Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Status
Not open for further replies.
According the information gathered by you guys, I'm optimistic to believe the nnMP to be announced at the WWDC with deliveries early Q3.

What I'm not sure is about HBM2 memory, I think Apple will opt for ECC memory for the Higher end NMP and non-ecc for the lower end and dont bit the HBM bait this year, any case being apple so conservative w/o regard HBM2 supply constrains I doubt they will integrate it before a full Zen Mac Pro next year.
 
According the information gathered by you guys, I'm optimistic to believe the nnMP to be announced at the WWDC with deliveries early Q3.

What I'm not sure is about HBM2 memory, I think Apple will opt for ECC memory for the Higher end NMP and non-ecc for the lower end and dont bit the HBM bait this year, any case being apple so conservative w/o regard HBM2 supply constrains I doubt they will integrate it before a full Zen Mac Pro next year.
koyoot - this one is yours.... ;)
 
Hah, I chuckled when I saw that GTC ad on Macrumors :D

Ill watch it anyway. I hope to see another Asynchronous Compute Engine and at least one Hardware Scheduler on Pascal Hardware.
 
Keeping in mind that Apple often gets early access to components what are the odds of any of this showing up March 15?
This "urban legend" that Apple gets pre-release access to silicon keeps coming up among the faithful.

Can you give two or three examples where this in fact happened, instead of just repeating the folklore?
[doublepost=1455590777][/doublepost]
Hah, I chuckled when I saw that GTC ad on Macrumors :D
You should have signed up - we could have lunch!
 
I think it's more likely the Roman Senate will Assassinate Caesar, Again, on March 15th.

hahaha

Et tu, Brute?
[doublepost=1455594179][/doublepost]
This "urban legend" that Apple gets pre-release access to silicon keeps coming up among the faithful.

Can you give two or three examples where this in fact happened, instead of just repeating the folklore?

It's happened over the years. Google is your friend.
 
Yeah, but 5 years ago?
Not the best or most relevant example most likely..

That's so long ago that we are back into the Jobs era of negotiations and operational details. I'd imagine some of this stuff is *wildly* different now, no?

I doubt it. Apple under Cook is, if anything, even more cutthroat than it was under Jobs when it comes to exerting leverage on their suppliers. Jobs would suddenly change suppliers if he was disappointed. Under Cook Apple uses its enormous cash holdings to give zero-interest loans to suppliers to build out new capacity in order to get first dibs on new silicon. Turning down an Apple request can, literally, cost you seven figures.

That said, their influence in the mobile space is far larger than it is in the desktop/laptop space. Intel tried, for a decade, to lure Apple away from PPC, so I imagine Intel doesn't matter giving Apple early access. Given that Apple has pretty much zero footprint in PC gaming, I don't know if nVidia would feel the same kind of pressure to show favors to Apple as would someone like TMSC, especially that Apple is an all AMD company now. You could make the argument that nVidia could be induced to share pre-release silicon with Apple in order to, possibly, pry them away from AMD, but that's the kind of thing that, if it happens, we won't know about until it's a done deal.
 
  • Like
Reactions: Derived
I doubt it. Apple under Cook is, if anything, even more cutthroat than it was under Jobs when it comes to exerting leverage on their suppliers. Jobs would suddenly change suppliers if he was disappointed. Under Cook Apple uses its enormous cash holdings to give zero-interest loans to suppliers to build out new capacity in order to get first dibs on new silicon. Turning down an Apple request can, literally, cost you seven figures.

That said, their influence in the mobile space is far larger than it is in the desktop/laptop space. Intel tried, for a decade, to lure Apple away from PPC, so I imagine Intel doesn't matter giving Apple early access. Given that Apple has pretty much zero footprint in PC gaming, I don't know if nVidia would feel the same kind of pressure to show favors to Apple as would someone like TMSC, especially that Apple is an all AMD company now. You could make the argument that nVidia could be induced to share pre-release silicon with Apple in order to, possibly, pry them away from AMD, but that's the kind of thing that, if it happens, we won't know about until it's a done deal.

I guess the real question of late is...

Where is the evidence in the released products (not iOS based) that they are getting and using that early access to any advantage?

Other than 5k screens - everything they release is never really ahead of any hardware curve is it?

(honestly asking...perhaps people have some examples?)
 
I guess the real question of late is...

Where is the evidence in the released products (not iOS based) that they are getting and using that early access to any advantage?

Other than 5k screens - everything they release is never really ahead of any hardware curve is it?

(honestly asking...perhaps people have some examples?)

