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

[unseenmorbidity]

Please tell me why Intel (or any company) would want to compete on price.

How has competing on price worked out for AMD the last decade or so?

Never said they would... I said the opposite.

Intel will throw it's monopoly power around as always, and the brain dead consumer will buy their overpriced products because they don't have a choice and/or they are ignorant.

My point was that this isn't a product "response" (A product that will compete with AMD's zen), but a marketing "response".

 

[VirtualLarry]

I see Intel "sponsoring" some benchmark reviews, after Ryzen comes out, by seeding reviewers with 6850K/6900K CPUs, as well as 7700K CPU, and coming out with a "reviewers guide", that causes the benchmarks to paint Intel's CPUs in the best possible light. Basically, taking a page from NVidia's playbook.

 

[arandomguy]

CPU reviewers could do with review guides from either company as I find that CPU gaming reviews are really poorly done for the most part.

If Intel suggests CPU reviews test Fallout 4 at locations like Corvega or Overwatch at high FPS for competitive gaming or AMD suggests testing titles like Watch Dogs 2 in worst case scenarios such as on the bridge or using actual settings people play with, not low, this is much more useful in highlighting the effect of CPU and related component performance in gaming than the majority of tests currently being used.

If AMD is going to push core count as advantage they should try to showcase why that might make gaming better based on how actual real gamers are not going to be closing out every single background task while gaming, streaming while gaming is now a growing and very common activity. Why should they not suggest in a reviewers guide that reviews highlight this advantage of their CPU?

Likewise if Intel high ST performance will still be relevant for people playing competitively looking for max FPS maybe htey should push to show why think 60fps+ or even 144fps+ is important and can only be achievable on their CPUs.

In either case while those are bias work loads that would likely paint each companies product in a better light they are actually relevant work loads.

Or I guess we can have more reviews spam things like AOTS because of all those people looking forward to CPU upgrades to play that game. Or how about any of those titles tested in locations and settings where the CPU basically makes no difference. I know what I will be looking for in any CPU purchase for gaming is how well they can run titles at 640x480 at minimum settings, oh yeah!

 

[Arachnotronic]

I see Intel "sponsoring" some benchmark reviews, after Ryzen comes out, by seeding reviewers with 6850K/6900K CPUs, as well as 7700K CPU, and coming out with a "reviewers guide", that causes the benchmarks to paint Intel's CPUs in the best possible light. Basically, taking a page from NVidia's playbook.

Do you think AMD isn't going to provide a reviewers guide for Ryzen?

 

[LTC8K6]

Will intel have something price competitive when Zen 2 comes around? I mean they may not make a major move now, but not doing anything when Zen 2 comes out would seem odd right? Unless intel intends to move the hexacore into mainstream.
Still though, that octocore not being mainstream makes it Zen 2 and beyond look strong.

I thought Intel's chips were way overpriced? This was the recurring story long before Zen info came out.
It means that Intel can easily compete on price right now, right?

 

[realibrad]

Then how is he refuting "hardly", which in english means "just barely". Its not a true/false statement.

In this context it is used as a way of saying false. Look at the 3rd bullet point for the Google definition.

no or not (suggesting surprise at or disagreement with a statement

So in this context he is using it to denote he disagrees with the statement.

 

[tential]

I thought Intel's chips were way overpriced? This was the recurring story long before Zen info came out.
It means that Intel can easily compete on price right now, right?

Not necessarily. It doesn't mean that the average consumer is educated yet.

So it may not be a good price to performance chip by Intel. But it may (and most likely will) sell well enough that it would hurt Intel at the revenue line to make the chip price competitive since the additional sales gained wouldn't make up the revenue lost per chip.

This is what puts amd at a good position is they can take market share of the most informed customers who tend to set trends for other customers.

 

[Phynaz]

I must be reading "hardly" to mean something different then.

I saw it as refuting my claim, as is the definition of hardly as "With little likelihood".

 

[Phynaz]

In this context it is used as a way of saying false. Look at the 3rd bullet point for the Google definition.

no or not (suggesting surprise at or disagreement with a statement

So in this context he is using it to denote he disagrees with the statement.

Beat me to it

 

[Phynaz]

Never said they would... I said the opposite.

Intel will throw it's monopoly power around as always, and the brain dead consumer will buy their overpriced products because they don't have a choice and/or they are ignorant.

