Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I wonder if this would ever give Intel a shot at a win in the next generation console space. AMD has been running away with that one. Something like this with HBM with a TDP of 100W could be quite promising.
 
  • Like
Reactions: Ener Ji
It does make one wonder why Iris Pro died. If Apple wasn't enough volume for the Iris solution it is hard to believe that this part will have the volume. Of course the R & D for the GPU core actually get spread a scores lots of implementations so that is a thing all to itself.

I suppose the big win here is that Intel no longer has to invest their own R&D manpower into a GPU that was always inferior to nVidia and AMD competition. Perhaps in the long run, they'll go back to Imagination for the low-end and scale AMD to anything above that.

In any event if combined with the Zen based APU's from AMD this chip could be a nice solution for Apple, Zen covers the very low power end while this guy is the chip for intro class performance machines.

I don't see what a Zen APU would bring to Apple's table that this does not. Competition and thus lower prices, but other than that?
 
Cool! This should lead to some interesting Mac Book Pros. The interesting question here is how many variants are planed. For example will there be a 16 GB HBM variant and is this HBM memory shared between the GPU and the CPU? I pose these questions because it would be huge win for Apples designs if they got rid of the external RAM space completely. If people think this is strange it will happen sooner or later in laptop computers, if for nothing else it results in a faster RAM subsystem for the CPU/GPU.

It's a single stack of HBM2, at least from the imagery shown. The GPU portion is not massive either.
I would expect the HBM2 to be GPU specific, and likely come in a 4GB stack, with an 8GB option at the high end (+$100).
The GPU is rumoured (very vaguely) to be around 3.3 TFLOPS, so not as fast as the RX580 or 570.
This is about having decent graphical performance in a very small footprint for small devices - MBP, Mac Mini, etc.
 
  • Like
Reactions: Kopa and Ener Ji
So this kind of performance will be the "new" IRIS pro level?
[doublepost=1510007975][/doublepost]
It's a single stack of HBM2, at least from the imagery shown. The GPU portion is not massive either.
I would expect the HBM2 to be GPU specific, and likely come in a 4GB stack, with an 8GB option at the high end (+$100).
The GPU is rumoured (very vaguely) to be around 3.3 TFLOPS, so not as fast as the RX580 or 570.
This is about having decent graphical performance in a very small footprint for small devices - MBP, Mac Mini, etc.
So if its lower than 570, and we already have MBP with 560....whats the point?
 
  • Like
Reactions: Ener Ji
So this kind of performance will be the "new" IRIS pro level?
[doublepost=1510007975][/doublepost]
So if its lower than 570, and we already have MBP with 560....whats the point?

560 is a lot less powerful than the 570 however. And it takes a lot of board space.

This is about integration, power reduction, size reduction. Not cost reduction, given EMIB and HBM2, although there may be knock-on effects there that reduce system level costs. Iris wasn't exactly cheap either.

I expect we will see a new MBP in H1 2018 then. The top end SKUs may not use this - 6C CPU with discrete graphics, but under that this may find a home.
 
Hmmmmmmm it is an interesting concept, but also worrying that neither company feels they can take on Nvidia on there own. Nvidia have been killing it with their 10 series GPUs, and the SOC in the Nintendo Switch can handle Doom even, the new one not the original.

Still this is all good competition.
 
The partnership will allow AMD and Intel to better compete with Nvidia in the high-end laptop/compact desktop market.

I also think it has to do with ensuring that it can compete with the theoretical idea of the Mac line up using a customer ARM processors so that Apple thinks twice if the gap between Intel and their own design is so narrow that the move away from Intel isn't worth the risk/costs involved.

Side issue: I find it funny how when Apple went exclusively with AMD that we had arm chair CEO's claim it had to do with cost - maybe, just maybe, that the move to AMD exclusively and the deal with Intel were disclosed to Apple and thus Apple viewed it as a safe bent along with the fact that AMD provides hardware specifications and source code to OEM's - that the argument was one of being technically better rather than just saving a few bucks.
 
