Intel's Future Chips: News, Rumours & Reviews

[goldstone77]

https://www.computerbase.de/2018-02/intel-skylake-d-cpu-supermicro-mainboard/

Supermicro leads the first new motherboards with the Broadwell DE successor Skylake-D. The solutions available as Xeon D in the market will soon be a potent successor, already the first offshoots were awarded at the start as " absolute low-power monster " .

A big limitation Intel will pick up on Skylake-D over its predecessor. Because this was particularly limited on the memory front with a maximum of 128 GB, the Skylake D motherboards should now be able to address a total of 512 GB DDR4-2666 and thus address completely new areas. Because the CPU performance of the small offshoot in many cases is more than sufficient, had already shown the previous Xeon D. But Intel also wanted to prevent that the Xeon D poaching too much in the market of small regular Xeon processors. However, Xeon SP with Skylake SP CPUs has significantly boosted both features and costs, leaving more room for the Skylake-D - also considering the growing competition from AMDand in special server markets also ARM .
Skylake-D is Xeon D-2000
The Supermicro Product Support Site, denoted X11SDV-16C-TP8F, X11SDV-12C-TP8F, and X11SDV-8C-TP8F, discloses at least three Skylake-D family processors, available in 16 cores, 12 cores, and also eight cores should. The search also brings three CPU names to light: Xeon D-2146NT (8 cores), Xeon D-2166NT (12 cores) and Xeon D-2183IT (16 cores). What the letter abbreviations are at the end is not yet known. However, the new 2000 series would replace the Xeon D-1500, under which all Broadwell DE trade. Intel had already confirmed in a presentation that they should come in early 2018 - but this was before Meltdown and Specter.
How exactly the technical details for Skylake-D is ordered, there is so far little information. As a high-end spin-off, Skylake-SP has a distinctly different structure than Skylake-S / H / U for the mainstream market, which is why it's interesting to see what middle ground Intel chooses for Skylake-D. Broadwell-DE already offered a maximum of 16 cores and 32 threads, plus dual-channel memory and an additional 32 PCIe 3.0 lanes - all with TDPs between 20 and 65 watts.

Also Xeon E (Coffee Lake-S) in the starting blocks
For some days now, Supermicro has also been running the X11SCZ-F, a first mainboard for the upcoming Xeon E. Behind it, there are Coffee Lake-S , as they are already on the desktop. The similarities of the platform are almost identical to a desktop SKU on a Z370 motherboard. The launch of the Coffee Lake for the professional environment is expected soon as the second wave of core processors with up to six cores .

Interesting!

 

[juanrga]

All that aside, since it's an i9 I wonder if that means it will have quad channel ram?

It is dual-channel.

 

[goldstone77]

All that aside, since it's an i9 I wonder if that means it will have quad channel ram?

It is dual-channel.

That's a shame, because it would be nice to see laptops/mobile/smaller form factors mimicking extreme desktop processors. Smaller form factors are my personal favorite future technology.

 

[gamerk316]

All that aside, since it's an i9 I wonder if that means it will have quad channel ram?

It is dual-channel.

That's a shame, because it would be nice to see laptops/mobile/smaller form factors mimicking extreme desktop processors. Smaller form factors are my personal favorite future technology.

The problem is that to do so, you need to increase power draw. And in mobile, power draw determines literally EVERYTHING that you do. Sticking with Dual Channel is one of the easier things they can do to keep power draw down.

 

[goldstone77]

Spoiler

All that aside, since it's an i9 I wonder if that means it will have quad channel ram?

It is dual-channel.

That's a shame, because it would be nice to see laptops/mobile/smaller form factors mimicking extreme desktop processors. Smaller form factors are my personal favorite future technology.

The problem is that to do so, you need to increase power draw. And in mobile, power draw determines literally EVERYTHING that you do. Sticking with Dual Channel is one of the easier things they can do to keep power draw down.

