[Canard PC Hardware] Intel prepares Ryzen's response behind the scenes

[scannall]

I don't expect Intel will do much more than juggle the SKU's a bit, announcing 'New and Improved!'. They won't take a big hit on their margins. They really don't need to. They have the 'mindshare', and the marketing clout to pretty much bully the market into submission. They will likely lose a bit of market share to Ryzen, but probably not a lot of sales, as there is a lot of pent up demand for more cores that are unwilling to pay the Intel tax. Early sales of Ryzen in particular will likely be to people unwilling to pay Intel's price for HEDT parts.

 

[frozentundra123456]

That's your opinion, please don't state it as fact.



Define "competitive"?



Huh. So you assume the absolute worst case for Intel ("Skylake-X is HEDT for 24 months"), the best case for AMD ("Zen+ based Summit Ridge successor coming in 1Q 2018) and then conclude that Intel's competitive position will be poor. I think you might want to rethink your assumptions.



Only in your opinion. If you look at what has actually been happening, Intel has been putting out new processors for notebooks, servers, and desktops at a regular pace. They did screw up by not bringing Broadwell to desktop, that was a case of a poor decision that came back to haunt them (though they recovered by getting Skylake out when it was supposed to come out -- more or less), but to say that Intel has been complacent when their R&D spending is at all time highs and they are taking actions to make sure to put out new products each year (14nm+/Kaby Lake is a good example) is just unfair.

I urge you to look at the situation with a little more objectivity.



OK, so first of all, I never argued that 10nm is in good shape, I actually think it's in pretty poor shape right now. But let's talk about Coffee Lake. If Coffee Lake is similar to what we saw with Kaby Lake, then we should see Intel improve the performance of its transistors again (14nm++) and the leaks show that they are planning a 6 core die. How is this "complacent" if they are adding more cores, tweaking the process, and probably tweaking the physical design to potentially improve maximum frequency, too?

It seems to me that they are going to deliver a product that will drive performance in desktops and high performance notebooks pretty nicely up. Working to build better products each year does not signal complacency.

As far as AMD "tocks" you are again introducing assumptions that I don't think are necessarily given. In AMD's mind Bristol Ridge was a "tock". Do you know what Bristol Ridge was? An optimization of Carrizo, in the exact same way that Kaby Lake was an optimization of Sky Lake. They improved the transistor performance (working with GloFo on this, obviously) and it is likely that they had to redo the physical design to support the enhanced 28nm process.

I also see you are now referring to Ice Lake as mid-2019 rather than late 2019 as I have seen you post in the past -- an improvement, but I think once again you are being overly pessimistic about Intel's product launches. Intel seems to be executing to a "one product family every year" cadence. Kaby Lake in early 2017, Coffee Lake in early 2018. Why would Ice Lake be mid-2019? Based on the release cadence, early 2019 seems more likely and late 2018 if they want to be super aggressive.

I don't know if they are conplacent, greedy, or just made bad decisions, but they should have had a hex core on the mainstream desktop for sure with skylake, or even sooner. I have argued this a long time in this forum against those who are saying HEDT is enough. I think this lame response to Zen shows I was correct. I mean, who cares about a slight clock speed bump in an overclockable chip? It might look good in marketing and stock benchmarks but that is all.

 

[Arachnotronic]

I don't know if they are conplacent, greedy, or just made bad decisions, but they should have had a hex core on the mainstream desktop for sure with skylake, or even sooner. I have argued this a long time in this forum against those who are saying HEDT is enough. I think this lame response to Zen shows I was correct. I mean, who cares about a slight clock speed bump in an overclockable chip? It might look good in marketing and stock benchmarks but that is all.

Agreed they should have debuted a six core with Kaby Lake, but perhaps given the time line they had to work on it, it just wouldn't have been feasible without too much risk to the schedule.

As far as the speed bump, like I said, the 7740K itself isn't likely a response to the Zen products but the 7640K, if it has HT enabled, likely is. The 7740K is there just to make buying an i7 worthwhile even in the presence of an HT enabled 7640K.

