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

cube

Suspended
May 10, 2004
17,011
4,973
The $250-$300 price rumor tells me that it won't match the performance of Broadwell-E. If Zen really did match Intel, they would price it higher. From what I understand, this is supposed to be an enthusiast part, which normally commands prices > $300.
If people are used to paying $150 for FX, asking for $250 is already a lot, especially if you just bought an AM3+ motherboard.

They would also be trying to convince APU customers to buy a graphics card.
 

koyoot

macrumors 603
Original poster
Jun 5, 2012
5,939
1,853
220$ Price tag is for 4 core/8 Thread CPUs. 8 core/8 thread can cost around 300$ to counter quad core Intel i7's. 8 core/16 thread will most likely cost around 399-499$, and Unlocked, 125-140W version of the same CPU around 699$.

Thinking that AMD will price 8 core/16 thread CPU on around 300$ price tag if they offer same performance as HEDT 8 core Haswell CPU is delusional.
 

cube

Suspended
May 10, 2004
17,011
4,973
As someone who is still running a Phenom II X6 and just bought an AM3+ motherboard, I will only buy an unlocked 8/16 Zen and would not pay $400.
 

koyoot

macrumors 603
Original poster
Jun 5, 2012
5,939
1,853
The rumored 3.15 GHz base/3.6 GHz Boost, 8 core/16 thread, 95W CPU will most likely cost 499$.

Unless AMD in last minute will pull something out of their hats and improve the core clocks even more.

From what rumors say, the 4 core/8 Thread CPU will also get unlocked version.

My information from few weeks ago was that Zen 8 core/16 thread unlocked version would have 125W TDP and 3.5 GHz base/3.9 Boost clock. Now they say that it can be at 4GHz base@140W TDP.

If this is true, it appears that AMD may have made some progress with the process. But If I would bet anything on unlocked version, it would be 3.5 GHz base@125W TDP.
 

cube

Suspended
May 10, 2004
17,011
4,973
If they could cheaply produce a 22FDX 8-core Excavator AM3+ FX, I would be quite happy buying it.
 

deconstruct60

macrumors G5
Mar 10, 2009
12,493
4,053
Yes they would because it's a cross licensing agreement. Intel licences their x86 IP and AMD licences their x64. The agreement on IP sharing is void if either company changes hands it's one of the big reasons why AMD that has been struggling for years hasn't been bought.

Cross licensing on what? AMD had (has ) a basic, classic x86 license. Intel handed a couple of those out long ago. The cross licensing is probably written for what goes on top ( augmentations of x86). Some AMD additions and some Intel additions with each track each other on the stuff layered on top. That is cross licensed. The old , "grandfather", license could easily still have the same terms. There is also likely more clear 3rd party foundry licensing terminology ( since both Intel and AMD are doing more of that now. AMD solely doing it. Intel just dipping baby toe in. )

So Intel can use x86_64 and AMD gets all of he AVX stuff. Virtualization stuff swapped both ways as needed. The goal is to keep up a dual source platform.

The layered on top licensing had not reason to put a change of ownership clause into it. AMD gets sold and Intel still has rights to the layered on top stuff. They don't need a grandfather license because they own it. It is only a poison pill for one side. There is absolutely ZERO rational reason why Intel would sign an agreement that crushed themselves. AMD didn't really have choice with their relatively weak position technologically ( an layered on top augment to someone else's core technology. ) and financially ( so 'sell' x86_64 to get access to AVX et. al. ). There is nothing in the law that cross licensing agreements have to be 100% bilateral. They very often are not (e.g., one side having to make payments while the other side doesn't. Terms that are limited in one side with respect to 3rd parties, etc. )
[doublepost=1479501373][/doublepost]
x86 is key to computing. The world is locked-in.

No, it isn't. Smartphones computers x86 is hardly present. Really big iron boxes . No lock-in. You are trying to narrow down computing just to classic PC form factors. That isn't computing. Windows isn't.

It is likely trying to nail down the Mac market subset to being a market and then applying monopoly rules to it. It is a submarket of a larger market.


P.S. Note that AMD even with serve cash limitations is still pursuing ARM based CPUs. Both indicative that computing isn't solely x86 based and that they do need a fall back CPU architecture if did get bought out.
 
Last edited:

cube

Suspended
May 10, 2004
17,011
4,973
No, it isn't. Smartphones computers x86 is hardly present. Really big iron boxes . No lock-in. You are trying to narrow down computing just to classic PC form factors. That isn't computing. Windows isn't.

It is likely trying to nail down the Mac market subset to being a market and then applying monopoly rules to it. It is a submarket of a larger market.
I am not trying to narrow down anything. The personal computer market is key and is locked-in to x86.

Note that I did not even mention the small server market, because in this case it would be easier to replace x86 with some mythical powerful ARM.
 

deconstruct60

macrumors G5
Mar 10, 2009
12,493
4,053
I am not trying to narrow down anything. The personal computer market is key and is locked-in to x86.

The Personal computer ( a computer for personal use) market isn't simply classic PC form factors anymore.
That is exactly the position the lawyers will take and they will very likely get traction on that. Arguments like "but I can't run super AAA upscale PC game foobar on those other systems" or "I can't do my 32GB Maya model on those other systems " point to narrower niches not to definition of the overall market segment. What do normal people do on computers is what the personal computer market is. That isn't classic PC form factor anymore.
 