That makes sense, but since they are putting 4.8GHz hexa core processors in these laptops/smaller form factor devices eventually you would think they might go all the way with it. Which I would love to see of course! 😀

 

[juanrga]

All that aside, since it's an i9 I wonder if that means it will have quad channel ram?

It is dual-channel.

That's a shame, because it would be nice to see laptops/mobile/smaller form factors mimicking extreme desktop processors. Smaller form factors are my personal favorite future technology.

The problem is that to do so, you need to increase power draw. And in mobile, power draw determines literally EVERYTHING that you do. Sticking with Dual Channel is one of the easier things they can do to keep power draw down.

In fact standards as LPDDR3 and LPDDR4 has been designed just for mobile. LP = Low Power.

 

[juanrga]

Spoiler

All that aside, since it's an i9 I wonder if that means it will have quad channel ram?

It is dual-channel.

That's a shame, because it would be nice to see laptops/mobile/smaller form factors mimicking extreme desktop processors. Smaller form factors are my personal favorite future technology.

The problem is that to do so, you need to increase power draw. And in mobile, power draw determines literally EVERYTHING that you do. Sticking with Dual Channel is one of the easier things they can do to keep power draw down.

That makes sense, but since they are putting 4.8GHz hexa core processors in these laptops/smaller form factor devices eventually you would think they might go all the way with it. Which I would love to see of course! 😀

New process node 14nm++ can help pushing clocks 14% higher, but cannot account for the power consumption from doubling the number of memory channels. Moreover few consumer applications benefit from quad-channel, so most of the time it would be only extra power consumption affecting batteries.

 

[8350rocks]

Spoiler

GlobalFoundries is claiming they are making 5GHz operating devices. How else do you expect performance to increase when you shrink down the process? Faster moving transistors! How do you make them operate faster? Design them to function faster! What do we know? GlobalFoundries said they are making 5GHz operating Server, Data Center, ASICS parts! GlobalFoundries have made massive beneficial changes to the design of their transistors! It's not rocket science it's deductive reasoning based on the information provided.
Here is the claim of 5GHz operating devices:

Here is the transistor performance at 7nm compared to 14nm

Here is what the transistor looks like at 14nm vs. what they look like at 7nm. Notice the drastic change in shape of the transistors!

We know that a long rectangular slightly rounded edges provides the best characteristics for high performance transistors! GlobalFoundries 14nm transistors look to be closer to the 22-nm on the chart below, while the 7nm looks closer to the 10nm. While being a much more refined smooth design as well.

Here is Intel's 10nm transistor design. It's looks really similar to GlobalFoundries 7nm.

Here are the details of different cells or IP blocks that can be implemented of which include 7nm HPC. It's as simple as saying okay build on this IP foundation for my 7nm chips and they select 9T libraries for the process:

Don't confuse the following:

1: "We can allow you to make a single core special purpose CPU that operates in the 5GHz range at some power envelope"

and

2 "We can allow you to make a multi-core general purpose CPU that operates in the same power envelope as the above, but properly will run 20-30% slower as a result"

Seriously people, I still remember when people were claiming Intels 45nm would allow them to clock up the Core architecture up to the 10GHz they promised for Netburst. We go through this EVERY power node; please stop.

Of course design plays it's roll in frequency. What did we learn from GlobalFoundries licensing Samsungs process? That process/IP will restrict frequency. Improved process and improved transistor designs while shrinking this process will fundamentally lead to higher operating frequencies. Compare the technological improvements being made, not Intel PR.

I'd say that is a non-trivial point to take.

For better or for worse, whenever Intel talks about die shrinks, everyone listens carefully and takes notes. No matter the reason, they're still leading it (by how much, not really worth discussing for this point), so all current "common" knowledge comes from whatever Intel has been doing and tweaking for themselves. That includes the "reasoning" why. So, it is not a trivial point to take.

Cheers!

I recall when Intel was behind on process and no one cared or listened to what they were doing...

Is not that Easy to recall when Intel has been ahead since their launch of their Core 2 Duo processors back in 2006. Your name actually cracks me up, AMD took a misstep in its Bulldozer / Piledriver architecture (FX-series) been the worse modern processor architecture ever created by AMD.