Another quick Google: https://www.macrumors.com/2013/07/2...-high-end-haswell-processors-for-macbook-pro/

I don't want to give the impression that Apple has all the newest toys before anyone else, as I just don't think that's true. However, I do think they have enough influence to demand, and get, early access, or special consideration, for chips they think will give them an advantage. Whether or not those could include next-gen nVidia GPUs is something I can't even guess at.
 
Another quick Google: https://www.macrumors.com/2013/07/2...-high-end-haswell-processors-for-macbook-pro/

I don't want to give the impression that Apple has all the newest toys before anyone else, as I just don't think that's true. However, I do think they have enough influence to demand, and get, early access, or special consideration, for chips they think will give them an advantage. Whether or not those could include next-gen nVidia GPUs is something I can't even guess at.

So did that result in getting something more amazing out to the public before others?
Or that wasn't offered to others?

I guess I just don't really "feel" that they end up ahead of anyone in any areas other than their own industrial design really (which is a pro and a con depending upon who you ask)
 
So did that result in getting something more amazing out to the public before others?
Or that wasn't offered to others?

I think I may have misunderstood your question, if only because the days of amazing advances in desktop hardware are over. CPUs continue to get slightly faster and more efficient from one generation to the next, but the only real question now is whether Zen will allow AMD claw back some of the IPC performance it's lost to Intel. On the GPU side, AMD's introduction of HBM was a good step, but it's an evolution, not a revolution, and we can make good educated guesses about what the next generation of GPUs will look like. I don't think any of them will make me trade in my GTX 980, though.

So, the fact that Apple's early access doesn't allow them to introduce something amazing before others is just a sign that truly amazing advances on the desktop are pretty rare.

That said, in a field where marginal gains are most of what you can count on, Apple being able to reap those marginal benefits before others is a advantage, even if it's something as small as 10% better performance. I see this in all the money they're throwing at battery tech, because a laptop which lasts 11 hours has a definite advantage over one which dies after nine.
 
Ahh, yes, that's part of the "Vapourware 5000" line, yes?

Tahiti went public at end of 2011, got announced by Apple as part of 2013 nMP in June of 2013, and finally shipped in Dec of 2013, a full 2 years after it launched for PCs. (And amazingly STILL SHIPPING TODAY in their "Flagship" product)

Are we really to believe that after epic foot dragging of that sort that Apple will introduce a GPU for 7,1 WHEN IT LAUNCHES FOR PC, or are you saying we can expect it 1-2 years after launch, per usual?

Don't you think it is likely that the nascent Hawaii and Fiji drivers we see in OSX already might be more likely as watered down versions of those cards? They fit much better into Apple's typical GPU scheduling.

This is exactly what I was trying to say with my previous post. People need to start getting their mindset into thinking like this otherwise they will be sorely disappointed.
 
Ok, assume Apple updates the Pro later this year. What next? Will there be three years before another update? I may be fine with longer upgrade cycles but then either 1) Apple need to make smaller updates in-between and/or 2) succesively lower the price. Like they used to to with the old Pro. As it is now the Pro is ridiculously over-priced. If buying a Pro in the beginning of its upgrade cycle the price may be fair, but not everyone is in sync with Apple's updates. Even though I don't pay for computers with my own money (I use it for work), currently it is very hard to motivate buying a Pro compared to e.g. a workstation from HP. Moreover, I would like to see nvidia options since CUDA is by far more stable and mature compared to OpenCL. For me the small form-factor is not a big problem, but I am not so sure it appeals to the target group. I think Apple is making a big misstake by not paying more attention to Pro users. It is a creative target group that can bring additional values to Apple that is not directly measured in sales of the Pro.
 
  • Like
Reactions: Hank Carter
... I would like to see nvidia options since CUDA is by far more stable and mature compared to OpenCL.

Agreed. I bought a Mac Pro 3013. The huge gripe I have is the AMD cards in it, more so now than when I bought it. Promises of OpenCL this and that never happened. It was all hype.

Updates (Bootcamp) are so slow and AMD drivers are about as solid as a pack of butter on a summer day. OpenCL is supported in a few apps like Blender (Cycles) but is nowhere near as fast as CUDA due to, I assume better drivers. My laptop with a GTX 980M renders 40-50% faster using Cuda GPU than my dual D700's and OpenCL which is exasperating.

I will not upgrade unless they go Nvidia. Or at least offer NVidia cards for us 2013 owners.

Apart from the GPU's the machine is gorgeous.
 
  • Like
Reactions: rdav
I wonder if the D300/D500/D700 naming convention was used to throw people off as to how old those GPUs are. All a marketing ploy.

