Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Hmmmm. AMD hires away top Apple execs. Apple hires AMD GPU engineers in tampa..

AMD earns major design win for 7970/Firepro architecture in next-gen mac pro, not to forget xbox 1, PS4 design wins.

Haswell MacBooks set to launch sans Nvidia architecture. I'd place $$$ on a AMD variant in the next iMac. Both cuts could pose a good hit to Nvidia's bottom line and ties them even closer to Tegra and it's Android offerings..
 
Hmmmm. AMD hires away top Apple execs. Apple hires AMD GPU engineers in tampa..

AMD earns major design win for 7970/Firepro architecture in next-gen mac pro, [...]
Haswell MacBooks set to launch sans Nvidia architecture.
Hey - let's make a rumor out of this: Apple planning to take over AMD! :p:D
 
Perhaps I don't understand the limitations as well as you all, but couldn't they mean that NVIDIA isn't in the base configuration?

The limitations is that Apple does the embedded GPU work. Apple doesn't generally do redundant work. You don't see MBP with Nvidia and AMD components in the same generation. Likewise iMac. Each generation is a design win bake-off for AMD and Nvidia. Who ever wins gets that year's Mac model. The loser comes back and tries again next year.

The Mac Pro is in the same boat at this point. Pragmatically the GPU "cards" are far more so just daughtercards to the one logical motherboard present. Conceptually it is just like all the other Macs at this point were Apple does all the work of embedding the GPU(s) into the design.

This round Nvidia lost (if they even competed ). Just like AMD lost out on the MBP/iMac revisions in 2012. ( and probably will in 2013 also since neither AMD nor Nvidia has anything substantively new to offer; just clock speed bumps. )

Most likely Apple has done W7000 , W8000, and W9000 equivalent "cards" for the Mac Pro 2013 to flush out a range of performance and price points for the Mac Pro to cover. It is quite unlikely they would have don't another 2-3 cards for Nvidia to cover about the exact same set of price points and very similar performance.

My guess that Nvidia might get interested in winning if these Mac Pro sell in substantive numbers relative to their Pro market (i.e, AMD raises their Pro market share substantively). Otherwise not. The much larger graphics control is to win the laptop and iMac design bake-off. As long as they keep winning that one they won't worry much about the Mac Pro. If AMD keeps loosing the laptop/iMac bake-offs then at least the Mac Pro helps them keep a foot in the door.

----------

High end pro graphic cards like the Quadro are a specialized niche market, especially the top $$$$ model. Why should they start a whole new production line for a tiny subset of a niche market.

Who is "they"? The new Mac Pro's card is neither AMD's or Nvidia's. It is a card that Apple produces ( OK technically that contract someone to make it and probably modified a reference design, but its Apple's design work. ).

It is far more a question of whether Nvidia and AMD want to allow Apple to make cards to sell into the "Pro" market that uses Macs or not. Not whether they (GPU vendors) make the card. Apple doesn't need any of their help in getting a graphics system built. Apple's corporate contract manufacturing budget is about as big then their respective graphics businesses. This is a question of whether AMD and Nvidia want to be component suppliers or not.
 
Hmmmm. AMD hires away top Apple execs. Apple hires AMD GPU engineers in tampa..

It is Orlando.

https://www.macrumors.com/2013/05/2...es-for-an-orlando-gpu-team-still-hiring-more/

And it isn't clear if this is Mac related or not. If look at that article one of the qualifications was

".. 5+ years proven experience building/hiring medium/large (30+) complex ASIC IP development teams ..."

Intellectual property (IP) is more so designs than finished product. The iOS SoC CPU/GPU packages have GPU IP in them. This can be more so highly tweaking/customizing the Imagination GPUs that Apple gets ..... or getting off of Imagination GPUs... ( not sure whether their recent jump into MIP as an ARM competititor has them looking at other options (e.g., tweaking ARM's Mali faster) or not. )

What the Mac systems needs is not particularly ASICs but drivers (which they did get someone with that kind of background from AMD Orlando) and work incorporating ASICs (not making them). I suspect the work is a bit split with perhaps small numbers of folks doing Mac but most are in SoC work. Basically the move is to make the group that was in Cupertino bigger by hiring talent where they happen to live (Orlando) rather than trying to drag them to CA ( and higher cost of living. ) .



AMD earns major design win for 7970/Firepro architecture in next-gen mac pro, not to forget xbox 1, PS4 design wins.

AMD has been willing to license to 3rd parties at deeprer level than Nvidia. That is in part because they are playing "catch" up in terms of revenues and growth.




Haswell MacBooks set to launch sans Nvidia architecture.

There are a bunch of 2012 Ivy Bridge MacBooks that launched sans Nvidia Architecture.