My point was that this isn't a product "response" (A product that will compete with AMD's zen), but a marketing "response".

Considering CPUs take about four years to design, what do you think an appropriate product response would be?

As a corollary, why would Intel even respond to a product that isn't even on the market yet?

 

[Arachnotronic]

Considering CPUs take about four years to design, what do you think an appropriate product response would be?

People seem to forget this. For example, the folks who say, "Oh, Intel got lazy with Haswell/Skylake because no competition," they forget that these products came out in, what? 2013 and 2015 respectively?

Haswell and Skylake had roughly 5 year development cycles and the CPU cores themselves were defined likely in the 2008 and 2010 timeframes (and pretty much frozen at those points), respectively. Neither of them had likely even taken Bulldozer into consideration when they were defined.

What's also interesting is that Zen went into development in ~2012, so AMD had the benefit of knowing exactly what Sandy/Ivy Bridge brought to the table and I believe Intel had even disclosed the full architecture of Haswell (and David Kanter had a full writeup on it, complete with performance expectations):

http://www.realworldtech.com/haswell-cpu/

 

[LTC8K6]

Not necessarily. It doesn't mean that the average consumer is educated yet.

So it may not be a good price to performance chip by Intel. But it may (and most likely will) sell well enough that it would hurt Intel at the revenue line to make the chip price competitive since the additional sales gained wouldn't make up the revenue lost per chip.

This is what puts amd at a good position is they can take market share of the most informed customers who tend to set trends for other customers.

I think all of these grand pronouncements about Intel and AMD need to wait until the fact ship docks and unloads.

 

[LTC8K6]

Considering CPUs take about four years to design, what do you think an appropriate product response would be?

As a corollary, why would Intel even respond to a product that isn't even on the market yet?

Yes, Haswell was demonstrated in September of 2011. So you can figure when it must have been designed.
If AMD has caught up to Haswell/Broadwell, that's good and all, but let's keep things in perspective.

 

[Arachnotronic]

Yes, Haswell was demonstrated in September of 2011. So you can figure when it must have been designed.
If AMD has caught up to Haswell/Broadwell, that's good and all, but let's keep things in perspective.

Yep, it was demo'd at IDF 2011 by Mooly Eden, and considering that it takes quite a bit of time to go from RTL to physical implementation, then it takes another ~3 months to run the wafer through the fab, you are really looking at CPUs that were rolling off the line in June-ish 2011, and physical implementation probably began in June-ish of 2010, meaning that the actual core microarchitecture was in the works beginning sometime in 2009 and finished by June 2010.

Haswell is ancient and was designed long before Intel knew how Bulldozer would perform. If anything, it was probably designed based on Intel projecting out steady improvements from AMD off of the Phenom II baseline.

 

[lolfail9001]

Ay there folks, you are not seriously implying that Intel has some follow-up designed, frozen and waiting to be actually done in silicon for few years already. Are you? Because if you are then Intel's mode of not updating architecture with process updates makes even less sense.
OTOH, Skylake was out so quickly after Broadwell quad precisely because Intel were waiting to get 14nm rolling, but nonetheless.

 

[raghu78]

Arachnotronic I would say looking at Zen's 4 year development timeline for a clean sheet design it would take roughly same time or lesser for incremental architectural enhancements like Haswell->Skylake. Nonetheless the problem for Intel is the gaping hole in terms of a lack of tock or IPC improvements from Aug 2015 when Skylake launched to early 2019 when Icelake launches (roughly 3.5 years). This gap is what is going to allow AMD to catch up as they have yearly tocks lined up for 2018-2020. We are likely to see a Zen+ on 14 LPP in Q1 2018 and a Zen++ on GF 7nm in H1 2019.

http://www.anandtech.com/show/11117/globalfoundries-to-expand-capacities-build-a-fab-in-china

"As the company is preparing to start high-volume manufacturing (HMV) of chips using its 7 nm FinFET technology in the second quarter of next year (so, several months ahead of the plan), the actual output of the Fab 8 remains to be seen. Initially, GlobalFoundries plans to use deep ultraviolet (DUV) lithography with quadruple patterning to produce chips using its 7 nm process, but sometime in 2019 it intends to start using extreme ultraviolet (EUV) lithography for a new wave of 7 nm designs. Usage of EUV will not eliminate multi-/quadruple-patterning, but will be used for cirical layers and will thus help to increase output of leading-edge chips. At present, the company does not talk about its 7 nm capacity, but it is logical to assume that the current expansion will have a positive effect on it as well."

