Intel Skylake / Kaby Lake

[swilli89]

If this is solely due to AMD, please explain for me with your chart why Broadwell went from 8 to 10 cores in 2016 before Ryzen or TR had any real released data? Why did Haswell go from 6 cores to 8 cores in 2013? Also due to AMD's strength? Why did Westmere go from 4 to 6 cores in 2010? Was this because of Ryzen or AMD's strength in 2010 as well? Why did Penryn go from 2 cores to 4 in 2007? Was this also due to Ryzen? Were these all AMD caused? Or did Intel just add 2 cores every couple of years?

I think you are arguing a pretty extreme "false equivalency". Just because Intel has increased core counts in the past, doesn't explain away the much larger increased core counts that they are rolling out on HEDT. Again, I'm willing to agree the CoffeLake 6-core chips could have been a planned upgrade in a vacuum outside of AMD. BUT if you believe in trends and patterns if you look at that chart Intel would have stayed at 10, maybe 12 core count for a few more years. Afterall, they stayed at 6-core for multiple years and 8-core for four more years. Most of that time AMD didn't have a competitive offering. Now all of a sudden Intel is pulling in HCC Xeon chips and instead of increasing by 2-cores they are adding 8 more? I'd say pretty much any business analyst looking at that would overwhelmingly point to increased competition as the catalyst for the change. Again.. no harm in admitting Intel is reacting competitively ESPECIALLY when its been accused of acting in non-competitive ways in the past.

8th generation Core, or code name Coffee Lake, are client (mainstream) names for the 14-nm++ Skylake/Kaby Lake optimized replacement for 10-nm Cannon Lake. HEDT and server will also get their Cannon Lake replacement, code named differently.

Thank you for clearing that up.. I definitely got the code names confused. So CannonLake is the 10nm Tick then, got it.

 

[Arachnotronic]

Ok wait so what is this 8th generation then if Cannon lake comes after Kaby Lake.. I am genuinely confused. And will these new 8th gen's have soldered IHS? Base clocks of 3.6ghz + non soldered IHS may actually bring a regression compared to 7700k.. how will they fair against 1600X?

The 8700K, which is Coffee Lake-S, which is 8th gen Core, will have the following turbo frequencies:

1-core: 4.7GHz
2-core: 4.6GHz
4-core:: 4.4GHz
6-core: 4.3GHz.

So, no, not a regression from 7700K.

 

[dullard]

I think you are arguing a pretty extreme "false equivalency".

Again.. no harm in admitting Intel is reacting competitively ESPECIALLY when its been accused of acting in non-competitive ways in the past.

It was your posted graph that excluded most of the core count changes that had nothing to do with AMD that was the "false equivalency". Not only that, but the orange line was incorrect drawing 6 cores when the Extreme Editions had far fewer than that.

Yes, Intel is responding to AMD with using the larger core count Xeon. I said that already. No harm taken as it is the truth. But the chips themselves have nothing to do with AMD. Just the prices.

 

[Malogeek]

It is not a new design, it is not rushed, it is not panic, it is not being forced, it is not new turbos.

Can't wait to see Intel's new Turbo policy for Coffee Lake. No rush, I'll wait. (and maybe even buy)

What is the picture meant to prove for or against your comment?

 

[swilli89]

The 8700K, which is Coffee Lake-S, which is 8th gen Core, will have the following turbo frequencies:

1-core: 4.7GHz
2-core: 4.6GHz
4-core:: 4.4GHz
6-core: 4.3GHz.

So, no, not a regression from 7700K.

Nice. Are these published somewhere?

 

[swilli89]

It was your posted graph that excluded most of the core count changes that had nothing to do with AMD that was the "false equivalency". Not only that, but the orange line was incorrect drawing 6 cores when the Extreme Editions had far fewer than that.

Yes, Intel is responding to AMD with using the larger core count Xeon. I said that already. No harm taken as it is the truth. But the chips themselves have nothing to do with AMD. Just the prices.

When was the last extreme edition with fewer than 6 cores? Maybe I was mistaken

 

[coercitiv]

What is the picture meant to prove for or against your comment?