AMD was very competitive up to Ivy Bridge...then they went years without releasing a proper HEDT processor.

 

[goldstone77]

TECH TRADER DAILY
Intel’s CFO Swan Explains The Transformation
ByTiernan Ray Feb. 1, 2018 1:40 p.m. ET
https://www.barrons.com/articles/intels-cfo-swan-explains-the-transformation-1517510410

While Krzanich can offer magical sound and light shows of “data” applications, as he did at the Consumer Electronics Show in Las Vegas this month, it is up to Swan to explain the delicate financial trade-off for Intel in moving from its rich fiefdom in PC chips to parts for the Internet of Things that in many cases come with lower gross profit margin.

In an extended interview this week with Tech Trader, Swan makes the case that margins can go down, and well may, as long as it means that higher revenue growth, and higher growth in absolute profit dollars, are the result.

All of that, of course, backed by higher cash flow following U.S. tax reform, which funds Intel's commitment to increasing dividends, offsetting stock compensation dilution, and buying back stock “opportunistically."

Swan, 57, was for nine years CFO at eBay (EBAY), and took over the books from Intel’s Stacy Smith with the Q4 report a year ago. He is a veteran of private equity, having come to Intel from General Atlantic LLC.

It's a short interview.

 

[gamerk316]

Spoiler

GlobalFoundries is claiming they are making 5GHz operating devices. How else do you expect performance to increase when you shrink down the process? Faster moving transistors! How do you make them operate faster? Design them to function faster! What do we know? GlobalFoundries said they are making 5GHz operating Server, Data Center, ASICS parts! GlobalFoundries have made massive beneficial changes to the design of their transistors! It's not rocket science it's deductive reasoning based on the information provided.
Here is the claim of 5GHz operating devices:

Here is the transistor performance at 7nm compared to 14nm

Here is what the transistor looks like at 14nm vs. what they look like at 7nm. Notice the drastic change in shape of the transistors!

We know that a long rectangular slightly rounded edges provides the best characteristics for high performance transistors! GlobalFoundries 14nm transistors look to be closer to the 22-nm on the chart below, while the 7nm looks closer to the 10nm. While being a much more refined smooth design as well.

Here is Intel's 10nm transistor design. It's looks really similar to GlobalFoundries 7nm.

Here are the details of different cells or IP blocks that can be implemented of which include 7nm HPC. It's as simple as saying okay build on this IP foundation for my 7nm chips and they select 9T libraries for the process:

Don't confuse the following:

1: "We can allow you to make a single core special purpose CPU that operates in the 5GHz range at some power envelope"

and

2 "We can allow you to make a multi-core general purpose CPU that operates in the same power envelope as the above, but properly will run 20-30% slower as a result"

Seriously people, I still remember when people were claiming Intels 45nm would allow them to clock up the Core architecture up to the 10GHz they promised for Netburst. We go through this EVERY power node; please stop.

Of course design plays it's roll in frequency. What did we learn from GlobalFoundries licensing Samsungs process? That process/IP will restrict frequency. Improved process and improved transistor designs while shrinking this process will fundamentally lead to higher operating frequencies. Compare the technological improvements being made, not Intel PR.

I'd say that is a non-trivial point to take.

For better or for worse, whenever Intel talks about die shrinks, everyone listens carefully and takes notes. No matter the reason, they're still leading it (by how much, not really worth discussing for this point), so all current "common" knowledge comes from whatever Intel has been doing and tweaking for themselves. That includes the "reasoning" why. So, it is not a trivial point to take.

Cheers!

I recall when Intel was behind on process and no one cared or listened to what they were doing...

Is not that Easy to recall when Intel has been ahead since their launch of their Core 2 Duo processors back in 2006. Your name actually cracks me up, AMD took a misstep in its Bulldozer / Piledriver architecture (FX-series) been the worse modern processor architecture ever created by AMD.

AMD was very competitive up to Ivy Bridge...then they went years without releasing a proper HEDT processor.