  • Like
Reactions: tuxon86

cube

Suspended
May 10, 2004
17,011
4,973
The Personal computer ( a computer for personal use) market isn't simply classic PC form factors anymore.
That is exactly the position the lawyers will take and they will very likely get traction on that. Arguments like "but I can't run super AAA upscale PC game foobar on those other systems" or "I can't do my 32GB Maya model on those other systems " point to narrower niches not to definition of the overall market segment. What do normal people do on computers is what the personal computer market is. That isn't classic PC form factor anymore.
The Post-PC market is irrelevant to this application.
 

deconstruct60

macrumors G5
Mar 10, 2009
12,493
4,053

20+ mm^2 bigger dies. That's probably why not jumping to 10nm because need to going to have very mature yields when the number of dies on a wafer goes down. That die size increase will be a bump up in cost even if mature. Likely what may see is the smaller 4 core die on mature, optimized process would be cost competitor with Zen not this 6 core.
The 6 core will probably gap them on performance/$ .
[doublepost=1479612637][/doublepost]
...
Zen seems to be making Intel move.
Still DP 1.2?!

Not really. The pain (yield) and expense of 10nm is keeping them at 14nm for desktop (and bigger dies ).

It is still DP 1.2 because still the Kaby Lake graphics subsystem. Perhaps optimized for that 14++ nm process but same stuff. There are still no DP 1.3 required monitors so DP 1.2 works just fine. The 2016 MBP 13" is driving 5K just fine with skylake graphic subsystem.
 
  • Like
Reactions: tuxon86

cube

Suspended
May 10, 2004
17,011
4,973
If they could cheaply produce a 22FDX 8-core Excavator AM3+ FX, I would be quite happy buying it.
It should officially support DDR3 2400 fully populated (1T, DR) as this speed is already possible with some 970 boards and can be bought almost as cheap as 1866 memory.
 
Last edited:

koyoot

macrumors 603
Original poster
Jun 5, 2012
5,939
1,853
https://www.heise.de/newsticker/mel...enCompute-mit-ROCm-aber-kein-Zen-3491050.html

Random Post on Anandtech said:
- No samples were presented/showcased at SC16 conference, not even behind the public
- In NDA sessions only slides were presented
- In SPECint2006 AMD replaced Intel with gcc compiler and the performance for 32 core Naples was slightly below Haswell
- Results are without Hyper-Threading because of somes issues
- When all is working AMD claims they will be on par with Skylake EP (Xeon E5 2698v5), but there is no info on what this is based on or if this is even based on some real comparisons rather than predictions, also AMD remained silent about some other important HPC performance metrics (e.g. SPECfp, Linpack)
That doesnt spill confidence, AMD.

Come on, give some samples. Give some fu**** proofs for your claims.

Xeon E5-2698v5 would be 24-26 core CPU design.

32 core vs 24-26 core design. I suppose they might have in mind in Multithreading performance.

Why the f*** remain so silent if you have so good proposition?
 
Last edited:

cube

Suspended
May 10, 2004
17,011
4,973
If Zen does not officially support DDR4 3600, it is already late.

AMP RAM is not something one usually finds in shops.
 
Last edited:

deconstruct60

macrumors G5
Mar 10, 2009
12,493
4,053
Having to use 2 ports and 2 cables is not just fine.

The 2016 MBP uses just one cable. So your point is?????

Even with DP 1.3 ( 1.4) monitors most likely the first generation is *still* going to be taking two DP v1.2 MST streams over one cable as the display controllers are probably not going to 5K right away. Just like the 1st generation 4K. 5K is going to get into a "race to bottom" on pricing quite quickly and folks most likely are going to use the same controllers to drive panels as the ones currently out.

Thunderbolt v3 is just a way of deploying it early and quicker on one cable. DPv1.3 will follow later on same path.
 

cube

Suspended
May 10, 2004
17,011
4,973
The 2016 MBP uses just one cable. So your point is?????

Even with DP 1.3 ( 1.4) monitors most likely the first generation is *still* going to be taking two DP v1.2 MST streams over one cable as the display controllers are probably not going to 5K right away. Just like the 1st generation 4K. 5K is going to get into a "race to bottom" on pricing quite quickly and folks most likely are going to use the same controllers to drive panels as the ones currently out.

Thunderbolt v3 is just a way of deploying it early and quicker on one cable. DPv1.3 will follow later on same path.
My point is that 5K monitors using DP 1.2 require 2 cables.

I will never buy a monitor with only 1 input, or that can only be used by select computers.
 

ManuelGomes

macrumors 68000
Dec 4, 2014
1,617
354
Aveiro, Portugal
DP 1.3/1.4 monitors are taking their time. With GPUs out it would be expected that monitors would come along, and not 2 DP 1.2 stitched together.
Let's hope Apple will still make the TB3D, and a proper one.
 

deconstruct60

macrumors G5
Mar 10, 2009
12,493
4,053
My point is that 5K monitors using DP 1.2 require 2 cables.

The context here is Intel iGPUs. Intel presents TBv3 as a preferred solution. It only needs one cable for 5K. The '5K world' isn't going to collapse because the next gen Intel iGPU only has DP v1.2.

I will never buy a monitor with only 1 input, or that can only be used by select computers.

if Apple isn't suppressing the TB display market there will be more 3rd party solutions in the future. Doubtful LG is going to be the only one. Apple jumpstarting the first one more likely will expand not restrict the market.

If only select computers is a disqualification then don't buy 5K monitors because there are many millions of computers that can't.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.