MBA 11" sans Nvidia.
MBA 13" sans Nvidia
MBP 13" sans Nvidia
rMBP 13" sans Nvidia
MBP 15 Nvidia
rMBP 15" Nvidia

2/3's of the whole line up doesn't have them. That same thing is probably true of the 2013 Macbook/laptop line up. The fact is *MOST* of the Mac laptop line up needs better integrated graphics. That has nothing to do with dropping Nvidia. Nvidia isn't even a player in integrated x86 graphics anymore. And AMD is only a player is can dislodge Intel also on the x86 core.

The whole "classic" MBP 15" product might get tossed in 2013. That also has little to do with Nvidia or not. That would bump the percentage up to 4/5's of the laptop line up. There is no "good, better , best" offering of the 15" anymore. It is one model and all BTO for variation. Last Apple stores visited there is only one demo model in the store like the Mac Pro. It is probably on a slippery slope.


The key issue for Nvidia is really whether the MBP 15" survives at all or not. Apple may fall back to a entry level MBP 15" that has integrated graphics only ( 2009 9400M only entry model ) to push the price down a bit.
The other blow will be the entry iMac if it dumps a dGPU; again to lower price a bit.

It is probably a generation too soon to dump the dGPU from the rMBP 15". I know there are claims Apple is dumping now rather than later, but we'll see later in Fall.

the MBP 13" and rMBP 13" are the two models most in need of the Haswell IrisPro. The combination of those two is also a large chunk of the Mac laptop volume. Nvidia is a non player there now just as much as later this year in that block of Mac laptop sales.

The only real question is wither Apple is going to kneecap the classic MBP 13" (which still reportedly the best selling laptop model ) without a Haswell IrisPro update and use that to push the rMBP 13" in front on volume.


I'd place $$$ on a AMD variant in the next iMac.

Probably not. The same reasons why AMD lost last year's design win for iMac are still in place. Both AMD and Nvidia updates for 2013 are just clock speed bumps for their 2012 models. $ , performance, and drivers that won last year likely in place for this year too. 2014 the door is wide open. It depends upon AMD's and Nvidia's tweaks and pricing they exposed to Apple in NDAs this year.


Both cuts could pose a good hit to Nvidia's bottom line and ties them even closer to Tegra and it's Android offerings..

Apple's cuts in and of themselves? Not really. The overall PC market going stagnant and dropping in some areas is a much bigger hit to Nvidia's bottom line.

There are alot more than just Apple's systems that Nvidia is going to get squeezed out of by the IrisPro. But that is a trend that has already been in place for last 2-3 years.

Nvidia will need to lean harder on CPU-GPU combos that are possible for them in the ARM SoC space. But Qualcomm is no slouch there either. ATI's old mobile division (which AMD didn't get ) is helping Qualcomm take a large chunk of the ARM market. ( e.g., new Nexus 7 dumped Tegra and has substantially better graphics than last year. )

Nvidia can print money for another couple of years with Pro cards and trapped in tarpit CUDA library software. But that is a short term crutch.
 
Probably not. The same reasons why AMD lost last year's design win for iMac are still in place. Both AMD and Nvidia updates for 2013 are just clock speed bumps for their 2012 models.

You've forgotten GK110 again.

Nvidia brought out game changing architecture that truly leaves AMD in the dust.

Unfortunately Apple forgot it too.
 
You've forgotten GK110 again.

For the iMac? Not hardly.

You are flapping your arms again and if anything the out of context mobile comparision I mistakenly made before is all the more relevant here. And frankly, the GK110 was introduced in 2012. Rebrands and configuration reintroductions isn't new and isn't going to change the likely parameters of a design win.


Nvidia brought out game changing architecture that truly leaves AMD in the dust.

It is late (K6000 just launching later ). That is likely a yet another contributing reason why lost the design win. I doubt Nvidia offered reasonable pricing also. For better or worse the HPC market soaking up most of the GK110 product actually makes it not a candidate for the Mac Pro, because it delays the Pro market display card. The Mac Pro will be attached to a display, so that is a critical factor.


Unfortunately Apple forgot it too.

The are multiple dimensions to getting a design bake-off win. Cost and being able to deliver on time are also factors. On time means meeting the other design time constraints that have to be meet because this is in embedded card. ( K6000 arriving in Fall and Mac Pro arriving in Fall means it is too late to redesign off the reference design to meet the Mac Pro's deadline. )
 
Last edited:
For the iMac? Not hardly.

You are flapping your arms again and if anything the out of context mobile comparision I mistakenly made before is all the more relevant here. And frankly, the GK110 was introduced in 2012. Rebrands and configuration reintroductions isn't new and isn't going to change the likely parameters of a design win.




It is late (K6000 just launching later ). That is likely a yet another contributing reason why lost the design win. I doubt Nvidia offered reasonable pricing also. For better or worse the HPC market soaking up most of the GK110 product actually makes it not a candidate for the Mac Pro, because it delays the Pro market display card. The Mac Pro will be attached to a display, so that is a critical factor.