It shows Intel presumably holding back info on SKL-X from their partners while simultaneously sharing critical info on upcoming CFL-S. So they had an apparently carefully planned SKL-X product stack which they chose to hide, and an apparently expedited CFL-S which they chose to share down to number of cores being launched in Aug-Sep 2017: 6 and 4 core, 95W and 65W TDP.

Both SKL-X and CFL-S were planned ahead with some wiggle room available from start, but execution tempo seems reactionary to me, and so does the final chip lineup.

 

[Shivansps]

Having ring bus vs. not having ring bus completely depended on whether Intel planned on having a 6c-10c Xeon SP.

Xeon, Xeon SP, SL-X are and were always going to be the same products. The only reason LCC might have been ring bus is that at 6-10c ring bus is quicker and if they weren't going to offer and SP unit in that size then the 6-10c Ring bus Xeon would make a great SL-X.

But that doesn't really matter in regards to x299 options. I would challenge that Intel has always stayed on the smaller X299 option. But even then we only need to look at availability. If Intel was going to launch SL-X with MCC, then why didn't it launch with MCC? In fact when has Intel ever launched a product and left all the data and the release date as TBD?

The fact is SKL-X launch was rushed forward, you just cant rant about not being able to move forward EVERY cpu on the lineup. 12 to 18C SKL-X are petty much coming out on the original date.

Maybe, just maybe, if Intel decided to wait and launch the entire SKL-X lineup when it was originally planned (no TBD), this whole rumor may have never existed. But considering there is now people saying CFL-S is thanks to AMD as well, im going to say this looks more like a religion than anything else at this point.

 

[dullard]

When was the last extreme edition with fewer than 6 cores? Maybe I was mistaken

That was in my post to you above. Move to 6 cores in 2010:
https://ark.intel.com/products/4793...Edition-12M-Cache-3_33-GHz-6_40-GTs-Intel-QPI

In 2009 it was only 4 cores:
http://ark.intel.com/products/37153...-Edition-8M-Cache-3_33-GHz-6_40-GTs-Intel-QPI

Move to 4 cores in late 2006 (sorry I accidently posted 2007 above):
https://ark.intel.com/products/2802...ocessor-QX6700-8M-Cache-2_66-GHz-1066-MHz-FSB

Before that it was just 2 cores in mid 2006:
http://ark.intel.com/products/27258...rocessor-X6800-4M-Cache-2_93-GHz-1066-MHz-FSB

These were in periods when AMD wasn't very strong (especially 2010). It just seems more like Intel releases more cores at a fairly steady rate whether or not AMD is weak (2010, 2013) or strong (2017).

 

[swilli89]

The fact is SKL-X launch was rushed forward, you just cant rant about not being able to move forward EVERY cpu on the lineup. 12 to 18C SKL-X are petty much coming out on the original date.

Maybe, just maybe, if Intel decided to wait and launch the entire SKL-X lineup when it was originally planned (no TBD), this whole rumor may have never existed. But considering there is now people saying CFL-S is thanks to AMD as well, im going to say this looks more like a religion than anything else at this point.

Again.. no. What we are saying is Intel made a major leap to 18-core SKU which was WAY out of line for them. Then they, after almost an entire decade.. a literal eternity in CPUs, to move from 4-6 core on the mainstream. BOTH happening after AMD has the first competitive product again in almost a.. decade. Its these massive, unavoidable "coincidences" that we are discussing. I'd say to deny those almost impossible coincidences as the real religion of faith.

 

[Glo.]

It's really hilarious that some people think that Intel releases new products at all because of AMD and not because, you know, that's what chip companies do.

If it wasn't for AMD, Intel would never release 4C/8T Core i3's.

And that is best thing that could've happened for the market.

 

[Topweasel]

The fact is SKL-X launch was rushed forward, you just cant rant about not being able to move forward EVERY cpu on the lineup. 12 to 18C SKL-X are petty much coming out on the original date.

Maybe, just maybe, if Intel decided to wait and launch the entire SKL-X lineup when it was originally planned (no TBD), this whole rumor may have never existed. But considering there is now people saying CFL-S is thanks to AMD as well, im going to say this looks more like a religion than anything else at this point.

