Intel's Future Chips: News, Rumours & Reviews

[goldstone77]

Probably correct about that. Although I wouldn't mind if Intel decided to price cut AMD a bit to push prices down more. I still liked it when it was simple but thats just me. Simplicity.

Instead now we will have i7/i9 for mainstream and HEDT and then i5/i3 for desktop with varying cores and threads. Hard for me to remember all of this.

Yeah, just follow the money trail. It gets more expensive as your go up, i3, i5, i7, i9, HEDT.

 

[goldstone77]

i9 9700K 8/8 12 MB cache, 3.6 GHz clock

http://ranker.sisoftware.net/show_device.php?q=c9a598d1bfcbaec2e2a1cebcd9f990a78ab384b484cfe98eb39eaf89fbc6f7d1b885a3cbf6d0a895b3d6b38ebe98ebd6ee&l=en

 

[jimmysmitty]

Great all we need now is pricing to see if it will keep pushing more for less.

 

[goldstone77]

David Schor
‏@david_schor
5h5 hours ago
This is really bad news. Intel is not expecting Ice Lake SP until Q1/2 2020? (Source: Intel's published presentation from China "Intel High Performance Computing and Intelligent Computing") #10nm #IceLake

https://twitter.com/david_schor/status/1022142835989118977

Notice the 2019 and 2020 timelines on the roadmap.

 

[jimmysmitty]

I'm more interested in Barlow Pass DIMM. Apache Pass is the NVDIMMs Intel has been talking about. 512GB per DIMM for up to 2TB of NVDIMMs.

Really want that to trickle down to consumer end.

 

[gamerk316]

Great all we need now is pricing to see if it will keep pushing more for less.

Told you this was going to happen.

As the end of the day, Intel decided it was going to focus on increasing clocks as much as they can, which essentially meant trading off HTT on it's non HEDT chips. Intel's pretty much reached the end of their architectures lifetime, and they are now at the point where they have to take something out (HTT) in order to increase performance in other areas (clocks).

Basically, you're seeing the result of Intels 10nm problems.

 

[jimmysmitty]

HTT has never been a deciding factor for me. At best it can increase performance 20%. At worst it does nothing. Better than the first implementation, which made performance worse, but 16 threads seems a bit overkill.

Of course this is if the i9 with HTT is not real. Plus the base clock isn't that fast.

 

[gamerk316]

HTT has never been a deciding factor for me. At best it can increase performance 20%. At worst it does nothing. Better than the first implementation, which made performance worse, but 16 threads seems a bit overkill.

Of course this is if the i9 with HTT is not real. Plus the base clock isn't that fast.

Yields are likely the reason for all this. i9s are the ones that can reach those clocks without disabling HTT. I fully expect the 8/8 i7's are just going to fuse off the HTT circuits.

I'll say it again: Core is EOL, and Intel better be working on a replacement architecture and not just hoping that fixing their foundries will make these issues go away.

 

[jimmysmitty]

HTT has never been a deciding factor for me. At best it can increase performance 20%. At worst it does nothing. Better than the first implementation, which made performance worse, but 16 threads seems a bit overkill.

Of course this is if the i9 with HTT is not real. Plus the base clock isn't that fast.

Yields are likely the reason for all this. i9s are the ones that can reach those clocks without disabling HTT. I fully expect the 8/8 i7's are just going to fuse off the HTT circuits.

I'll say it again: Core is EOL, and Intel better be working on a replacement architecture and not just hoping that fixing their foundries will make these issues go away.

Pretty sure they do have a new uArch in the works. But they are probably super cautious as last time they went in a new direction it wasn't as well received.. Hell the last two times. Netburst and Itanium both were greeted with much hate.

That's the sad state of PCs though. Trying to push past x86 is like trying to move a mountain. It can be done but it would take so much more than just sticking more than just building around the mountain.

 

[Supernova1138]

HTT has never been a deciding factor for me. At best it can increase performance 20%. At worst it does nothing. Better than the first implementation, which made performance worse, but 16 threads seems a bit overkill.

Of course this is if the i9 with HTT is not real. Plus the base clock isn't that fast.