The are multiple dimensions to getting a design bake-off win. Cost and being able to deliver on time are also factors. On time means meeting the other design time constraints that have to be meet because this is in embedded card. ( K6000 arriving in Fall and Mac Pro arriving in Fall means it is too late to redesign off the reference design to meet the Mac Pro's deadline. )

Just nice hearing you admit you were wrong.

Sign of good character.

Someone should start a poll, who would have preferred a K6000 over this FirePro thingy.

Would be interesting.
 
Do you realize that a K6000 will be priced in the range of 5000/7000$ alone? Probably only very few people are going to buy that card, the vast majority of users(where the money come from) are going to buy "cheaper" pro cards just because that's all they need.
Only time will tell if apple have made a good decision to choose amd instead nvidia.
In the meanwhile I remember very well lots of years ago when apple switch to intel processors, lots of whiners where furious because they choose intel(xeon) over amd(opteron).
 
Do you realize that a K6000 will be priced in the range of 5000/7000$ alone? Probably only very few people are going to buy that card, the vast majority of users(where the money come from) are going to buy "cheaper" pro cards just because that's all they need.
Only time will tell if apple have made a good decision to choose amd instead nvidia.
In the meanwhile I remember very well lots of years ago when apple switch to intel processors, lots of whiners where furious because they choose intel(xeon) over amd(opteron).

I doubt very much that it will be that much.

THat would mean a Quadro 6000 that was $3K now is going to be replaced by a K6000 for double that? Not likely.

And has been pointed out, there is no way Apple is going to pass $3,300 x 2 for the FirePro along either. They buy the bare GPU chip and have their folks slap the board together.
 
Unfortunately they are some much better then anything AMD has to offer they can set their own price.

But I still doubt it will be that high.

And even if it is, there will be discounts from day one. I got a Mac K5000 on release day for under retail.

But blame AMD for coming up with lackluster competition. Nvidia was able to hold their best cards close to vest for months because AMD had NOTHING and couldn't bluff. Nvidia was able to field GK104 to counter 7970 precisely BECAUSE it was lackluster.

GK110 is a game changer that AMD has nothing to counter. Hence $1K for a Titan. And hence the fire sale FirePro prices that got them in new consoles and nMP.
 
....
Someone should start a poll, who would have preferred a K6000 over this FirePro thingy.

Would be interesting.

Not particularly interesting. It is just as likely to attract AMD/ATI vs. Nvidia fan boy noise as to highlight anything particularly new or informative.
Without prices on either one much of the discussion is going to be spec porn driven as opposed to anything connected to utility value.


But blame AMD for coming up with lackluster competition. Nvidia was able to hold their best cards close to vest for months because AMD had NOTHING and couldn't bluff. Nvidia was able to field GK104 to counter 7970 precisely BECAUSE it was lackluster.

7970 up against a GK104 in double precision isn't exactly lackluster.
Nvidia did two implementations of the same general architecture this round. GK104 and GK110. Both are aimed at different markets.

AMD did one and is far more aiming at trying to tag both markets with one implementation of the general architecture. At the edges AMD's approach doesn't fit as well. Toward the middle it does. Given AMD's track record in 2010-2011 with OpenCL drivers it is doubtful a highly skewed HPC market implementation would have done them much good 2012-2013. Especially given TSMC's quirky fab track record early in a specific process level lifecycle.


The relevant question would be where is the bulk of the Mac Pro GPUs going to be. At the upper bleeding edge or toward the middle of the GPU performance spectrum?


GK110 is a game changer that AMD has nothing to counter. Hence $1K for a Titan. And hence the fire sale FirePro prices that got them in new consoles and nMP.

FirePro isn't in consoles. "Fire sale" versus "reasonable value pricing" is debatable for the Apple FirePro cards ( assuming they are reasonably priced). The Nvidia solution being substantially more expensive isn't necessarily a big plus. Especially when there are two GPUs.

AMD has nothing to counter with now, but then again NVidia isn't even shipping the K6000. The rumor mill (and leaked AMD slides ) for last couple of months have pointed to AMD's shipping their next generation (Volcanic Islands) by end of 2013. ( Of course their fab being primarily TSMC can take that with a grain of salt of the 20nm process getting to decent yields on time. Nvidia using the same folks means smaller implementation probably goes to market first. ) AMD and Nvidia leapfrog each other at times.

I don't think AMD's 'FirePro' version is coming in 2013, but it up in the air whether Nvidia will revise their Pro card line up before AMD revises theirs. Throw into the mix of timing for the new E5 v3 (Haswell) and it is a toss up as to which one wins the next Mac Pro design bake off. If the E5 v3 slides to early 2015 then it is even money. If the E5 v3 makes it out in Fall 2014 then scales might tip to AMD again.

Just gazing at spec porn isn't 100% of why parts win in design bake offs.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.