 

[LTC8K6]

I don't know if they are conplacent, greedy, or just made bad decisions, but they should have had a hex core on the mainstream desktop for sure with skylake, or even sooner. I have argued this a long time in this forum against those who are saying HEDT is enough. I think this lame response to Zen shows I was correct. I mean, who cares about a slight clock speed bump in an overclockable chip? It might look good in marketing and stock benchmarks but that is all.

There was no visible reason for a hex-core mainstream processor. The unlocked hex core HEDT chips did not run games faster.

Even now the justification for it will probably be slim, and mostly market-speak for a while.

It seems like the 4c/8t chips at high clocks with fast ram will probably remain Intel's gaming kings for a while yet.

Unless Intel has something up their sleeve...

 

[Atari2600]

So Intel has no ability to respond with new products? New SKUs?

One or two additional SKU at the top of the i5 and i7 (non HEDT) line won't come even close to alleviating the pricing/marketshare pressure that an any way competitive (i.e. even with -10% on IPC and -10% on clocks) Zen will bring to the market.

Surely you understand that?

The topic of this thread is about the new SKUs that Intel is supposedly launching, and then when Skylake-X/Kaby Lake-X come out, Intel has quite a few options available to it.

When they come out, maybe 6 months down the line... and how much do they bring over BWe?

 

[LTC8K6]

Agreed they should have debuted a six core with Kaby Lake, but perhaps given the time line they had to work on it, it just wouldn't have been feasible without too much risk to the schedule.

As far as the speed bump, like I said, the 7740K itself isn't likely a response to the Zen products but the 7640K, if it has HT enabled, likely is. The 7740K is there just to make buying an i7 worthwhile even in the presence of an HT enabled 7640K.

I would guess that the 7740K and 7640K will be in short supply and priced unattractively. To me, they seem very unlikely to be responses to RyZen.
Plus, the 7640K already will be down to 6Mb of cache, being an i5. For some reason people seem to like more cache over less.

 

[LTC8K6]

So Intel has no ability to respond with new products? New SKUs? The topic of this thread is about the new SKUs that Intel is supposedly launching, and then when Skylake-X/Kaby Lake-X come out, Intel has quite a few options available to it.

I'm certain that Intel has the ability to respond with new products, but I'm concerned that they really won't alter their road map very much at all. I think they believe they have RyZen covered well enough.

 

[dacostafilipe]

PCGamesHardware.de posted CPCHardware news this morning and they just added some more information based on their own sources.

http://www.pcgameshardware.de/Kaby-...ntels-Antwort-Core-i7-7740K-i5-7640K-1220177/

The i7-7740K and i5-7640K do exists, but they are 112W Kaby Lake-X (Socket 2066) that should be released for Gamescom. It's the same Kaby-Lake-S die "just" with a disabled iGPU. TIM will still be used.

There will be a 100 Mhz base clock improvement on the 7740K, but the turbo should stay the same @ 4.5Ghz.

 

[Atari2600]

The i7-7740K and i5-7640K do exists, but they are 112W Kaby Lake-X (Socket 2066)

That is interesting.

From 91W to 112W with 100 MHz increase? Any wonder they cannot push another 100 MHz if they have already specified minimum 130W cooler for the line.

 

[coercitiv]

Don't know about Intel, but the forums are indeed in full panic mode

 

[[DHT]Osiris]

Hyperthreading i5's ? Yea no, I dont believe it . That would signal death of at least i3 and down .. or resegmenting ala :
i3 = 4c/4t
i5 = 4c/8t
i7 = coffeelake++
HEDT : 8 cores+

Came here to post this. I think Intel will do a full SKU shift to stay competitive on core counts. i5 is the new i7, i7 will be shifted to 6c+HT or 8c+HT. Either one of the 10/12/16c 'extreme' versions will collapse or they'll push a little further (20c?)

 

[sm625]

The i7-7740K and i5-7640K do exists, but they are 112W Kaby Lake-X (Socket 2066) that should be released for Gamescom. It's the same Kaby-Lake-S die "just" with a disabled iGPU.

