Intel: Tick-Tock cycle dead, now instead "Process-Architecture-Optimization"

[tsk]

5 hours ago, DocSwag said:

One thing I am wondering-other silicon chip manufacturers like TSMC seem to still be somewhat following Moore's law. While Intel has postponed 10nm to 2017 and 7nm to at least 2019 and most likely 2020, TSMC has instead said it will begin 10nm this year and have 7nm in 2018. Intel has always had the superior manufacture process at each node, but this probably won't hold true when comparing TSMC 7nm vs Intel 10nm. If Intel is moving to a 3 year cycle while companies like TSMC are not, could this mean Intel will start to fall behind when it comes to manufacturing process? Only time will tell, I guess.

No and no.

From your own source;

 

Quote

The manufacturing lead Intel has had over the past few years over rivals such as Samsung, TSMC and Global Foundries, has put them in a commanding position in both home computing and enterprise. One could argue that by elongating the next two process nodes, Intel might lose ground on their advantage, especially as other companies start hitting their stride. However, the research gap is still there - Intel introduced 14nm back in August 2014, and has since released parts upwards of 400mm2, whereas Samsung 14nm / TSMC 16nm had to wait until the launch of the iPhone to see 100mm2 parts on the shelves, with Global Foundries still to launch their 14nm parts into products. While this relates to density, both power and performance are still considered to be on Intel’s side, especially for larger dies.

 

[patrickjp93]

Just now, huilun02 said:

And then Zen is good and Intel has second thoughts...

Zen won't change Intel's mind at all. It already has stark competition from IBM Power 8 and Oracle Sparc 7 in ultra-high performance compute in financial algorithms and huge databases.

[patrickjp93]

2 minutes ago, huilun02 said:

Neither of which are for consumers

Which is the least important market for Intel (discounting the new focus on IOT). Intel needs its new server parts out as fast as it can make them, and yet they are still the last products deployed in their stack because of validation time and getting yields up. You're making assumptions which are entirely invalid when making your conclusion. And every innovation Intel makes for servers is applicable to consumer software. The problem is the developers not keeping up with new instructions (or hell, multithreading until the most recent years).

[patrickjp93]

7 minutes ago, huilun02 said:

And the only thing important for me here

Read the edits. Everything Intel does to innovate for servers ends up validated in consumer chips anyway, so you're getting the innovation with or without AMD.

[SImoHayha]

I want 1nm CPU's next

[LordArcturus]

6 hours ago, Maybach123 said:

thats a given rule like never try to invade Russia  

Always a natural rule for something, Russia is never invade, period, especially in winter, Intel should never be underestimated, I think Intel just did this to motivate more companies to become competition and force advances faster, it would make sense, especially as it also motivates your own employees to make better and greater advances themselves.

 

If someone can understand that, you can speak Namikaze apparently, cuz I have issues wording my sentences to not sound like useless drivel XD.

[shdowhunt60]

1 hour ago, patrickjp93 said:

Zen won't change Intel's mind at all. It already has stark competition from IBM Power 8 and Oracle Sparc 7 in ultra-high performance compute in financial algorithms and huge databases.

Right. Zen won't "change" Intel's minds at all in those sectors. While I have faith that Zen will be a good architecture, Intel is still going to pull ahead by sheer virtue of Intel STILL having a better process and a more heavily optimized architecture to boot. AMD simply cannot match Intel, they just don't have the means to.

 

AMD knows this. That's why they're targeting the consumer desktop market first.

[Tenelia]

Moore's Law was never a law in the first place. I remember sitting through Computing and Electronics History, learning that it was just a product development strategy statement that was taken way way way out of context...

[DocSwag]

10 hours ago, tsk said:

No and no.

From your own source;

 

 

Yes, I know, but I am looking into the future. I know Intel currently still holds the lead in manufacturing, and will most likely continue to at each node, but my point is if Intel moves to a 3 year cycle while the others are still on a two year cycle, won't TSMC/Samsung/GloFo eventually pass Intel? The main reason they aren't creating big parts yet is cause no one is demanding them yet. With the new GPUs though we can expect larger dies to come out of GloFo/Samsung/TSMC.

 

Intel currently holds the lead, but that may only be for a few more years.

[patrickjp93]

8 hours ago, shdowhunt60 said:

Right. Zen won't "change" Intel's minds at all in those sectors. While I have faith that Zen will be a good architecture, Intel is still going to pull ahead by sheer virtue of Intel STILL having a better process and a more heavily optimized architecture to boot. AMD simply cannot match Intel, they just don't have the means to.

 

AMD knows this. That's why they're targeting the consumer desktop market first.

And in that case it just comes down to a price war if Intel wants to keep an iron grip on the market.

[T8z5h3]

honestly,

X86_64 is dead and it's because of arm.

even Microsoft knows that.... we the gamer do not.

[Coaxialgamer]

10 hours ago, Ramamataz said:

I want 1nm CPU's next

i don't think we will ever get there unfortunately

[DocSwag]

25 minutes ago, Coaxialgamer said:

i don't think we will ever get there unfortunately

I think I saw somewhere that carbon nanotubes can be used as transistors and that they can be 1nm but I could be wrong.