Replace Ivy Bride with Nahalem, and you'd be right. The Phenom II lineup was "ok", certainly better then C2Q, but couldn't compete directly against Nahalem in most situations.

 

[juanrga]

AMD was very competitive up to Ivy Bridge...then they went years without releasing a proper HEDT processor.

AMD was uncompetitive before Sandy. And it has very little to do with releasing HEDT processors. AMD was totally uncompetitive on mobile and server spaces as well.

 

[YoAndy]

Spoiler

GlobalFoundries is claiming they are making 5GHz operating devices. How else do you expect performance to increase when you shrink down the process? Faster moving transistors! How do you make them operate faster? Design them to function faster! What do we know? GlobalFoundries said they are making 5GHz operating Server, Data Center, ASICS parts! GlobalFoundries have made massive beneficial changes to the design of their transistors! It's not rocket science it's deductive reasoning based on the information provided.
Here is the claim of 5GHz operating devices:

Here is the transistor performance at 7nm compared to 14nm

Here is what the transistor looks like at 14nm vs. what they look like at 7nm. Notice the drastic change in shape of the transistors!

We know that a long rectangular slightly rounded edges provides the best characteristics for high performance transistors! GlobalFoundries 14nm transistors look to be closer to the 22-nm on the chart below, while the 7nm looks closer to the 10nm. While being a much more refined smooth design as well.

Here is Intel's 10nm transistor design. It's looks really similar to GlobalFoundries 7nm.

Here are the details of different cells or IP blocks that can be implemented of which include 7nm HPC. It's as simple as saying okay build on this IP foundation for my 7nm chips and they select 9T libraries for the process:

Don't confuse the following:

1: "We can allow you to make a single core special purpose CPU that operates in the 5GHz range at some power envelope"

and

2 "We can allow you to make a multi-core general purpose CPU that operates in the same power envelope as the above, but properly will run 20-30% slower as a result"

Seriously people, I still remember when people were claiming Intels 45nm would allow them to clock up the Core architecture up to the 10GHz they promised for Netburst. We go through this EVERY power node; please stop.

Of course design plays it's roll in frequency. What did we learn from GlobalFoundries licensing Samsungs process? That process/IP will restrict frequency. Improved process and improved transistor designs while shrinking this process will fundamentally lead to higher operating frequencies. Compare the technological improvements being made, not Intel PR.

I'd say that is a non-trivial point to take.

For better or for worse, whenever Intel talks about die shrinks, everyone listens carefully and takes notes. No matter the reason, they're still leading it (by how much, not really worth discussing for this point), so all current "common" knowledge comes from whatever Intel has been doing and tweaking for themselves. That includes the "reasoning" why. So, it is not a trivial point to take.

Cheers!

I recall when Intel was behind on process and no one cared or listened to what they were doing...

Is not that Easy to recall when Intel has been ahead since their launch of their Core 2 Duo processors back in 2006. Your name actually cracks me up, AMD took a misstep in its Bulldozer / Piledriver architecture (FX-series) been the worse modern processor architecture ever created by AMD.

AMD was very competitive up to Ivy Bridge...then they went years without releasing a proper HEDT processor.

Not really: If i remember correctly since the first i7 920 released in 2008 intel was already way ahead. The i7 920 had 4 cores and 8 threads, released in 2008. The Phenom ii x4 920 had only 4 cores without hyperthreading, also released 2008. Long story short multi-core speed of the i7 was +92% after both been overclocked, and single core speed was +54%
😉

http://cpu.userbenchmark.com/Compare/Intel-Core-i7-920-vs-AMD-Phenom-II-X4-920/1981vsm1945

 

[8350rocks]

Spoiler

GlobalFoundries is claiming they are making 5GHz operating devices. How else do you expect performance to increase when you shrink down the process? Faster moving transistors! How do you make them operate faster? Design them to function faster! What do we know? GlobalFoundries said they are making 5GHz operating Server, Data Center, ASICS parts! GlobalFoundries have made massive beneficial changes to the design of their transistors! It's not rocket science it's deductive reasoning based on the information provided.
Here is the claim of 5GHz operating devices:

Here is the transistor performance at 7nm compared to 14nm

Here is what the transistor looks like at 14nm vs. what they look like at 7nm. Notice the drastic change in shape of the transistors!