Its quite possible we see 7nm GPUs from AMD in Q4 2018 and 7nm CPUs in H1 2019. Lisa Su confirmed that they are designing currently for 7nm. AMD would have done 2 architectural generations by the time Icelake launches in just a span of 2 years. Compare that to Intel which would have done a tock in 3.5 years and you know thats a major problem going forward especially with this tick->tock->process optimization(n) where n>1.

The advantage for Intel all along from 2006 till now was that once AMD fell behind Core 2 and Nehalem widened the gap it turned into outright disaster for AMD with Bulldozer. Intel was under no competitive pressure for the last 5 years and even in the 2006-2011 timeframe Intel was firmly ahead. Intel is going to have to rethink their strategy and go back to being very aggressive on tocks.

 

[Arachnotronic]

Arachnotronic I would say looking at Zen's 4 year development timeline for a clean sheet design it would take roughly same time or lesser for incremental architectural enhancements like Haswell->Skylake. Nonetheless the problem for Intel is the gaping hole in terms of a lack of tock or IPC improvements from Aug 2015 when Skylake launched to early 2019 when Icelake launches (roughly 3.5 years). This gap is what is going to allow AMD to catch up as they have yearly tocks lined up for 2018-2020. We are likely to see a Zen+ on 14 LPP in Q1 2018 and a Zen++ on GF 7nm in H1 2019.

http://www.anandtech.com/show/11117/globalfoundries-to-expand-capacities-build-a-fab-in-china

"As the company is preparing to start high-volume manufacturing (HMV) of chips using its 7 nm FinFET technology in the second quarter of next year (so, several months ahead of the plan), the actual output of the Fab 8 remains to be seen. Initially, GlobalFoundries plans to use deep ultraviolet (DUV) lithography with quadruple patterning to produce chips using its 7 nm process, but sometime in 2019 it intends to start using extreme ultraviolet (EUV) lithography for a new wave of 7 nm designs. Usage of EUV will not eliminate multi-/quadruple-patterning, but will be used for cirical layers and will thus help to increase output of leading-edge chips. At present, the company does not talk about its 7 nm capacity, but it is logical to assume that the current expansion will have a positive effect on it as well."

Its quite possible we see 7nm GPUs from AMD in Q4 2018 and 7nm CPUs in H1 2019. Lisa Su confirmed that they are designing currently for 7nm. AMD would have done 2 architectural generations by the time Icelake launches in just a span of 2 years. Compare that to Intel which would have done a tock in 3.5 years and you know thats a major problem going forward especially with this tick->tock->process optimization(n) where n>1.

The advantage for Intel all along from 2006 till now was that once AMD fell behind Core 2 and Nehalem widened the gap it turned into outright disaster for AMD with Bulldozer. Intel was under no competitive pressure for the last 5 years and even in the 2006-2011 timeframe Intel was firmly ahead. Intel is going to have to rethink their strategy and go back to being very aggressive on tocks.

Do you think that the competitive analysis teams at Intel are completely clueless? Do you think they have been both:

1. Twiddling their thumbs for years, telling the architecture people to go enjoy a nice multi-year nap; and
2. When AMD said back in 2015 that they are doing a new core with +40% IPC over XV, SMT, etc. that Intel just didn't do any planning at all?

Come on.

 

[Justinbaileyman]

I am very surprised that Intel didnt just launch a 6c/12t cpu for the new Z270 chipset to counter AMD... They still could but I dont see it happening..

 

[Arachnotronic]

I am very surprised that Intel didnt just launch a 6c/12t cpu for the new Z270 chipset to counter AMD... They still could but I dont see it happening..

Honestly, I think such a SKU would have been really nice, and it would have allowed Intel to essentially transition the i5 lineup to 4C/8T and then the i3 lineup to 4C/4T.

But there are probably good reasons for Intel to not have done that.

So, for example, if you make the die sizes bigger across the board, you need more capacity in place to be able to produce enough chips to meet demand. Intel may simply not have enough 14nm capacity in place right now to do that. It gets easier for Intel when they transition a reasonable portion of its product line to 10nm in 2018 (Cannon Lake-Y and possibly low-end Cannon Lake-U). With some of the unit volume shifted from 14nm -> 10nm, some 14nm capacity naturally is left unused. By making the dies a little bit bigger with four cores (4+2 and 4+3e) in U and 6+2 in H and 6+2/4+2 in S, they can expand their 14nm wafer usage to fill up the preexisting capacity.