Just because people say stupid things (like CFL-S being influenced by Ryzen) doesn't mean there is no basis. I mean multiple people board companies didn't even know about the 14c+ options at all and the roadmaps added 12c pretty late once Ryzen rumors started hitting. I am sure either side can turn "rumors" and unofficial "slides" to prove their point.

But it's obvious that Intel made moves they wouldn't have without competition. What those are can be debated on a item by item basis. But either way it cause Intel to do something they haven't done before in paper releasing a product. I mean when it comes to SL-E and SL-SP they had been shipping those for almost a year before they launched. Maybe it was because they screwed up with the timeline and didn't have the specs figured out yet as those would have been binned last. Maybe it's because they didn't have enough stock of chips due to enterprise requirements. Maybe it's because they originally weren't going to offer them. Who knows. But their mere existence has broken a trend they had built up over a nearly 10 year period.

It's not hard to see how competition might have caused the creation of those SKU's when you start adding all of the different weird ways this launch been handled over the last year.

 

[formulav8]

People still don't think AMD had any affect on Intel's plans?

 

[Ajay]

Again.. no. What we are saying is Intel made a major leap to 18-core SKU which was WAY out of line for them. Then they, after almost an entire decade.. a literal eternity in CPUs, to move from 4-6 core on the mainstream. BOTH happening after AMD has the first competitive product again in almost a.. decade. Its these massive, unavoidable "coincidences" that we are discussing. I'd say to deny those almost impossible coincidences as the real religion of faith.

ARGH! This is starting to get annoying! CPUs/SoCs have a four to five year design window! Intel's products where already in the pipeline. The only thing Intel did it change positioning and pricing - same for AMD. That is all they could do at this point. Please stop these stupid unicorn stories

The real competition, in term of architecture - probably started a couple a years ago, when Intel had early AMD samples in hand to analyze. Even then, Intel has it's own design targets based on what it's customers actually need and want. This is the same for both companies. Please don't confuse marketing and sales with engineering. /rage off

 

[Ajay]

People still don't think AMD had any affect on Intel's plans?

Positioning, pricing any to a small extent, timing. That's it. The competition is a Marketing and Sales driven event at this juncture because of the long lead time in the high performance semiconductor market. Going forward there will be some competition based on architectures (because of the maturity of x86 CPUs) and a bit more on process tech, but it will also still be very much customer driven.

 

[inf64]

ARGH! This is starting to get annoying! CPUs/SoCs have a four to five year design window! Intel's products where already in the pipeline. The only thing Intel did it change positioning and pricing - same for AMD. That is all they could do at this point. Please stop these stupid unicorn stories

The real competition, in term of architecture - probably started a couple a years ago, when Intel had early AMD samples in hand to analyze. Even then, Intel has it's own design targets based on what it's customers actually need and want. This is the same for both companies. Please don't confuse marketing and sales with engineering. /rage off

All we are saying is that intel would have never made a jump from 10C BDW-E to 18 SKL-X in one year time span and priced it just 200$ more than 6950X if there wasn't a product like TR 1950X. IT is very simple point we are making. I cannot believe you guys are so blind you cannot understand this point. The MOST we would have gotten without Ryzen and TR would have been 12 or 14C and that is it, probably with 1700 and 2500$ price tags respectively. This way you get 80% more cores than 6950X for just 200$ more, insane.

 

[formulav8]

IIRC CFL-s wasn't originally do out till Q1 '18. SKL-X 3rd/4th Q this year. They were both moved forward. One reason SKL-X was in bad shape when first released. The pricing definitely looks to show where AMD had an affect.

 

[swilli89]

All we are saying is that intel would have never made a jump from 10C BDW-E to 18 SKL-X in one year time span and priced it just 200$ more than 6950X if there wasn't a product like TR 1950X. IT is very simple point we are making. I cannot believe you guys are so blind you cannot understand this point. The MOST we would have gotten without Ryzen and TR would have been 12 or 14C and that is it, probably with 1700 and 2500$ price tags respectively. This way you get 80% more cores than 6950X for just 200$ more, insane.