This makes no sense. When has Intel ever released an i5 for a HEDT platform?

 

[LTC8K6]

This makes no sense. When has Intel ever released an i5 for a HEDT platform?

Maybe it's i7-7640K and i7-7740K for HEDT?

 

[Burpo]

Making changes. They just EOL'd Haswell-E..

http://www.pcgameshardware.de/Haswe...re-i7-5820K-5930K-und-5960X-sind-EOL-1220185/

Core i7-5960X, the i7-5930K and the i7-5820K. Of those you have to say goodbye, because Intel has classified the models as EOL (End Of Life). As usual with Intel, the processors can be ordered for a while by dealers, but in principle is now concluded. The latest orders are received by Intel until August 2017,

 

[LTC8K6]

They will deliver them until Feb 2018, so quite a while yet.

 

[raghu78]

That's your opinion, please don't state it as fact.

We will see how reality pans out. I am betting Intel's revenues and profits in Client Computing and probably even DCG will fall in 2017. In 2018 it will get worse. Any improvement or turnaround is 2019 or later.

Define "competitive"?

seriously you think Intel is invincible ? Anyway I think AMD Ryzen CPUs will be within 10% of max Kabylake ST performance . I expect IPC to be in between Broadwell and Skylake and clocks to max out around 4.5-4.6 Ghz with good cooling. Intel 7700k will face serious competition from a 8C/16T SKU with 3.4 base / 3.8 Ghz turbo as it will be priced around USD 400. Stock performance will be significantly in 7700k's favour but clock both these to max and you are going to see the Ryzen 8C come close behind in ST while murdering it in MT.

Huh. So you assume the absolute worst case for Intel ("Skylake-X is HEDT for 24 months"), the best case for AMD ("Zen+ based Summit Ridge successor coming in 1Q 2018) and then conclude that Intel's competitive position will be poor. I think you might want to rethink your assumptions.

What assumption. Aug 2015- Skylake launches. Aug 2017 - No tock. Aug 2018 - No tock (heck if they cannot get 10nm Cannonlake on desktops in 2018 whats the hope for a 10nm tock). Its going to be roughly 4 years between Skylake and the next tock. Lets see Intel update their roadmap on Feb 9th Investor update meeting and then we will talk about a 10nm tock Icelake in 2018. More importantly if you look at the past Intel's HEDT and Xeon launches atleast a year after their mainstream SKUs. So yeah I think Intel does not have a Skylake-X replacement till late 2019. btw AMD CTO confirmed they have the tocks lined up. Its not speculation. AMD have already showed with Bulldozer - > Piledriver -> Steamroller -> Excavator that process node changes do not affect their steady IPC improvements.

Only in your opinion. If you look at what has actually been happening, Intel has been putting out new processors for notebooks, servers, and desktops at a regular pace. They did screw up by not bringing Broadwell to desktop, that was a case of a poor decision that came back to haunt them (though they recovered by getting Skylake out when it was supposed to come out -- more or less), but to say that Intel has been complacent when their R&D spending is at all time highs and they are taking actions to make sure to put out new products each year (14nm+/Kaby Lake is a good example) is just unfair.

I urge you to look at the situation with a little more objectivity.

dude there are others like canard PC and semiaccurate who have said Intel's 10nm position is not so good. Their product roadmap is largely uninspiring. My question to you is what the hell they have been doing since Aug 2015 and when their next tock Icelake will land sometime in 2019. There are many reasons for their situation . But mostly arrogance , complacency and a bad CEO.

OK, so first of all, I never argued that 10nm is in good shape, I actually think it's in pretty poor shape right now.

Something we can finally agree on.

But let's talk about Coffee Lake. If Coffee Lake is similar to what we saw with Kaby Lake, then we should see Intel improve the performance of its transistors again (14nm++) and the leaks show that they are planning a 6 core die. How is this "complacent" if they are adding more cores, tweaking the process, and probably tweaking the physical design to potentially improve maximum frequency, too?

