all the CPUID points to Coffeelake, not Kabylake for iMac and Mac mini.
Only last two generations of MM have had "affordable" price point. Previously - Mac Mini cost 800-900$ for base model.
 

ksec

Senior member
Mar 5, 2010
420
117
116
Only last two generations of MM have had "affordable" price point. Previously - Mac Mini cost 800-900$ for base model.

I can still remember, as a matter of fact with his voice is in my head now, his joke about having a penny every time someone ask him about affordable Mac. It was at MacExpo. That was the original Mac Mini, still using PowerPC at the time starting at $499.

It wasn't exactly cheap at the time, but it certainly was affordable.

And I still remember when they released the new Intel Mac Mini, they bump the price to $599, that course quite a stir.

The Mac Mini has always been about the entry to Mac ecosystem. And it never did start at $800 for base model.
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
As a matter of fact, I dont understand why i want a Desktop Chip that has this combination and not a normal CPU + Vega.

I have said before, it is just a Normal CPU + Vega.

This is not Vega inside Intel, it is Vega besides Intel, and not even that close.

The packaging benefits are Vega + HBM. The Intel CPU is really kind of pointless in the package. It could have been left off, and packaged separately with no difference in functionality.

I think Intel wanted an EMIB showcase, because otherwise this product doesn't really make sense.

It should really just be GPU+HBM, but then there is very little that is Intel about it.
 
  • Like
Reactions: Phynaz

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
It should really just be GPU+HBM, but then there is very little that is Intel about it.

Really? I thought you knew better than this. Assuming it can come at a reasonable price, integration has advantages in power management. Also because the HBM2 memory isn't shared with the GPU, it can have all the advantages of an iGPU and a discrete GPU without the disadvantages of using either of them.
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
Really? I thought you knew better than this. Assuming it can come at a reasonable price, integration has advantages in power management. Also because the HBM2 memory isn't shared with the GPU, it can have all the advantages of an iGPU and a discrete GPU without the disadvantages of using either of them.

What integration? The CPU is not integrated with the dGPU. It's connected via PCIe relatively far away on a mini circuit board. I don't see how it is significantly different that if you have a separate CPU directly on the MB.

Just look at laptops with dGPU, they allow the option of running with iGP or dGPU depending on performance and Power desires, there will be no change with this part.
 
Last edited:

ksec

Senior member
Mar 5, 2010
420
117
116
I have said before, it is just a Normal CPU + Vega.

This is not Vega inside Intel, it is Vega besides Intel, and not even that close.

The packaging benefits are Vega + HBM. The Intel CPU is really kind of pointless in the package. It could have been left off, and packaged separately with no difference in functionality.

I think Intel wanted an EMIB showcase, because otherwise this product doesn't really make sense.

It should really just be GPU+HBM, but then there is very little that is Intel about it.

So what you are suggesting is that Intel just wanted to showcase EMIB usage, which provide little to no technical and price benefits?

I think I will just wait for Anandtech 's technical analysis.
 

jpiniero

Lifer
Oct 1, 2010
14,591
5,214
136
So what you are suggesting is that Intel just wanted to showcase EMIB usage, which provide little to no technical and price benefits?

I think I will just wait for Anandtech 's technical analysis.

Not a showcase but more of a proof of concept. They needed something that would have decent enough volume to improve the manufacturing process.

Plus Apple wanted it.
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
So what you are suggesting is that Intel just wanted to showcase EMIB usage, which provide little to no technical and price benefits?

I think I will just wait for Anandtech 's technical analysis.

There are some benefits on the GPU side, since EMIB is integrating the HMB memory with the GPU, for less cost than Silicon Interposer.

But for the CPU connection, it appears to just be a standard PCIe connection, meaning the technical benefits really don't exist. That part is more about Marketing.
 

Glo.

Diamond Member
Apr 25, 2015
5,707
4,551
136
16 ROPs. Not a good information.

Wait... 65W TDP Package is delivering better performance than Core i78550U+ GTX 1050?

That is... extremely promising.

If AMD APU has 32 ROPs - it will be even better.
 

IRobot23

Senior member
Jul 3, 2017
601
183
76
16 ROPs. Not a good information.

Wait... 65W TDP Package is delivering better performance than Core i78550U+ GTX 1050?

That is... extremely promising.

If AMD APU has 32 ROPs - it will be even better.

RX 580 has 32ROPs yet it is beating R9 390X with 64ROPs. RX VEGA 64 has 64 ROPs so it could be at least 2x RX 580 @ clock per clock.

Does this "Quad Geometry Engines" mean that front end is same as VEGA 64?
 
Last edited:

Glo.

Diamond Member
Apr 25, 2015
5,707
4,551
136
Max-Q GTX 1060 has 1280 CUDA Cores, and 1265 MHz boost clock@70W TDP.

Im wondering whether AMD will be able to do the APU with 1792 GCN cores, and sell it.
 

CatMerc

Golden Member
Jul 16, 2016
1,114
1,149
136
16 ROPs. Not a good information.

Wait... 65W TDP Package is delivering better performance than Core i78550U+ GTX 1050?

That is... extremely promising.

If AMD APU has 32 ROPs - it will be even better.
That's 64 ROPS.

It has 16 RBEs, which translates to 64 ROPS. Which also points to 4 Geometry Engines and Shader Engines. Edit: It even says quad geometry engine, so no need to speculate.

This design has as many render back ends and geometry engines as full fat Vega 64. Those 24 CU's well have no problems being fed at all.
 
  • Like
Reactions: Olikan

Glo.

Diamond Member
Apr 25, 2015
5,707
4,551
136
16 ROPs/ 6 CUs then?

That is interesting design.

I wonder if 1792 GCN core APU from AMD will have similar pattern.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,339
10,044
126
If it's 8th gen Core, then it's vulnerable...so why buy it now?
I guess we can pray for a new, bug-fixed stepping? Though, I expect that if the "vulnerability" affects their entire OoO pipeline, then it would be non-trivial to fix, so I'm guessing that they didn't fix it.