Exactly, no one is claiming Intel architected in response to a product that has been out not even 6 months. Claiming anyone is saying that is a strawman argument at best. Even still, these Coffee lake "8th" generation chips are literally 6 core Skylakes with iGPU. Process node has improved slighty, like it always has since the beginning of lithography and so now we have "new generations" of products.

Honest question, with all the massive R&D money Intel throws at problems, how long would it take them to take a chip, architected as a 4 core, change nothing to the core itself and add 2 cores? They have the manpower to dedicate a team to this easy. KEEP IN MIND this 6-core skylake was already in development for HEDT.. so really its just adding an iGPU, is THIS the 4-5 year development cycle that is being quoted? 4-5 years to add an existing iGPU module to an existing 6-core monolithic design?

Anyways this has diverged to an argument in completely unprovable semantics. Some gonna say the blatant market conditions changing caused Intel to finally upgrade cores, some will say its just a massive coincidence.

 

[Shivansps]

There was never any rumor like that. The rumors or better said actual slides like the one i linked dealt strictly with core counts. There was no mention of MESH until maybe 2 months ago:

http://wccftech.com/intel-skylake-x-skylake-sp-mesh-architecture-interconnect/

Just for the recapitulation:

Intel originally released info, which pointed toward the possibility of 10C being the top Skylake-X consumer part. Its on that previously linked slide.
Later, rumors came up, that the top HEDT chip will actually be 12 cores, not just 10C. This was reinforced by this photo:

This is the first mention of 12 core Skylake-X part, although as you can see, its properties were unknown yet, which kinda makes it look like afterthought, not surprising given the original Intel slides.
At this point of time, there was no talk about LCC and HCC and whatnot, this stuff came into light later when the whole line-up up to 18C was revealed . There was a bit of confusion, whether 12C will be still LCC, or already HCC, which was later definitely put to rest with the Mesh info.

Finally the Computex slide:

Threadripper was officially mentioned by AMD for the first time on 16th May, 2 weeks before Computex.

http://www.amd.com/en-us/press-releases/Pages/amd-unveils-expanding-2017may16.aspx


this is how it all happened to my knowledge. You are free to dismiss it as fake.

As you said, first we knew about 10C max, them 12C(before Ryzen launch, i think it was about January?), then AMD mentioned TR with 16C, then Computex happen.

Before we knew about computex disscusions on the forums were about how 16C was better than 12C, after Computex, those people jumped on the defensive claiming 12 to 18C cpus were thanks to AMD TR, based on another 2 rumors that turned out to be false. If you dont belive me you are free to search this thread before and after computex.


How lets think about this a little, SKL-X launch was moved forward, it is possible that they were unable to move forward 12C to 18C SKL-X (maybe because of Xeon production) ? To me it looks like a yes here.

Now, lets think a little more 12C was already TBD before Ryzen when we already expected 10C max, it is possible they had more cpus and decided not to show them before official launch? To me is another yes.

Another issue here, from May to September there are just 4 months, it is enoght time to take a Xeon, validate higher clock speeds on a socket that, in theory was never designed for more than 10C, and move production schedules around so they could deliver them on the original date?

One more, if the SKL-X 12 to 18C core launch was created in a panic move, it makes sence to go out and try to validate higher clock speeds than their Xeon models? They already had Xeon SKU validated by that time, whats the point of making SKL-X 12C+ TBD and try to figure out later? What i whould have done in they place if this was done in an hour, is just to release SKL-X with the same clocks as the Xeons. Problem solved. No need to re-validate or show them as TBD at computex, Xeon clock speed is already high enoght.

Now lets do another step forward here, they even need to do what pleople claim they did? They have Xeons targeting the same market like the TR 16C, Intel even need SKL-X up to that many core count because of TR?

 

[dullard]

Nice. Are these published somewhere?

This thread is confusing since it covers far too many chips. But, the current thread title (until it is updated) answers your question: see page 554 of this thread. Specifically see this post:
https://forums.anandtech.com/thread...s-out-page-554.2428363/page-554#post-39005848

There are multiple sources of that (all just reposting the same thing) such as this:
https://www.extremetech.com/computi...lake-cpus-will-require-new-motherboard-socket

Intel usually keeps to their old price tiers, but just gives you more processor for the dollar at about the same price tier than Intel's last chip. If Intel sticks to this decades long pattern, then the 8700K would be about $350.