It seems to me that they are going to deliver a product that will drive performance in desktops and high performance notebooks pretty nicely up. Working to build better products each year does not signal complacency.

dude Coffeelake is a a reactive response to Zen and whats coming ahead. Intel should have had 6 cores ready by now if they were proactive.

As far as AMD "tocks" you are again introducing assumptions that I don't think are necessarily given. In AMD's mind Bristol Ridge was a "tock". Do you know what Bristol Ridge was? An optimization of Carrizo, in the exact same way that Kaby Lake was an optimization of Sky Lake. They improved the transistor performance (working with GloFo on this, obviously) and it is likely that they had to redo the physical design to support the enhanced 28nm process.

Bristol Ridge was not a tock. It had the same Excavator core as Carrizo core but just speeds were better due to process maturity. Moreover Bristol Ridge was a mild update to remain as a placeholder till Raven Ridge landed in late 2017.

I also see you are now referring to Ice Lake as mid-2019 rather than late 2019 as I have seen you post in the past -- an improvement, but I think once again you are being overly pessimistic about Intel's product launches. Intel seems to be executing to a "one product family every year" cadence. Kaby Lake in early 2017, Coffee Lake in early 2018. Why would Ice Lake be mid-2019? Based on the release cadence, early 2019 seems more likely and late 2018 if they want to be super aggressive.

I do not know when Intel will release Ice-lake. But I do think it will be 2019. If they get their 10nm process in good shape in 2018 and get aggressive they could hit Q1 2019 otherwise it could end up mid 2019 or later. The early signs are 10nm has problems some of which are engineering and some management. Lets see how soon Intel can fix those.

 

[Asterox]

Don't know about Intel, but the forums are indeed in full panic mode

Intel is in panic mode no doubt, or Bud Spencer is Ryzen CPU lineup vs Intel CPU lineup.

 

[Face2Face]

Making changes. They just EOL'd Haswell-E..

http://www.pcgameshardware.de/Haswe...re-i7-5820K-5930K-und-5960X-sind-EOL-1220185/

Core i7-5960X, the i7-5930K and the i7-5820K. Of those you have to say goodbye, because Intel has classified the models as EOL (End Of Life). As usual with Intel, the processors can be ordered for a while by dealers, but in principle is now concluded. The latest orders are received by Intel until August 2017,

Looks like I'll have to get Antique plates for my 5820K. I guess it will become a family heir loom...

 

[mikk]

What a fail from Canard PC, these SKUs are made for Kabylake-X!!!

https://benchlife.info/intel-kaby-lake-x-with-core-i7-7740k-and-core-i5-7640k-x299-02072017/


It's no surprise for me, it didn't make sense for S1151.

 

[richierich1212]

What a fail from Canard PC, these SKUs are made for Kabylake-X!!!

https://benchlife.info/intel-kaby-lake-x-with-core-i7-7740k-and-core-i5-7640k-x299-02072017/


It's no surprise for me, it didn't make sense for S1151.

The sad thing is these should have been on LGA1151 instead of the 7700K and 7600K.

 

[LTC8K6]

What a fail from Canard PC, these SKUs are made for Kabylake-X!!!

https://benchlife.info/intel-kaby-lake-x-with-core-i7-7740k-and-core-i5-7640k-x299-02072017/


It's no surprise for me, it didn't make sense for S1151.

Both i7 as I suspected. The title has i7-7640K but the chart has i5-7640K, but I think both are i7.

 

[dacostafilipe]

What a fail from Canard PC

It seems like the media tries to hard to get some kind of information ...

Luckily this will end in a few days

 

[Arachnotronic]

We will see how reality pans out. I am betting Intel's revenues and profits in Client Computing and probably even DCG will fall in 2017. In 2018 it will get worse. Any improvement or turnaround is 2019 or later.

