Kaby Lake information.

[Fjodor2001]

So speculation and no fact? We can just as well dream that Intel will have HMC memory attached.

You can wake up from your dream. Already in the pipeline for Intel:

http://wccftech.com/intel-altera-offering-hbm2-fpga-sip-stratix-10/

Only for server CPUs though. And HBM2, not HMC.

Also, anyone would be foolish not to understand that the goal for AMD is to combine APU + HBM to make the most of their assets. If done wisely, it will be awesome, powerful, and very competitive both performance and price-wise.

AMD should also have a competitive edge. nVidia can't compete, since they don't have an x86 CPU license and don't make APUs. Intel will have a hard time competing on the GPU side, since AMD is ahead of Intel on that. So if AMD just can come close enough to Intel on the CPU performance with Zen (no need to beat them), they'll be in a great position.

 

[tenks]

You can wake up from your dream. Already in the pipeline for Intel:

http://wccftech.com/intel-altera-offering-hbm2-fpga-sip-stratix-10/

Only for server CPUs though. And HBM2, not HMC.

Also, anyone would be foolish not to understand that the goal for AMD is to combine APU + HBM to make the most of their assets. If done wisely, it will be awesome, powerful, and very competitive both performance and price-wise.

AMD should also have a competitive edge. nVidia can't compete, since they don't have an x86 CPU license and don't make APUs. Intel will have a hard time competing on the GPU side, since AMD is ahead of Intel on that. So if AMD just can come close enough to Intel on the CPU performance with Zen (no need to beat them), they'll be in a great position.

You know every single one of your posts are the same? They always say the same old thing "AMD is better and going to do this, Intel is inferior and can't do that."

But in reality, it's the exact opposite. Do you pay attention to what is actually happening in the market place and in the real world?

I think there is a huge disconnect between what you post and what is reality.

 

[HexiumVII]

Optane sounds cool, but how much layers of caching do we need! We are at 10!

L1, L2, L3, then broadwell L4 graphics memory, then DDR4, the Xpoint, then DRAM cache on SSD and the SLC flash in front of TLC flash lol

 

[Arachnotronic]

You know every single one of your posts are the same? They always say the same old thing "AMD is better and going to do this, Intel is inferior and can't do that."

But in reality, it's the exact opposite. Do you pay attention to what is actually happening in the market place and in the real world?

I think there is a huge disconnect between what you post and what is reality.

Good point. Here's what reality looks like (time for the red pill):

 

[Arachnotronic]

You can wake up from your dream. Already in the pipeline for Intel:

http://wccftech.com/intel-altera-offering-hbm2-fpga-sip-stratix-10/

Only for server CPUs though. And HBM2, not HMC.

Also, anyone would be foolish not to understand that the goal for AMD is to combine APU + HBM to make the most of their assets. If done wisely, it will be awesome, powerful, and very competitive both performance and price-wise.

AMD should also have a competitive edge. nVidia can't compete, since they don't have an x86 CPU license and don't make APUs. Intel will have a hard time competing on the GPU side, since AMD is ahead of Intel on that. So if AMD just can come close enough to Intel on the CPU performance with Zen (no need to beat them), they'll be in a great position.

That's an FPGA that will make an Extreme Edition CPU look like a budget toy. Thanks for proving my point

 

[ShintaiDK]

Haha

 

[Fjodor2001]

That's an FPGA that will make an Extreme Edition CPU look like a budget toy. Thanks for proving my point

Hmmm...Your point being what exactly? And how does it contradict anything I wrote?

 

[Fjodor2001]

You know every single one of your posts are the same? They always say the same old thing "AMD is better and going to do this, Intel is inferior and can't do that."

But in reality, it's the exact opposite. Do you pay attention to what is actually happening in the market place and in the real world?

I think there is a huge disconnect between what you post and what is reality.

You're looking at historical market data where as the rest of us are disussing future scenarios and technology. Thus the disconnect between you and reality.

Also, if you actually would have bothered to read my previous post you would have seen that I actually made a positive remark about Intel and their possible use of HBM2. But I guess you are too biased and blinded by hate to notice that.

 

[happy medium]

rest of us are disussing future scenarios and technology.

No we were discussing Kaby Lake.........not AMD, wrong thread.

and

"Broadwell-E is launched in the spring of 2016

In product plans strengthened also reports that Broadwell-E goes into production in the early second quarter of 2016, which speaks for a launch later in the spring. New here, especially Intel's new standard-bearer Core i7-6950X, which comes with no less than 10 pips and 20 threads via Hyperthreading technology."

http://www.microsofttranslator.com/...e-med-pakostade-grafikdelar-och-l4-cacheminne

 

[ElFenix]

i came in here to read about kaby lake and half this thread is the usual suspects from the AMD brigade and the intel brigade derailing it to something else. started by the intel brigade.

 

[tenks]

You're looking at historical market data where as the rest of us are disussing future scenarios and technology. Thus the disconnect between you and reality.

Also, if you actually would have bothered to read my previous post you would have seen that I actually made a positive remark about Intel and their possible use of HBM2. But I guess you are too biased and blinded by hate to notice that.

Yeap, the past and the present has no bearing on the future, my bad. I totally expect AMD to come out with Zen on time and superior to anything Intel has in price and performance. I also expect Intel to fail miserably when launching their next chip, delay them even further, and then probably go near bankrupt.

It's backwards wednesdaaaaaaay!