Compare that to the AMD 1600X that you asked about. The 1600X would be ~$100 cheaper. Both would be 95W TDP (although AMD and Intel define TDP slightly differently). Both would be 6 cores. 12 threads. The base 8700K would be slightly faster than the 1600X. The turbo 8700K would be 7.5% faster than the 1600X. Then you have the IPC differences of the two chips. Note: the 8700K has faster memory than the 7700K, so keep that in mind when you estimate the IPC.

A more interesting comparison would be the 8600K to the 1600X. Both probably will cost about the same amount, both use the same TDP, both have 6 cores. Both have about the same base clock. The 8600K will have 7.5% faster turbo. But the 1600X would have the ability to run 12 threads.

 

[Timmah!]

As you said, first we knew about 10C max, them 12C(before Ryzen launch, i think it was about January?), then AMD mentioned TR with 16C, then Computex happen.

Before we knew about computex disscusions on the forums were about how 16C was better than 12C, after Computex, those people jumped on the defensive claiming 12 to 18C cpus were thanks to AMD TR, based on another 2 rumors that turned out to be false. If you dont belive me you are free to search this thread before and after computex.


How lets think about this a little, SKL-X launch was moved forward, it is possible that they were unable to move forward 12C to 18C SKL-X (maybe because of Xeon production) ? To me it looks like a yes here.

Now, lets think a little more 12C was already TBD before Ryzen when we already expected 10C max, it is possible they had more cpus and decided not to show them before official launch? To me is another yes.

Another issue here, from May to September there are just 4 months, it is enoght time to take a Xeon, validate higher clock speeds on a socket that, in theory was never designed for more than 10C, and move production schedules around so they could deliver them on the original date?

One more, if the SKL-X 12 to 18C core launch was created in a panic move, it makes sence to go out and try to validate higher clock speeds than their Xeon models? They already had Xeon SKU validated by that time, whats the point of making SKL-X 12C+ TBD and try to figure out later? What i whould have done in they place if this was done in an hour, is just to release SKL-X with the same clocks as the Xeons. Problem solved. No need to re-validate or show them as TBD at computex, Xeon clock speed is already high enoght.

Now lets do another step forward here, they even need to do what pleople claim they did? They have Xeons targeting the same market like the TR 16C, Intel even need SKL-X up to that many core count because of TR?

I have no clue how much time it takes to validate higher clock speeds. Perhaps its doable in the time-frame between May and September, i would not know.

Regarding socket, do not LCC and HCC Xeons fit socket 2066, i mean its server version? If yes, then clearly that socket was designed with up to 18 cores in mind. Does not neccessarily mean those HCC chips were intended to be released in consumer space as part of i7 or now i9 line-up. AFAIK you could put 14 core Xeon into X99 board, even though Intel never released i7 with more than 10 cores.

Regarding moving production schedules, perhaps they did not move anything at all. They are just gonna sell part of what they produced and originally intended to sell as Xeons under different name, with higher clocks, lower price and without stuff like ECC support.

Finally, Threadripper is HEDT consumer chip, not a competition to Xeons. Well it could be, if Intel did not have their own HEDT line-up on their own, instead just selling those chips as Xeons. I dont know why it even exists. I really dont care whether i pay for something called Xeon or i7/i9, if the price is same. But clearly Intel cares for whatever reason.

 

[raghu78]

Intel has always used Xeon LCC dies for their HEDT cpus. Intel did not require to use their MCC dies because there was no competition. Intel has responded to strong competition from AMD TR with the use of MCC based HEDT CPUs. Still the very fact that the 14-18 core cpus are launching 1.5 months after TR is a clear indication that Intel had no clue about AMD TR and were caught by surprise.

Sent from my SM-G935V using Tapatalk

 

[Topweasel]

As you said, first we knew about 10C max, them 12C(before Ryzen launch, i think it was about January?), then AMD mentioned TR with 16C, then Computex happen.