Yields are likely the reason for all this. i9s are the ones that can reach those clocks without disabling HTT. I fully expect the 8/8 i7's are just going to fuse off the HTT circuits.

I'll say it again: Core is EOL, and Intel better be working on a replacement architecture and not just hoping that fixing their foundries will make these issues go away.

Pretty sure they do have a new uArch in the works. But they are probably super cautious as last time they went in a new direction it wasn't as well received.. Hell the last two times. Netburst and Itanium both were greeted with much hate.

That's the sad state of PCs though. Trying to push past x86 is like trying to move a mountain. It can be done but it would take so much more than just sticking more than just building around the mountain.

Yeah, that's the 800lb gorilla that Intel would have to deal with, it's what killed Itanium with performance being abysmal unless your software was specifically written and optimized for it. I don't know what the solution would be, whether they'd need to develop a new architecture that allows for very good x86 emulation capabilities (way better than the x86 emulation we've seen on the Windows ARM devices) to keep legacy software compatible, or if they'd have to go so far as to build in x86 cores in the first few generations to provide that support for older software while they move on to something new.

 

[gamerk316]

HTT has never been a deciding factor for me. At best it can increase performance 20%. At worst it does nothing. Better than the first implementation, which made performance worse, but 16 threads seems a bit overkill.

Of course this is if the i9 with HTT is not real. Plus the base clock isn't that fast.

Yields are likely the reason for all this. i9s are the ones that can reach those clocks without disabling HTT. I fully expect the 8/8 i7's are just going to fuse off the HTT circuits.

I'll say it again: Core is EOL, and Intel better be working on a replacement architecture and not just hoping that fixing their foundries will make these issues go away.

Pretty sure they do have a new uArch in the works. But they are probably super cautious as last time they went in a new direction it wasn't as well received.. Hell the last two times. Netburst and Itanium both were greeted with much hate.

That's the sad state of PCs though. Trying to push past x86 is like trying to move a mountain. It can be done but it would take so much more than just sticking more than just building around the mountain.

Itanium as an architecture was fine; people just hated to 10% performance hit to x86 apps, ignoring the fact that x86 was going to be replaced in short order.

At this point, we're stuck with x86; Windows and it's outright massive library of stuff pretty much ensures this. If you want to replace x86 at this point, you'll need to design a CPU that fully emulates both x86 and x86-64, which is no easy feat. We had our chance to kill it off, now we're stuck with it for at least the next few decades.

 

[aldaia]

David Schor
‏@david_schor
5h5 hours ago
This is really bad news. Intel is not expecting Ice Lake SP until Q1/2 2020? (Source: Intel's published presentation from China "Intel High Performance Computing and Intelligent Computing") #10nm #IceLake

https://twitter.com/david_schor/status/1022142835989118977

Notice the 2019 and 2020 timelines on the roadmap.

Interesting, xeon phy line is being replaced by cascade lake

Also see footnote 1: Left Edge Box = Lead Silicon Die factory ramp begins. Add 1-2 quarters for end user availability

That moves 10nm ICE Lake-SP availability dangerously close to 2021 specially considering that AMD 7nm Rome is expected in 2019

 

[goldstone77]

Interesting, xeon phy line is being replaced by cascade lake

Also see footnote 1: Left Edge Box = Lead Silicon Die factory ramp begins. Add 1-2 quarters for end user availability

That moves 10nm ICE Lake-SP availability dangerously close to 2021 specially considering that AMD 7nm Rome is expected in 2019

Yep, you are seeing the same things I'm seeing. (Intel 10nm)106.1 MTx/mm2 vs. (TSMC 5nm)185.46 MTx/mm2 looking at SemiWiki could potentially happen.