You are free to bet on whatever you would like

seriously you think Intel is invincible ? Anyway I think AMD Ryzen CPUs will be within 10% of max Kabylake ST performance . I expect IPC to be in between Broadwell and Skylake and clocks to max out around 4.5-4.6 Ghz with good cooling. Intel 7700k will face serious competition from a 8C/16T SKU with 3.4 base / 3.8 Ghz turbo as it will be priced around USD 400. Stock performance will be significantly in 7700k's favour but clock both these to max and you are going to see the Ryzen 8C come close behind in ST while murdering it in MT.

It seems that you think AMD is invincible, that it will execute perfectly after years of disappointments, while Intel will execute extremely poorly (as in the worst it has ever executed).

What assumption. Aug 2015- Skylake launches. Aug 2017 - No tock. Aug 2018 - No tock (heck if they cannot get 10nm Cannonlake on desktops in 2018 whats the hope for a 10nm tock). Its going to be roughly 4 years between Skylake and the next tock. Lets see Intel update their roadmap on Feb 9th Investor update meeting and then we will talk about a 10nm tock Icelake in 2018. More importantly if you look at the past Intel's HEDT and Xeon launches atleast a year after their mainstream SKUs. So yeah I think Intel does not have a Skylake-X replacement till late 2019. btw AMD CTO confirmed they have the tocks lined up. Its not speculation. AMD have already showed with Bulldozer - > Piledriver -> Steamroller -> Excavator that process node changes do not affect their steady IPC improvements.

Believe what you'd like.

dude there are others like canard PC and semiaccurate who have said Intel's 10nm position is not so good. Their product roadmap is largely uninspiring. My question to you is what the hell they have been doing since Aug 2015 and when their next tock Icelake will land sometime in 2019. There are many reasons for their situation . But mostly arrogance , complacency and a bad CEO.

Do you think Intel told its CPU architects to go for a multi-year vacation? Do you really think that any company with just even baseline competence -- not talking Steve Jobs excellence, but just basic competence -- would, after hearing in May 2015 directly from AMD that it is building Zen (and all of the major performance info was known -- 40% more IPC than XV, SMT, 8 cores, etc.) -- would just sit around and do nothing in response?

That's on top of the many ARM-based SoC makers like Qualcomm, Cavium, Broadcom, HiSilicon, governments, etc. all announcing their own server SoC efforts targeting Intel, with ARM saying that they expect to take 25% marketshare in servers by 2020.

Come on, this defies common sense and basic logic. You know this.

dude Coffeelake is a a reactive response to Zen and whats coming ahead. Intel should have had 6 cores ready by now if they were proactive.

Coffee Lake is Intel's next annual product release. Not everything in the computing universe revolves around AMD. Intel needs to put out new products each year because that's what OEMs want and that's one way to get people to buy new stuff. Do you think that Apple releases new iPhones every year just because some other phone companies are planning new phones? Of course not! They put out new phones each year because when you bring new products to the market, you increase your chances of getting people to buy new products.

Bristol Ridge was not a tock. It had the same Excavator core as Carrizo core but just speeds were better due to process maturity. Moreover Bristol Ridge was a mild update to remain as a placeholder till Raven Ridge landed in late 2017.

Bristol Ridge was more than just process maturity. GloFo and AMD worked to enhance the 28nm process' performance by around 10%, which is no small feat. And since the process was changed, their circuit design/physical implementation teams had to do work to land the uArch onto the new process. Bristol Ridge delivered more performance/watt not because of process maturity but because of active process and circuit enhancements.

 

[hojnikb]

What a fail from Canard PC, these SKUs are made for Kabylake-X!!!

https://benchlife.info/intel-kaby-lake-x-with-core-i7-7740k-and-core-i5-7640k-x299-02072017/


It's no surprise for me, it didn't make sense for S1151.

That makes no sense whatsoever. Why would they bother making less than 6 cores for 2066 ?

 

[Arachnotronic]

What a fail from Canard PC, these SKUs are made for Kabylake-X!!!

https://benchlife.info/intel-kaby-lake-x-with-core-i7-7740k-and-core-i5-7640k-x299-02072017/


It's no surprise for me, it didn't make sense for S1151.

Ouch, they REALLY failed here.