Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

deconstruct60

macrumors G5
Mar 10, 2009
12,493
4,053
Hehe, I was talking about Mainstream i3, i5, i7. And that is because Zen will be offered in 6 and 8 core versions ;).

If look at AMD's 2015 slides here you'll see that Zen was to initially drop in a "FX" class
http://www.fudzilla.com/news/40627-first-amd-zen-chips-may-not-be-quad-core-parts
( second slide above).

I suspect that this (from article linked above )
".. t appears AMD will be initially focused on bringing back some high-end desktop (HEDT) market share from Intel’s stagnant performance numbers. ..."

Only not really aiming at these 6-8 core models more so at the 6700 (capped at 4) and the kneecapped 6 core model at the bottom (e.g. new i7 6800K capped at 28 PCIe lanes) .

They are going to save effort/expensive by reusing the APU socket AM4, but I suspect the addition 4 cores are going to come at expense of the iGPU.

If look at the floor plan of Bristol Ridge ( which is AM4 socket solution )
Bristol%20Ridge%20Tech_FINAL-page-004.jpg


http://www.anandtech.com/show/10362/amd-7th-generation-apu-bristol-ridge-stoney-ridge-for-notebooks
Then can get to double the amount of x86 cores by chopping the GPU in half and consuming one of those halves. If mostly aiming at the HEDT market then there likely will be 1-2 discrete GPUs in the standard PCI socket(s) anyway. The iGPU would be either be for computational only for main RAM based modeling or lightweight VGA/DVI socket duties. Putting hopes on DX12/Mantle like workloads to drive into more threads which the additional cores may help with while primarily focused on gaming for which the limited bandwidth to the GPU cards is tip-toed around in the normal desktop market. ( x16 sliced into two x8 backstopped by SLI/Crossfire is pretty standard stuff).


The initial Zen coming in at higher than mainstream Core i price point but under the Extrme/E5 1600 levels. AMD has to move their average selling price higher to survive. The ones which the much higher GPU budget likely will have just one of the 4 core "slices". (Hence in the same boat as Intel's mainstream offerings).


From an Apple perspective though .... a diminished iGPU isn't going to cut it. In the iMacs they aren't looking for a dGPU in most of the line up. In the future mini's even more so. For the Mac Pro this "mainstream" I/O bandwidth isn't going to work.

If Zen is decent then probably will get some movement of Intel to move the E5 1620-30 to fully 6 cores and non throttled clock speeds in next generation. ( the pricing might come back a bit or at very least not go up again. as it appears to have done for v4. ). At best AMD can stop the pricing climb . If a future iteration of Zen does even more better may see some reversal, but neither AMD nor Intel wants to see a "race to the bottom" on x86 package pricing.


Of course, I do not contradict anything what you have written about technology behind this. This is in addition to what you have written. But without Zen being good, Intel would not even think about putting 8 or even 6 cores on mainstream CPUs. You know, money from customers smells really good ;).

If AMD can bring 6-8 and outpace Intel on iGPU all on approximately the same transistor budget, then there is a threat. IF AMD throws the iGPU under the bus to get to 6-8 cores then there is only a highly limited threat here. Highly doubtful Intel will panic if that is the case. Much of weakness in the gap bettween 6700K and 6800K is Intel's marketing kneecapping the current designs to enhance market segmentation. It should be easy enough to just take that artificial throttle off.
 
Last edited:

Melodist

macrumors regular
Sep 30, 2015
150
90
Well, it looks like we won't be seeing an update to the Pro lineup, other than the opposite of pro. Uhhhh I'm so excited for the new Apple Watch, I really need this in my life right now oO.
 
  • Like
Reactions: pat500000

Dean Yu

macrumors regular
Mar 12, 2016
162
134
Toronto, Canada
Bristol%20Ridge%20Tech_FINAL-page-004.jpg


http://www.anandtech.com/show/10362/amd-7th-generation-apu-bristol-ridge-stoney-ridge-for-notebooks
Then can get to double the amount of x86 cores by chopping the GPU in half and consuming one of those halves. If mostly aiming at the HEDT market then there likely will be 1-2 discrete GPUs in the standard PCI socket(s) anyway. The iGPU would be either be for computational only for main RAM based modeling or lightweight VGA/DVI socket duties. Putting hopes on DX12/Mantle like workloads to drive into more threads which the additional cores may help with while primarily focused on gaming for which the limited bandwidth to the GPU cards is tip-toed around in the normal desktop market. ( x16 sliced into two x8 backstopped by SLI/Crossfire is pretty standard stuff).
 