 

[YoAndy]

Leaked Benchmarks Suggest Intel Will Drop Hyper-Threading from Core i7 and the new 8 core i9 9700K could retail for $450.

https://www.extremetech.com/extreme/274246-leaked-benchmarks-suggest-intel-will-drop-hyper-threading-from-core-i7?utm_medium=notification

 

[goldstone77]

Leaked Benchmarks Suggest Intel Will Drop Hyper-Threading from Core i7 and the new 8 core i9 9700K could retail for $450.

https://www.extremetech.com/extreme/274246-leaked-benchmarks-suggest-intel-will-drop-hyper-threading-from-core-i7?utm_medium=notification

A lot of people took a double take at the leak! "Intel is also preparing i7-9700K, which would also be the 8-core SKU, but interestingly, without hyper-threading."

https://videocardz.com/76808/intel-core-i9-9900k-i7-9700k-and-i5-9600k-possible-specifications-emerge

 

[goldstone77]

Intel Has Banner Quarter, Announces 10nm Systems Available In 2019, Stock Tumbles
by Paul Alcorn July 26, 2018 at 10:00 PM

Despite the record revenue, a 78% gain in profits to $5 billion, and increased guidance, the company's stock slipped 6% in after-hours trading due to slower-than-anticipated growth in its data center portfolio.

Intel cited recent tax cuts as one of the key factors in increased data center spending by its customers. Intel also benefited from tax cuts and is now paying a much lower rate, which helped boost profit. Margins remained flat at an impressive 63%.

Intel's faster ramp of 10nm data center products is going to be a critical component in fending off AMD's 7nm EPYC Rome processors that arrive early next year. Even with a shorter gap between Intel's desktop and data center processors, AMD has a relatively large window it can exploit with 7nm processors. Recently leaked roadmaps imply that Intel won't have 10nm server products on the market until mid-2020, but company representatives sounded more bullish on the roadmap during the call. Intel is already focusing on applying the lessons it learned with the 10nm delays to its future 7nm node. Intel's Murthy said "we're focusing on an optimum balance point between density, power and performance, and schedule predictability. So I think what you'll see is a more balanced approach across those three vectors."

https://www.tomshardware.com/news/intel-10nm-ceo-earnings-datacenter,37518.html

AMD was up ~18% for the day in after hour trading.

 

[goldstone77]

Intel (INTC) Q2 2018 Results - Earnings Call Transcript
Jul. 26, 2018 8:43 PM ET
https://seekingalpha.com/article/4190920-intel-intc-q2-2018-results-earnings-call-transcript

Vivek Arya - Bank of America Merrill Lynch

Thanks for taking my question. Bob, on 10-nanometer progress, any color on what these are doing? And systems you mentioned on shelf for second half 2019, I assume that's mostly PC Client. Any sense of when we can think of timing for your server products on 10-nanometer also?

Venkata S. M. Renduchintala - Intel Corp.

Hi, this is Murthy. I'll take that one. We continue to make progress on 10-nanometer. Yields are improving consistent with the timelines we shared in April. And yes, you're quite right. The systems on shelves that we expect in holiday 2019 will be client systems, with data center products to follow shortly after.

Vivek Arya - Bank of America Merrill Lynch

Thank you.

Robert Holmes Swan - Intel Corp.

So good progress on 10-nanometer and what we think is a very good lineup on 14-nanometer product for next year on both client and server that we think will deliver best-in-class performance as we continue to ramp 10-nanometer.

14nm for both client and server next year. This is consistent with the roadmap. Notice 10nm Icelake-SP mid to late 2020. Cooper Lake -SP is 14nm.

 

[goldstone77]

Intel Xeon Scalable Family Roadmap Revealed – Points Out 14nm Cascade Lake-SP in Q4 2018, Cooper Lake-SP in Q4 2019, 10nm Ice Lake-SP in 1H 2020
By Hassan Mujtaba
20 hours ago

Intel Xeon Scalable Family Roadmap Reveals 14nm Cooper Lake-SP in 2019, 10nm Ice Lake-SP in 2020 and Next-Generation Xeon Post-2020

The details show that Intel is currently planning the Cascade Lake-SP family for launch later this year. Based on the 14nm+ node, the Cascade Lake-SP Xeons would feature higher efficiency and retain support on the existing Purley platform, while offering new feature and I/O expandability. The following lineups are expected under the Cascade Lake family:
Now, we have known Cascade Lake-SP family for a while and the Cascade Lake-X family will have lots of similarities with it. Both Cascade Lake-SP and Cascade Lake-X processors will be featuring support for the LGA 3647 socket. Both families will have support for 6 channel memory and we can also expect faster memory support of DDR4-2800 (native). Cascade Lake-SP processors will be the first to feature Optane DIMM support which is shipping later this year but this cannot be said to be true for Cascade Lake-X series. The Cascade Lake Xeons would feature up to 28 cores and 56 threads, the same configurations would be made available to enthusiasts in the form of Cascade Lake-X.