We know that a long rectangular slightly rounded edges provides the best characteristics for high performance transistors! GlobalFoundries 14nm transistors look to be closer to the 22-nm on the chart below, while the 7nm looks closer to the 10nm. While being a much more refined smooth design as well.

Here is Intel's 10nm transistor design. It's looks really similar to GlobalFoundries 7nm.

Here are the details of different cells or IP blocks that can be implemented of which include 7nm HPC. It's as simple as saying okay build on this IP foundation for my 7nm chips and they select 9T libraries for the process:

Don't confuse the following:

1: "We can allow you to make a single core special purpose CPU that operates in the 5GHz range at some power envelope"

and

2 "We can allow you to make a multi-core general purpose CPU that operates in the same power envelope as the above, but properly will run 20-30% slower as a result"

Seriously people, I still remember when people were claiming Intels 45nm would allow them to clock up the Core architecture up to the 10GHz they promised for Netburst. We go through this EVERY power node; please stop.

Of course design plays it's roll in frequency. What did we learn from GlobalFoundries licensing Samsungs process? That process/IP will restrict frequency. Improved process and improved transistor designs while shrinking this process will fundamentally lead to higher operating frequencies. Compare the technological improvements being made, not Intel PR.

I'd say that is a non-trivial point to take.

For better or for worse, whenever Intel talks about die shrinks, everyone listens carefully and takes notes. No matter the reason, they're still leading it (by how much, not really worth discussing for this point), so all current "common" knowledge comes from whatever Intel has been doing and tweaking for themselves. That includes the "reasoning" why. So, it is not a trivial point to take.

Cheers!

I recall when Intel was behind on process and no one cared or listened to what they were doing...

Is not that Easy to recall when Intel has been ahead since their launch of their Core 2 Duo processors back in 2006. Your name actually cracks me up, AMD took a misstep in its Bulldozer / Piledriver architecture (FX-series) been the worse modern processor architecture ever created by AMD.

AMD was very competitive up to Ivy Bridge...then they went years without releasing a proper HEDT processor.

Not really: If i remember correctly since the first i7 920 released in 2008 intel was already way ahead. The i7 920 had 4 cores and 8 threads, released in 2008. The Phenom ii x4 920 had only 4 cores without hyperthreading, also released 2008. Long story short multi-core speed of the i7 was +92% after both been overclocked, and single core speed was +54%
😉

http://cpu.userbenchmark.com/Compare/Intel-Core-i7-920-vs-AMD-Phenom-II-X4-920/1981vsm1945

2nd gen construction cores were about 15-20% behind in single thread workloads, but even on multithread workloads when comparing to the same market segment (i.e. 8350 vs 3770k).

 

[-Fran-]

First gen "i" CPUs where the first jump Intel took from AMD. Sandy was just the second leap. Phenom II fought well up to the first "i" gen with the 1100BE and the really good last 45nm CPU. Llano used the first 32nm version that Bulldozer (Vishera) would end up using and try to compete with Sandy. We all know AMD got destroyed merciless with Vishera and only with Pilediver they were able to compete *at stock* with Sandy, but Sandy, as we all know, OC'ed like heaven, so anything K would just destroy PD anyway when OC was taken into account.

So... Why were you guys recalling the past again?

Cheers!

 

[juanrga]

2nd gen construction cores were about 15-20% behind in single thread workloads, but even on multithread workloads when comparing to the same market segment (i.e. 8350 vs 3770k).