Attachments

  • image.jpg
    image.jpg
    44.5 KB · Views: 114

Mago

macrumors 68030
Original poster
Aug 16, 2011
2,789
912
Beyond the Thunderdome
I suspect that this (from article linked above )
".. t appears AMD will be initially focused on bringing back some high-end desktop (HEDT) market share from Intel’s stagnant performance numbers. ..."

AMD plans is to re-build most their CPU family around the new Zen Core, from 4 to 32 cores also Mobile versions, re-cycling the same core as ARM vendors does when launch a 8 core cortex a53 or a 2 core cortex a53 its the same core, but teamed on a silicon fabric, AMD follows a similar strategy (intel too on its Xeon-D/Xeon Phy families), the good it's a consistent CPU core across the family, the bad is the same functions on laptops or servers.

Intel wont promise to increase IPC but cost efficiency, if you check AMD Zen core organization you'll see few interesting things which could give a bit better IPC than Skylake, even KabyLake.

but IPC and performance its an tricky questions, Intel peak performance an be achieved thru Vector instructions, not a general purpose set precisely, sure AMD Zen will suffer of such specialization and maybe better than Intel for some algorithms and worse for others.

Until we see replicable practical benchmarks, both AMD and Intel IPC are Vapor.
 

koyoot

macrumors 603
Jun 5, 2012
5,939
1,853
Intel wont promise to increase IPC but cost efficiency, if you check AMD Zen core organization you'll see few interesting things which could give a bit better IPC than Skylake, even KabyLake.
No chances for this.
 

rockyromero

macrumors 6502
Jul 11, 2015
468
147
Spending this kind of money on a Mac Pro is not justifiable if we have to wait until 2017 for stability if the mac pro 2016 is birthed during the next MacOS cycle in the fall.

Of course, it would run both OSs.

All MACs are compatible up & down.

Having bugs is a way of life in the technology world.

 

pat500000

Suspended
Jun 3, 2015
8,523
7,515
Well, it looks like we won't be seeing an update to the Pro lineup, other than the opposite of pro. Uhhhh I'm so excited for the new Apple Watch, I really need this in my life right now oO.
I need a violin.....
 

deconstruct60

macrumors G5
Mar 10, 2009
12,493
4,053
... die image ...

Oh yeah, forgot about those die shots which were actually discussed early in the thread. (http://www.guru3d.com/news-story/amd-zen-summit-ridge-die-says-hello.html ). There plenty of room for 8 cores if toss the whole GPU completely. ;) Somewhat puzzling though how AMD expects to get max performance out of both the "no GPU, high end" and "mid-low" APUs using exactly the same socket. Eight x86 cores sharing 2 memory channels is a bit different dynamic than 2-4 cores. 4:1 ratio versus 1:1 or 2:1 ratio.

Would likely incrementally "brute force" past Intel's 4 core, 2 channel offerings on highly concurrent stuff that, that happens to not be so memory throughput intensive.
[doublepost=1465569701][/doublepost]
...
Intel wont promise to increase IPC but cost efficiency, if you check AMD Zen core organization you'll see few interesting things which could give a bit better IPC than Skylake, even KabyLake. ...

Doubtful. Zen gets rid of sharing the FPU so that will be a big jump on any code that has average (or more) usage of FP instructions, but that isn't going to get them any closer to Intel (let alone jump out front). AMD has some pretty dubious things to unwind. That is where most of their "large" jump is going to come from.


One, AMD is touting number of 40% increase over their previous generation. When their current generation is 60-100% behind, that 40% makes them more competitive; not "better than". Coupled to performance/price they might pull ahead or perhaps base clock higher ( DEC Alpha style ), but 'raw' throughput per clock ... probably not? ( just based on AMD statements and the current reality. )

Intel is throwing hardware at IPC.