Moving on to the Cascade Lake-AP series, the new family will be the first to use the “Advanced Processor” badge. Expected to hit markets in the mid of 2019 and under the ‘Walker Pass’ brand platform, the Cascade Lake Advanced Processor lineup is Intel trying to regain their momentum on the server side after taking a huge hit by AMD’s disruptive EPYC line of CPUs that offer more cores, more memory, and more PCIe lanes while retaining Intel’s level of IPC. This has taken a toll on Intel’s market share and the effects are apparent with Intel’s CEO, Brian Krzanich, expecting his company to lose server market share to AMD’s new platform.

• 
  Intel Cascade Lake-AP “Advanced Processor Family” (BGA 5903)
  Intel Cascade Lake-SP “Scalable Processor Family” (LGA 3647)
  Intel Cascade Lake-X “HEDT Consumer Family” (LGA 3647)

With Cascade Lake-AP, Intel plans to execute their own MCM (Multi-Chip-Module) approach. An industry insider, Ashraf Essa, has previously heard rumors of the Advanced Processor lineup and mentioned that it wasn’t expected until the Ice Lake generation. But now with the EPYC series proving to be crushing Intel in one of their dominant markets, there’s no doubt that they’ll end up using an MCM approach a generation early.

https://wccftech.com/intel-xeon-scalable-family-roadmap-revealed-points-out-cascade-lake-sp-in-q4-2018-cooper-lake-sp-in-q4-2019-ice-lake-sp-in-1h-2020/

 

[gamerk316]

Intel Has Banner Quarter, Announces 10nm Systems Available In 2019, Stock Tumbles
by Paul Alcorn July 26, 2018 at 10:00 PM

Despite the record revenue, a 78% gain in profits to $5 billion, and increased guidance, the company's stock slipped 6% in after-hours trading due to slower-than-anticipated growth in its data center portfolio.

Intel cited recent tax cuts as one of the key factors in increased data center spending by its customers. Intel also benefited from tax cuts and is now paying a much lower rate, which helped boost profit. Margins remained flat at an impressive 63%.

Intel's faster ramp of 10nm data center products is going to be a critical component in fending off AMD's 7nm EPYC Rome processors that arrive early next year. Even with a shorter gap between Intel's desktop and data center processors, AMD has a relatively large window it can exploit with 7nm processors. Recently leaked roadmaps imply that Intel won't have 10nm server products on the market until mid-2020, but company representatives sounded more bullish on the roadmap during the call. Intel is already focusing on applying the lessons it learned with the 10nm delays to its future 7nm node. Intel's Murthy said "we're focusing on an optimum balance point between density, power and performance, and schedule predictability. So I think what you'll see is a more balanced approach across those three vectors."

https://www.tomshardware.com/news/intel-10nm-ceo-earnings-datacenter,37518.html

AMD was up ~18% for the day in after hour trading.

I'll say it again: The markets are 90% psychology. No other reason why Intels stock should go down on these results, except people think Intel should somehow be doing better.

 

[goldstone77]

Really, I was under a completely different impression. I've been assuming it was about how well a company/companies execute on there guidance, financial stability, and positive future prospects, which indecently are very poor for Intel at this moment! But, it couldn't have anything to do with that, could it? Who knew it was all just, "the mental and emotional factors governing a situation or activity." What a volatile world of uncertainty you must live in, since mind reading would be required to know how the market will react at any given time.... Come on, it's called economics!

 

[VladDracule]

I haven't read through the entire thread, is there any legitimate reason to wait for the 9700k? What kind of performance increase are we expecting?

 

[Supernova1138]

I haven't read through the entire thread, is there any legitimate reason to wait for the 9700k? What kind of performance increase are we expecting?