Hello new Macbook Pro :)

If there is a trend in all of this, weather be performance of CPU, Network cables, or anything performance wise to get better speeds, they all have one thing in common.. Shorter distances over shorter connections.
 
Hmmmmmmm it is an interesting concept, but also worrying that neither company feels they can take on Nvidia on there own. Nvidia have been killing it with their 10 series GPUs, and the SOC in the Nintendo Switch can handle Doom even, the new one not the original.

Still this is all good competition.

This is all about low power, increased performance and reducing the footprint.
 
If it's not a fat, powerful GPU for professional applications that competes with the likes of Nvidia 1080 GTX Ti & Kepler, than I'm not really interested. I'm glad they are making these advancements which will help baseline and mid-tier consumer-grade computers, but I am still waiting for Apple to pack some punch in a recent desktop release. The iMac Pro is coming soon, but I'm more interested in the upcoming Mac Pro for sure.
The bigger problem is the limitations of upgrading GPUs in Macs. They can include a powerful GPU, but it'll be outdated in a year or two. Best example: Latest Mac Pro had some of the most expensive GPUs possible, and then they became pointless quickly.

No current Mac has a PCIe slot. Apple only supports their own blessed GPUs well. You either go with an old Mac Pro + a non-Apple GPU (might not work as well as it should) or an external GPU (same problems plus whatever issues Thunderbolt adds).

It sucks. In any cheap PC, I can just stick a GPU in. Not hard.
 
Last edited:
Intel is doing this because they know what Apple is doing.

Probably just as much Apple knows what Intel is doing.


Intel is scared of Apple.

Probably not.


Intel talks to their relatively large customers about what they like and dislike about the products that Intel provides. This likely was an outgrowth of discussion of how Intel could do a better "Iris Pro" product (with eDRAM). Bigger GPU with more VRAM (measured in GBs not MBs) . There is a cap on on how large silicon dies can be and still be relatively inexpensive. If the GPU and CPU don't have to share die space each can be individually cheaper.

There were lots of designs ( Apple 15" and several from Dell , HP , etc. ) that coupled H-class CPU+limited GPU combos with discrete GPUs and GDDR5 VRAM. This all comes prepackaged by Intel. It makes the design process simpler for the system builders. It isn't about being 'scared' or 'competing'. It is about making the system vendors job easier and more effective.

Intel wasn't completely 'winning' all the CPU + dGPU + VRAM system designs before. Here they are supplying the one part that works. They'll have to give some of the money to AMD for their die but largely are taking the 'value add' away from the motherboard vendor/maker. Until Intel has a dGPU solution this is mostly a win-win-win for intel, AMD , and the system vendors ( the contract logic board vendor perhaps not so much).

Smaller, thinner logic boards with less work ... Apple will be all over that. ( by all appearences Apple liked the eDRAM "Iris Pro" solution more than most other vendors). This will likely be that much better (but a bit more expensive). Success will probably hinge on how many other designs and vendors can make it work in reasonably popular systems too. Most likely Intel has already asked and it is enough that they are doing the product. (whether the sometimes fickle system vendors follow through is an open question. )

P.S. At one point Intel was a larger player in the motherboard business. They bought other peoples chips and put them on Intel motherboards. This largely the same just at the chip packaging levels. It is one step closer to the "complete system on a chip" solution. More stuff from the motherboard gets sucked into the black hole of the CPU package. This is just the next iteration.
 
Only quad core is the current leak. However, there is a Geekbench floating around for a 6-core H series coffee lake, not sure if it’s real but it makes some sense. The specs are supposedly 6/12, 2.6/4.0 GHz, 9 MB L3 cache, presumably 45W TDP.

https://www.notebookcheck.net/First-Coffee-Lake-6-core-mobile-CPU-benchmarks-are-in.245804.0.html

Right, but with these (which are Kaby Lake) shipping in Q1, will Intel ship Coffee Lake-H in parallel? Or does it push Coffee Lake-H further back?
 
Intel is scared of Apple.

Screenshot_20171107-074106.jpg


Mac has 3.34% marketshare in the real get serious things done computer space.

Not worth being scared of.
 
  • Like
Reactions: rGiskard and NGruia
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.