24%20-%20OnTheFly_575px.jpg

http://www.anandtech.com/show/9582/intel-skylake-mobile-desktop-launch-architecture-analysis/5

You can't increase IPC past what is inherit in the binary code. There are code fragments that just don't have whole lot to exploit. The "drop off" by Intel per generation is primarily because all the easy 'low hanging fruit' is gone. ( all the more so if have code laid out by an optimizer designed for a 4-6 year old implementation of the architecture; i.e. running lowest common denominator optimized code. ). AMD can catch up over time more easily because x86 code is increasing a factor holding Intel back from maintaining a relatively large lead. ( and for the ARM arm flappers ... ARM code isn't particularly any better at latent IPC code constraints. Much of ARM's large jumps of late are due to actually allocating transistors to solutions researched 5-15 years ago. )


Until we see replicable practical benchmarks, both AMD and Intel IPC are Vapor.

The relative positions versus one another are vapor... but IPC is probably one of the design points tested by simulations intensively by both sides. More likely highly classified than vapor. They are also extremely likely not to change much once have taped out the design. You may fix some glitches but design is largely over; it is "remove defects" time.
 
Last edited:

beaker7

Cancelled
Mar 16, 2009
920
5,010
I'm using image sequences

What type? Yes, it makes a difference.

Not to mention that there's no actual effective way of figuring out what is slowing your render times

Of course there is. Start turning plugins off one at a time until it speeds up, or try a different file format. Look at your system usage to see if its a hardware bottleneck. These are pretty standard practices for workflow development in a professional environment.

*And* that Adobe actually *removed* multicore rendering from AE2015

Actually they removed the tick box allowing you to specify single-core rendering, or to define an arbitrary amount of cores to use. Most people set that incorrectly in prior releases due to not doing their workflow development homework per above.
 

pat500000

Suspended
Jun 3, 2015
8,523
7,515
Polaris info disappeared from AMD's website. Hum..
http://www.amd.com/en-us/innovations/software-technologies/radeon-polaris

you mean this?
[doublepost=1465590739][/doublepost]
I wonder if BT5 will be on the nMP?
Guess there won't be available parts in time, unless no new hardware needed, but I doubt it.
Looks to be aimed at phones and laptops mostly though.

http://wccftech.com/bluetooth-5-promises-range-groundbreaking-speed-announced-week/
Probably not..sounds like Apple is having trouble keeping up with tech.
 

ManuelGomes

macrumors 68000
Dec 4, 2014
1,617
354
Aveiro, Portugal
Cool looking concept of the OLED bar in rMBP by Hajek.
One thing is odd about this OLED thing - the Power button. When the Mac is off, how the hell do you power it on? Unless the OLED panel is always in stand-by.
Maybe it also comes to the wireless keyboard!!
[doublepost=1465590982][/doublepost]pat, yeah but I meant it disappeared from the front page. APU is still there.

Apple likes to use the tech when it's quite mature I guess. Let other take care of the initial quirks. Still, it's bad to see late tech adoption, specially if it's something they started out or championed. Like TB3.
 

Ph.D.

macrumors 6502a
Jul 8, 2014
553
479
Polaris info disappeared from AMD's website. Hum..

I do see Polaris and the RX 480 discussed on their site, but not prominently.

Although I still believe a 7,1 is coming, I've come to doubt that it will be announced on Monday. Now I'm thinking September. What's another 2 or 3 months? :(

Oh well, my 3,1 may very well last another 8 years, unless I tire of the wait and replace it with a Linux box.
 

pat500000

Suspended
Jun 3, 2015
8,523
7,515
I do see Polaris and the RX 480 discussed on their site, but not prominently.

Although I still believe a 7,1 is coming, I've come to doubt that it will be announced on Monday. Now I'm thinking September. What's another 2 or 3 months? :(

Oh well, my 3,1 may very well last another 8 years, unless I tire of the wait and replace it with a Linux box.
Maybe more like 2017.
 

ManuelGomes

macrumors 68000
Dec 4, 2014
1,617
354
Aveiro, Portugal
I'd bet AMD will adjust the clocks to have RX480 closer to the 1070.
And maybe the full fat RX480X or RX490 or whatever will match it or come even closer.
They are not competing directly but I believe they'll want to get as close as they can.
 

pat500000

Suspended
Jun 3, 2015
8,523
7,515
Doesn't appear like hardware is gonna be announced this summer. Time to move on to "what can I do to improve my current Mac Pro" subject....or fantasy upgrade.
 