9th generation is basically just a Coffee Lake refresh, so no IPC gains, only changes to the core/thread count. If the rumour is true and its an 8 core/8 thread CPU, then the performance boost over an 8700k is going to be variable, stuff that needs lots of cores but doesn't benefit from hyperthreading much will see a boost, but stuff that does benefit a lot from hyperthreading could perform fairly similar on both CPUs with the 9700k maybe only being a hair faster. If you're gaming, then probably no difference at all since virtually no games scale beyond 6 cores in any appreciable way and the only way you might see a performance boost is if the rumour of the 9700k being soldered is true, in which case it might be easier to push a higher overclock compared to the 8700k.

 

[VladDracule]

I haven't read through the entire thread, is there any legitimate reason to wait for the 9700k? What kind of performance increase are we expecting?

9th generation is basically just a Coffee Lake refresh, so no IPC gains, only changes to the core/thread count. If the rumour is true and its an 8 core/8 thread CPU, then the performance boost over an 8700k is going to be variable, stuff that needs lots of cores but doesn't benefit from hyperthreading much will see a boost, but stuff that does benefit a lot from hyperthreading could perform fairly similar on both CPUs with the 9700k maybe only being a hair faster. If you're gaming, then probably no difference at all since virtually no games scale beyond 6 cores in any appreciable way and the only way you might see a performance boost is if the rumour of the 9700k being soldered is true, in which case it might be easier to push a higher overclock compared to the 8700k.

In terms of multitasking wouldn't more physical cores be better? I.E I'm playing a game and ripping a movie at the same time, or rencoding one of those rips while gaming

 

[Supernova1138]

I haven't read through the entire thread, is there any legitimate reason to wait for the 9700k? What kind of performance increase are we expecting?

9th generation is basically just a Coffee Lake refresh, so no IPC gains, only changes to the core/thread count. If the rumour is true and its an 8 core/8 thread CPU, then the performance boost over an 8700k is going to be variable, stuff that needs lots of cores but doesn't benefit from hyperthreading much will see a boost, but stuff that does benefit a lot from hyperthreading could perform fairly similar on both CPUs with the 9700k maybe only being a hair faster. If you're gaming, then probably no difference at all since virtually no games scale beyond 6 cores in any appreciable way and the only way you might see a performance boost is if the rumour of the 9700k being soldered is true, in which case it might be easier to push a higher overclock compared to the 8700k.

In terms of multitasking wouldn't more physical cores be better? I.E I'm playing a game and ripping a movie at the same time, or rencoding one of those rips while gaming

It all depends on how CPU intensive these other tasks are and whether they can be done with spare execution resources on the core which is how hyperthreading works. All I can say is it will be interesting to see what the benchmarks look like, we'll have to wait unless someone has an i7 7820x and wants to try to simulate this by running benches with HT disabled, then disable two cores and turn HT back on and see what results he gets in the same benchmarks.

 

[VladDracule]

I haven't read through the entire thread, is there any legitimate reason to wait for the 9700k? What kind of performance increase are we expecting?

9th generation is basically just a Coffee Lake refresh, so no IPC gains, only changes to the core/thread count. If the rumour is true and its an 8 core/8 thread CPU, then the performance boost over an 8700k is going to be variable, stuff that needs lots of cores but doesn't benefit from hyperthreading much will see a boost, but stuff that does benefit a lot from hyperthreading could perform fairly similar on both CPUs with the 9700k maybe only being a hair faster. If you're gaming, then probably no difference at all since virtually no games scale beyond 6 cores in any appreciable way and the only way you might see a performance boost is if the rumour of the 9700k being soldered is true, in which case it might be easier to push a higher overclock compared to the 8700k.

In terms of multitasking wouldn't more physical cores be better? I.E I'm playing a game and ripping a movie at the same time, or rencoding one of those rips while gaming

It all depends on how CPU intensive these other tasks are and whether they can be done with spare execution resources on the core which is how hyperthreading works. All I can say is it will be interesting to see what the benchmarks look like, we'll have to wait unless someone has an i7 7820x and wants to try to simulate this by running benches with HT disabled, then disable two cores and turn HT back on and see what results he gets in the same benchmarks.

Its pretty likely that the next cpus coming out, along with the die shrink should be the same socket right?