[Coaxialgamer]

14 minutes ago, DocSwag said:

I think I saw somewhere that carbon nanotubes can be used as transistors and that they can be 1nm but I could be wrong.

maybe , but carbon nanotubes and similar technologies won't be ready when we reach the limits of silicon.

 

Edit : "CARBON NANOTUBE TRANSITORS" *lifts glasses*

[Coaxialgamer]

37 minutes ago, T8z5h3 said:

honestly,

X86_64 is dead and it's because of arm.

even Microsoft knows that.... we the gamer do not.

it's clearly not dead , as the vast majority of programs are written for it. You could argue that ARM is superior to x86_64 , but x86 is in no way dead.

[DocSwag]

9 minutes ago, Coaxialgamer said:

maybe , but carbon nanotubes and similar technologies won't be ready when we reach the limits of silicon.

True... still under development right now.

Of course that whatever germanium compound could work for 7nm and stuff.

[Coaxialgamer]

1 minute ago, DocSwag said:

True... still under development right now.

Of course that whatever germanium compound could work for 7nm and stuff.

I meant even Si based substrates like silicon germanium , so they probably wont be ready when we reach 5nm .

 

I personnally have faith in GaN as the next material , but i think we will eventually have to move away from current tech in favor of different ways to process logic ( ternary logic would be cool )

[SurvivorNVL]

1 hour ago, T8z5h3 said:

honestly,

X86_64 is dead and it's because of arm.

even Microsoft knows that.... we the gamer do not.

If we're lucky--we might see Itanium make a come-back.  

[patrickjp93]

1 hour ago, T8z5h3 said:

honestly,

X86_64 is dead and it's because of arm.

even Microsoft knows that.... we the gamer do not.

ROFL!!! Oh jeez my sides! No, x86 is far from dead and is still the performance leader EVERYWHERE but ultra low-power mobile where ARM reigns and transactional workloads that Power 8 and Sparc 7 lead in. Even tablets are now Intel's market to lose. And because of that 16-core Xeon D, Google is now no longer looking into using Qualcomm's chips for its email servers. x86 is still a force ARM can't reckon with outside of a small handful of markets like embedded systems and ULP Mobile.

 

46 minutes ago, Coaxialgamer said:

it's clearly not dead , as the vast majority of programs are written for it. You could argue that ARM is superior to x86_64 , but x86 is in no way dead.

It's only superior in one aspect, and Intel's closing the efficiency gap with every generation. ARM's performance is not growing at nearly the same rate as Intel is making efficiency gains.

[Coaxialgamer]

1 minute ago, patrickjp93 said:

 

It's only superior in one aspect, and Intel's closing the efficiency gap with every generation. ARM's performance is not growing at nearly the same rate as Intel is making efficiency gains.

exactly my point.

[LokiFire]

AMD ZEN releases.

 

 

[Tomsen]

Intel's tick-tock model is long buried dead in the ground. The tick-tock-toe model is already in place. We saw it with haswell, and the "devil's canyon" following.

 

14 hours ago, patrickjp93 said:

I'm not a photoshopper or even a user of Paint, so here's my two cents.

Process (get the new node producing the tweaked old architecture well, add in new graphics, boost up core count for servers and possibly enthusiasts)

Architecture (introduce the new one with new demanded feature sets, even if there isn't time to tune up clock speeds or power, add new graphics if ready, possibly add cores)

Optimization (shave instruction latencies, introduce more instructions as demanded, tune up clock speed, add new graphics if ready)

 

And pray IBM doesn't get a massive breakthrough at the wrong time.

In regards to your remark on optimization, you're not going to see changes in instruction latency. The core is the same (Only changes would be fixing some bug). I wont expect any new extentions, not already supported. If you are thinking of the individual SKUs having more "unlocked" features, like we saw with devil's canyon, then sure.

 

The changes will most likely be in uncore, and process and library optimizations. Intel will be fabbing bigger dies. Also be slightly more aggressive with the clock-speed.

That means higher core-count and clock-rate for servers, and higher clock-rate for consumers (and perhaps more "unlocked" features). 

[Coaxialgamer]

2 minutes ago, Tomsen said:

_snip_

I know this won't happen probably , but it would be great if intel released a 6 core mainstream SKU.

[SurvivorNVL]

2 minutes ago, Coaxialgamer said:

I know this won't happen probably , but it would be great if intel released a 6 core mainstream SKU.

I would prefer them keep the 4-cores and instead introduce their 4-way multi-threading and inverse-multi-threading for specific applications.  4c/16t and then inverse multi-threading for some specific applications.  That would be wonderful.

[Coaxialgamer]

Just now, SurvivorNVL said:

I would prefer them keep the 4-cores and instead introduce their 4-way multi-threading and inverse-multi-threading for specific applications.  4c/16t and then inverse multi-threading for some specific applications.  That would be wonderful.

wait. They have quad SMT in the works ?

 

But doing that would mean doing major changes to the core , and they arent going to.