Anyway, back on topic I'm interested to see what the kaby lake pch looks like with an HEDT chip. I wonder if it will have a different configuration and a higher speed DMI bus?

 

[jpiniero]

Anyway, back on topic I'm interested to see what the kaby lake pch looks like with an HEDT chip. I wonder if it will have a different configuration and a higher speed DMI bus?

HEDT using Kabylake PCH is not true. Some people misread a leak that for what was the Kabylake Xeon E3. It's limited to 4 cores.

 

[ShintaiDK]

HEDT using Kabylake PCH is not true. Some people misread a leak that for what was the Kabylake Xeon E3. It's limited to 4 cores.

E3 Xeons doesn't fit into Socket R

 

[ShintaiDK]

I'm interested to see what the kaby lake pch looks like with an HEDT chip. I wonder if it will have a different configuration and a higher speed DMI bus?

It will be the exact same. Higher DMI isn't possible without PCIe 4.0 either.

However you will most likely see M.2/U.2 slots directly connected to the CPU on the boards due to the more PCIe lanes by the HEDT platform.

 

[tenks]

HEDT using Kabylake PCH is not true. Some people misread a leak that for what was the Kabylake Xeon E3. It's limited to 4 cores.

Can you prove it's not true? Socket-R is not for Xeon e3s. Socket-R = LGA-2011. And in the last roadmap is clearly said Kabylake PCH, SOCKET-R

It will be the exact same. Higher DMI isn't possible without PCIe 4.0 either.

However you will most likely see M.2/U.2 slots directly connected to the CPU on the boards due to the more PCIe lanes by the HEDT platform.

That's a good point. m.2 direct to CPU is the only to go I guess.

 

[jpiniero]

Can you prove it's not true? Socket-R is not for Xeon e3s. Socket-R = LGA-2011. And in the last roadmap is clearly said Kabylake PCH, SOCKET-R

Well no, but I swore I saw something that said it was limited to 4 cores.

 

[Arachnotronic]

HEDT using Kabylake PCH is not true. Some people misread a leak that for what was the Kabylake Xeon E3. It's limited to 4 cores.

It is true

 

[dark zero]

Purley seems to be a strong reason to DOA Broadwell EX.

The issue will be on how thousands dollars will be more expensive... And also... It will be limited to few countries (China and the Mid East got banned, Korea and Europe will stick with Haswell chips and test their ARM ones)... USA and Israel to be exact.

 

[jpiniero]

If Cannonlake is SoC only and the mainstream is headed that way I suppose Intel could just merge high end socketed desktop, Xeon E3, HEDT and Xeon E5 1xxx into a new middle-tier like. I doubt the volume would justify doing this though.

Intel's new 16c+ Xeon-D processors should be an adequate replacement for 1U servers, it's just workstations and HEDT.

 

[ShintaiDK]

If Cannonlake is SoC only and the mainstream is headed that way I suppose Intel could just merge high end socketed desktop, Xeon E3, HEDT and Xeon E5 1xxx into a new middle-tier like. I doubt the volume would justify doing this though.

Intel's new 16c+ Xeon-D processors should be an adequate replacement for 1U servers, it's just workstations and HEDT.

Why would being a SoC or not change anything in relation to socket? And E3 in the same sentence as HEDT and E5 doesn't mix.

 

[mrmt]

You're looking at historical market data where as the rest of us are disussing future scenarios and technology. Thus the disconnect between you and reality.

I don't think one of the most incompetent management teams active on today's market will be able to oversight the successful research, development and deployment of this myriad of new technologies on the market, but it seems you do.

 

[Phynaz]

You're looking at historical market data where as the rest of us are disussing future scenarios and technology. Thus the disconnect between you and reality.

Past performance is the single most reliable indicator of future performance.

That's the reality.

 

[Systems analyst]

Past performance is the single most reliable indicator of future performance.

That's the reality.

April 2014, Intel reported $2 .8 billion earnings for 2014 Q1 PC group.
April 2015, Intel reported $1 .4 billion earnings in the combined PC + mobile group for 2015 Q1.
- This is a shortfall of $1 .4 billion earnings in a quarter.

July 2014, they reported $3.7 billion earnings for 2014 Q2 PC group.
July 2015, they reported $1 .6 billion earnings in the combined PC + mobile group for 2015 Q2.
- This is a shortfall of $2.1 billion earnings in a quarter.

October 2014, they reported $4.1 billion earnings for 2014 Q3 PC group.
October 2015, they reported $2.4 billion earnings in the combined PC + mobile group for 2015 Q3.
- This is a shortfall of $1.7 billion earnings in a quarter.


Projected 2015 Mobile/PC group shortfall c.f 2014 PC profits:
Q1 = 1.4, Q2=2.1, assume Q3=1.7, assume Q4=1.4
Total shortfall = 6.6 billion USD.
If 4.2 billion is attributable to Mobile,as per 2014, then PC shortfall for 2015 is 2.4 billion.
If 3.4 billion is attributable to Mobile, then PC shortfall for 2015 is 3.2 billion USD.

Following what you said, Intel will see a fall in client group profits of between $2.4 billion and $3.2 billion in 2016, and something similar in 2017. This will mean the end of PC CPUs from Intel, if true. You may well be right, we will see.

 

[Phynaz]

Way to miss the point.

 

[cytg111]

i came in here to read about kaby lake and half this thread is the usual suspects from the AMD brigade and the intel brigade derailing it to something else. started by the intel brigade.

+1