Another thing to keep in mind is that if you want more cores they are not free from a power perspective, especially if you want to be able to have them all fully-loaded at reasonable frequencies. Maybe Intel wanted another generation of transistor performance improvements (as well as more time for its physical design teams to optimize the implementation for improved perf/watt over the Kaby Lake dies) before six cores were feasible in a 45W notebook power envelope (or 4 cores in 15W/28W).

It's easy as enthusiasts for us to sit back and tell these companies what we want and when we want it, but there are technical and economic realities that these companies need to cope with.

 

[unseenmorbidity]

Considering CPUs take about four years to design, what do you think an appropriate product response would be?

As a corollary, why would Intel even respond to a product that isn't even on the market yet?

I am well aware of that, but that isn't the implication of the linked article.

I expect them to do what they are doing, blow smoke up our anal cavities. I just hoped at least some people would be able to see it for what it is.

I mean this will be the 3rd time they released this cpu, and they just released kabylake.

Do you think that the competitive analysis teams at Intel are completely clueless? Do you think they have been both:

1. Twiddling their thumbs for years, telling the architecture people to go enjoy a nice multi-year nap; and
2. When AMD said back in 2015 that they are doing a new core with +40% IPC over XV, SMT, etc. that Intel just didn't do any planning at all?

Come on.

There is nothing to hardware related that they can plan... They cannot wish new hardware into existence, and intel is almost hopelessly stuck on 14nm. The only "response" they have to amd and samsung catching up is marketing, and abusing their power as a monopoly. I am sure they made those"plans"...

They should have dropped the dual core i3 years ago, but they were making too much money on fools that kept perpetuating and rewarding them for their monopoly.

And now they are caught with their pants down, and their "plan" is to go back to abusing their power, until they finally do what they should have done years ago, if they didn't have the market by the balls, and offer more cores for the mainstream market.

 

[Mechanical Man]

Do you think that the competitive analysis teams at Intel are completely clueless? Do you think they have been both:

1. Twiddling their thumbs for years, telling the architecture people to go enjoy a nice multi-year nap; and
2. When AMD said back in 2015 that they are doing a new core with +40% IPC over XV, SMT, etc. that Intel just didn't do any planning at all?

Come on.

Nobody sane will believe such thing. But there really is truth in it. Having pressure to do and not having pressure to do is really big difference for many people and thus for many organizations.

Many intelligent persons need some pressure to be able to work at their greatest.

 

[Arachnotronic]

There is nothing to plan... They cannot wish new hardware into existence, and intel is almost hopelessly stuck on 14nm. The only "response" they have to amd and samsung catching up is marketing, and abusing their power as a monopoly. I am sure they made those"plans"...

The wonderful thing about opinions is that everybody in the world is entitled to them.

 

[PingSpike]

I am very surprised that Intel didnt just launch a 6c/12t cpu for the new Z270 chipset to counter AMD... They still could but I dont see it happening..

Would this even be possible without changing the socket?

 

[unseenmorbidity]

The wonderful thing about opinions is that everybody in the world is entitled to them.

There won't be a 10nm solution for pc for years to come, so yah, they are hopelessly stuck on 14nm. And as we saw with kabylake, they cannot squeeze additional ipc out of it either. Now clock speeds are likely maxed too.

The only trick left in the bag is more cores, and they don't want to do that because they have been grooming the consumers to accept ever higher prices. And their pricing is largely based on more cores being exponentially more expensive.

 

[raghu78]

Arachnotronic however big a company is its basically run by very few people at the highest level, the most important of whom is the CEO. They make all the important decisions which determine the future and direction of a company. We have seen CEOs like Hector Ruiz destroy AMD due to sheer arrogance and complacency and plain bad business decisions (like overpaying for ATI). Brian Krzanich seems to be one of those kinds according to canard pc who recently wrote about the general mood and state of projects within Intel. charlie also hinted at some major problems some of which are management related.

http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=thread...-to-issues-with-intel’s-10nm-process.2495934/

You might want to believe everything is fine at Intel but I think the next few years will point out how bad the problems are.