AidenShaw

macrumors P6
Feb 8, 2003
18,667
4,677
The Peninsula
I'd bet AMD will adjust the clocks to have RX480 closer to the 1070.
And maybe the full fat RX480X or RX490 or whatever will match it or come even closer.
They are not competing directly but I believe they'll want to get as close as they can.
But then Nvidia will move the goalposts when they announce the "ti" and "Titan" versions of Pascal ;)
 

Ph.D.

macrumors 6502a
Jul 8, 2014
553
479
I'm certain they have other designs in the works, and I doubt they will stretch the 480 much further right now. They have stated their intent to dribble them out starting from the lower end (priced well below the 1070, which is fine). They have also stated their intent to target proprietary platforms. This could include custom configurations for Apple. Or, for all we know, they may even be waiting for Apple before announcing other products.
 

ManuelGomes

macrumors 68000
Dec 4, 2014
1,617
354
Aveiro, Portugal
Of course, Ti and Titan would always come. And those will be GP102 based. and to counter those AMD will have Vega I figure.
But indeed AMD stated that they're not chasing the high end right now, and it's the best they can do really. If they gain market share (and they are apparently) in the mainstream that might provide for their survival, and that is good for everyone.
I hope they get it right this time, for their sake and everyone's.
It could be the first time I ever get something AMD, with the nMP. I've always been on the green team.
Hope I don't regret it.
 

koyoot

macrumors 603
Jun 5, 2012
5,939
1,853
I'd bet AMD will adjust the clocks to have RX480 closer to the 1070.
And maybe the full fat RX480X or RX490 or whatever will match it or come even closer.
They are not competing directly but I believe they'll want to get as close as they can.
It appears that 67DF:C7 model which is in the RX 480 has 1266 MHz core clock. Which gives 5.833 TFLOPs of compute power.

2560 GCN core version of this GPU, if exists, with this core clock will have 6.5 TFLOPs of compute power.

@Ph.D . There is strong rumor that AMD did not revealed much about their cards to this point because... they were NDA'd by...

And the speculation can begin right now.

P.S. I still would prefer Fiji or GP104 in Mac Pro because it would bring more compute power.
 
Last edited:
  • Like
Reactions: pat500000

pat500000

Suspended
Jun 3, 2015
8,523
7,515
It appears that 67DF:C7 model which is in the RX 480 has 1266 MHz core clock. Which gives 5.833 TFLOPs of compute power.

2560 GCN core version of this GPU if exists with this core clock will have 6.5 TFLOPs of compute power.

@Ph.D . There is strong rumor that AMD did not revealed much about their cards to this point because... they were NDA'd by...

And the speculation can begin right now.

P.S. I still would prefer Fiji or GP104 in Mac Pro because it would bring more compute power.
Damn the nda. Lol
 

Ph.D.

macrumors 6502a
Jul 8, 2014
553
479
@Ph.D . There is strong rumor that AMD did not revealed much about their cards to this point because... they were NDA'd by...

If Global Foundries - and hence AMD - is having 14nm startup or yield problems, then I could see AMD's whole production of higher-end chips going to a customer like Apple for at least a short while. So yeah...
 

koyoot

macrumors 603
Jun 5, 2012
5,939
1,853
If Global Foundries - and hence AMD - is having 14nm startup or yield problems, then I could see AMD's whole production of higher-end chips going to a customer like Apple for at least a short while. So yeah...
Looks like they don't. Guy from ONE of the GloFo fabs on reddit said that from their only one fab they provided AMD with 177 THOUSAND dies(177 000). And AMD has 2 fabs for this from GloFo, and one from Samsung.

That is why they have been silent. All of past 7 months GloFo was producing AMD GPUs to feed the demand for them.

And the demand will be huge, by the words in silicon industry(MSI, Sapphire, etc...).
 

Ph.D.

macrumors 6502a
Jul 8, 2014
553
479
Looks like they don't. Guy from ONE of the GloFo fabs on reddit said that from their only one fab they provided AMD with 177 THOUSAND dies of Polaris 10(177 000). And AMD has 2 fabs for this from GloFo, and one from Samsung.

Is there any idea how many Mac Pro's Apple has been selling? Anyway, I'd say 480's could be a good chip for new iMacs (or is it still too hot?), which might use all those and more.

OK, a quick calculation: Assuming the wafers are 300mm, and each die is 250 square mm (pretty big), we're looking at roughly 250-300 wafers. Not staggering, but significant.
 
Last edited:
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.