Piledriver was 30--40% behind in single thread

And when we add SMT and the CMT penalty we can see how a quad-core Ivy Bridge could keep even with an eight-core Piledriver on multithread

4 core * 1.3 * 1.2 / 0.8 = 7.8 core

 

[Isaiah4110]

First gen "i" CPUs where the first jump Intel took from AMD. Sandy was just the second leap. Phenom II fought well up to the first "i" gen with the 1100BE and the really good last 45nm CPU. Llano used the first 32nm version that Bulldozer (Vishera) would end up using and try to compete with Sandy. We all know AMD got destroyed merciless with Vishera and only with Pilediver they were able to compete *at stock* with Sandy, but Sandy, as we all know, OC'ed like heaven, so anything K would just destroy PD anyway when OC was taken into account.

So... Why were you guys recalling the past again?

Cheers!

Because 10+ year old news is still news I guess?


Or maybe because we have an ongoing AMD vs. Intel war in this thread instead of being focused on new news...

 

[logainofhades]

I would suggest stopping the AMD vs Intel bickering. Keep this thread on topic which is future Intel chips.

 

[goldstone77]

ASRock Rack D2100D8UM Intel Xeon D-2100 Series Launch Platform
By Cliff Robinson - February 7, 2018
https://www.servethehome.com/asrock-rack-d2100d8um-intel-xeon-d-2100-series-launch-platform/

Since it is launch day for the Intel Xeon D-2100 series SoC family, we have a number of new platform announcements. ASRock Rack’s initial offering will center around a mATX platform with a ton of expansion. In this short article, we are going to take a look at the ASRock Rack D2100D8UM Skylake-D platform.

mATX!

 

[goldstone77]

Living On The Edge: Intel Launches Xeon D-2100 Series SoCs
by Ian Cutress on February 7, 2018 9:00 AM EST
https://www.anandtech.com/show/12409/intel-launches-xeon-d-2100-series-socs-edge
Those price segments...

 

[goldstone77]

https://twitter.com/WikiChip/status/961249989690314752

 

[logainofhades]

I want to see a mini-itx quad channel board that uses laptop dimms.

 

[goldstone77]

I want to see a mini-itx quad channel board that uses laptop dimms.

I never thought about that, but it would be a great implementation for going into smaller form factors!

 

[logainofhades]

I have seen dual channel setups using laptop dimms. I just cannot figure out why we never see even those anymore.

 

[goldstone77]

by Brandon Hill — Friday, February 09, 2018
Leaked Intel Core i7 Coffee Lake H CPU Benchmarks Show 35% Uplift Over Previous Gen
https://hothardware.com/news/intel-core-i7-8850h-coffee-lake-leaked-benchmarks-claim-35-uplift-over-last-gen

Leaked Intel Core i7 Coffee Lake H CPU Benchmarks Show 35% Uplift Over Previous Gen
When it comes to upcoming hot hardware, we love leaks. Thankfully for us, one of Intel's upcoming mobile gaming chips -- the Core i7-8850H -- is making its presence known early via the SiSoft SANDRA benchmarking suite.
The Core i7-8850H is based on Intel's 8th generation 14nm++ Coffee Lake architecture, has a base clock of 2.6GHz, and contains 9MB of L3 cache. The chip has six physical cores and is HyperThreading-enabled, meaning that it can execute a total of 12 threads
On the SANDRA processor arithmetic benchmark, the Core i7-8850H was able to spit out a score of 159.21GOPS, while the processor multimedia score was pegged at 373.98Mpix/s. For comparison, the 4-core/8-thread Core i7-7820HQ (Kaby Lake) put up scores of 117.78GOPS and 347.32Mpix/s respectively in the same benchmarks.

Other rumored specifications for the Core i7-8850H include an integrated Intel UHD Graphics 630 GPU, and the ability to Turbo to 4GHz on all six cores. The chip is also alleged to Turbo all the way up to 4.3GHz on a single core. Those are lofty clocks for a mobile processor (45W TDP), so we'll be interested to see it in action in a cramped notebook chassis (or AIO desktop).

 

[8350rocks]

I have seen dual channel setups using laptop dimms. I just cannot figure out why we never see even those anymore.

Probably not a ton of sales volume would be my guess. LP dimms are hard to find, and can be expensive comparatively if you are not aware of where to look.