Before we knew about computex disscusions on the forums were about how 16C was better than 12C, after Computex, those people jumped on the defensive claiming 12 to 18C cpus were thanks to AMD TR, based on another 2 rumors that turned out to be false. If you dont belive me you are free to search this thread before and after computex.


How lets think about this a little, SKL-X launch was moved forward, it is possible that they were unable to move forward 12C to 18C SKL-X (maybe because of Xeon production) ? To me it looks like a yes here.

Now, lets think a little more 12C was already TBD before Ryzen when we already expected 10C max, it is possible they had more cpus and decided not to show them before official launch? To me is another yes.

Another issue here, from May to September there are just 4 months, it is enoght time to take a Xeon, validate higher clock speeds on a socket that, in theory was never designed for more than 10C, and move production schedules around so they could deliver them on the original date?

One more, if the SKL-X 12 to 18C core launch was created in a panic move, it makes sence to go out and try to validate higher clock speeds than their Xeon models? They already had Xeon SKU validated by that time, whats the point of making SKL-X 12C+ TBD and try to figure out later? What i whould have done in they place if this was done in an hour, is just to release SKL-X with the same clocks as the Xeons. Problem solved. No need to re-validate or show them as TBD at computex, Xeon clock speed is already high enoght.

Now lets do another step forward here, they even need to do what pleople claim they did? They have Xeons targeting the same market like the TR 16C, Intel even need SKL-X up to that many core count because of TR?

The problem is that we don't know when the plans changed. The 12c showed up on the slides in January when vendors were first seeing R7 final silicon. Rumors of the TR hit us in mid March. So maybe even earlier for Intel. So if MCC was already being tested prior to January and sometime in Q1 they changed their plans. Then all of sudden its 6-9 months and not 4. Half a year sounds about right. Remember AMD didn't get Ryzen to Mobo companies till a week or two before the Chinese new year.

You seemed to be focused on the the fact that the because Intel released different products the slides were wrong. When to a lot of people which includes the motherboard manufacturers its seems like they ended up being wrong because Intel released different products. Or better put that the slides were right till Intel decided they needed to change the lineup and not that they were off the entire time.

As for TR vs. SL-EP vs. SL-X. The Xeon lineup isn't competing (though it can compete) in the same market as TR and SL-X. SL-X and TR are about high performance desktops. Not "workstations" and not server work. AMD if they wanted to could fill most of the market just like you are saying with the Xeon lineup with EPYCs. If the rumors are right even the 8c EPYC would still offer great margins. So they would just need mobo companies to make a Desktop SP workstation board. But TR is about performance on top of core count. It has aggressive clocks to match their lower core count brethren which they couldn't do with EPYC. This is what SL-X is about competitive clock speeds using an enterprise chip. These are High performance platform which mean they need to be the jack of all trades. The fact that they didn't know and where still trying to figure out TDP and clockspeeds after announcing the product should tell you that. Which goes back to point 1 that considering the three major points. 1. No reference to the CPU's in the roadmap, 2. The surprise by the mobo makers of their existence. 3 That the specs were not finalized even though they had been producing the chip for a while in the form of SL-EP, that it is more likely than not that their inclusion in the x299 launch was a newer development and not that Intel was always planning it but wasn't ready yet.

Just like today's announcement probably has more to do with Threadripper being days from being available with chips with 60% more cores and Intel needing to tell everyone to remember that they have the 12c+ stuff still coming.

 

[PeterScott]

KEEP IN MIND this 6-core skylake was already in development for HEDT.. so really its just adding an iGPU, is THIS the 4-5 year development cycle that is being quoted? 4-5 years to add an existing iGPU module to an existing 6-core monolithic design?

There he was claiming this was only about 12 to 18C chips and now you are implying Intel went back to the drawing board to tapeout the 6C coffee lake again, but with an IGP, because of Ryzen.

If you want to claim this is only about, 12-18 core HEDT, you really need to stop implying this is also about Coffee Lake.

 

[Zucker2k]

Intel wasn't "trying to figure out TDP and clockspeeds." Intel didn't want to show its hands first to give AMD a target to go after. AMD committed to an earlier date to release threadripper so Intel could afford to play the waiting game. I'm not surprised Intel released final specs soon after AMD released theirs.