My bet is firmly that any new Mac Pro released this year would be based on Fiji; down-clocked dual Fury Nanos on a non-reference board.

Late 2013 model launched with Intel Xeon Q3'13 CPUs, so reasonable to assume that Intel will provide the latest CPUs available.

6x Thunderbolt 3 could be a push because of bandwidth, but if they're just PCIe switches on the back it might not be; there just won't be enough to run many demanding things at once.

SSD would have to be an SM951/950 Pro

DDR4 RAM seems certain.
 
  • Like
Reactions: Xteec
The thing I wonder is where are all these PCIe lanes coming from? TB3/SSD/NIC's all sharing 8 lanes? If the PLX the heck out of it are they going to get TB3 bandwidth? Which leads to my next thought of what if cards aren't in the box and all external? Buy the base box and choose then choose your expansion box with Apple cards and fully rely on TB3 instead of stealing all the lanes for two cards.
 
Tahiti went public at end of 2011, got announced by Apple as part of 2013 nMP in June of 2013, and finally shipped in Dec of 2013, a full 2 years after it launched for PCs. (And amazingly STILL SHIPPING TODAY in their "Flagship" product)

Are we really to believe that after epic foot dragging of that sort that Apple will introduce a GPU for 7,1 WHEN IT LAUNCHES FOR PC, or are you saying we can expect it 1-2 years after launch, per usual?

Don't you think it is likely that the nascent Hawaii and Fiji drivers we see in OSX already might be more likely as watered down versions of those cards? They fit much better into Apple's typical GPU scheduling.

The workstation versions are on a longer release cycle and the Firepro series wasn't released until 6 months later. Considering its in a new form factor for the first time could take longer.
 
Last edited:
First hit on google: http://www.techspot.com/news/43676-apples-new-imac-gets-early-access-to-intels-z68-chipset.html

For me, the more interesting question is who gives Apple early access and who doesn't.

ASrock announced motherboards with the Z68 5 days before that story. And the official announcement was 5 days after.

Perhaps technically a scoop, but does a couple of days really matter?


Other than 5k screens - everything they release is never really ahead of any hardware curve is it?

(honestly asking...perhaps people have some examples?)

Dell was already [strike]shipping[/strike] showing the same 5K panel when the 5K Imac was announced.

That's creating a new top bin, not any change to the silicon.

The thing I wonder is where are all these PCIe lanes coming from? TB3/SSD/NIC's all sharing 8 lanes? If the PLX the heck out of it are they going to get TB3 bandwidth? Which leads to my next thought of what if cards aren't in the box and all external? Buy the base box and choose then choose your expansion box with Apple cards and fully rely on TB3 instead of stealing all the lanes for two cards.
The GbE, SSD, WiFi and USB 3.0 share the 8 PCIe 2.0 lanes that come out of the PCH (southbridge).

The PLX switch multiplexes the 8 PCIe 3.0 lanes coming from the CPU package into 16 PCIe 2.0 (full bandwidth) lanes. 12 of those lanes feed the three T-Bolt 3 controllers.

There are more than 8 lanes left over "for everything else", but that seems to be ignored.
 
Last edited:
You sure?

http://www.hitc.com/en-gb/2014/11/06/dells-5k-ultrasharp-monitor-gets-release-date/

"Dell's monitor will attempt to rival Apple's newly announced next generation iMac with Retina 5K Display."

Apple released their 5K iMac three months before Dell. Seems something got switched. Ok, post updated.

There are more than 8 lanes left over "for everything else", but that seems to be ignored.

Technically, 8 PCIe 3.0 lanes left over. Obviously, any lanes branching off of that are lower bandwidth. At 16 PCIe 2.0, which you already mentioned.
 
Last edited:
ASrock announced motherboards with the Z68 5 days before that story. And the official announcement was 5 days after.

Perhaps technically a scoop, but does a couple of days really matter?




Dell was already [strike]shipping[/strike] showing the same 5K panel when the 5K Imac was announced.


That's creating a new top bin, not any change to the silicon.


The GbE, SSD, WiFi and USB 3.0 share the 8 PCIe 2.0 lanes that come out of the PCH (southbridge).

The PLX switch multiplexes the 8 PCIe 3.0 lanes coming from the CPU package into 16 PCIe 2.0 (full bandwidth) lanes. 12 of those lanes feed the three T-Bolt 3 controllers.

There are more than 8 lanes left over "for everything else", but that seems to be ignored.
Are those 12 pcie 2 lanes enough to feed 6@tb3
 
Status
Not open for further replies.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.