[FUDZILLA] No Broadwell for Desktop

[CakeMonster]

I have a feeling its just a new platform with SATA Express. Maybe a speedbin or 2. And Broadwell Xeon support.

Yeah, I guess. Probably.

But I would suspect that the desktop upgrading game won't really stop with this. There will be new CPUs released in the two-year period. And people will want them because of more mature steppings with more OC potential. And if Intel releases new chipsets like they are actually already doing in this two-year cycle we just began, there's yet another reason there.

 

[Phynaz]

In other news, desktop is dead.

 

[CakeMonster]

If we don't feel like upgrading, then yes. But not because of Intel. As I pondered in the post before yours, there will probably be both new CPUs released and new chipsets within the tock-cycle.

 

[Arachnotronic]

It's not normal, that's why they took capacity underutilization charges two quarters in a row.

Weren't you the one still claiming that they were still at 50% utilization even after the excess capacity charges stopped?

 

[blackened23]

If they had capacity issues they would do the exact opposite of their current plan. They would ship LGA processors which sell a miniscule amount to OEMs in comparison to mobile and enterprise products.

But they're not doing that. They're shipping their 14nm processors to the largest segments which ship by far the most volume. But, again, they're shipping to the largest volume markets (MOBILE, ENTERPRISE), and that's that.

 

[blackened23]

In other news, desktop is dead.

Eh. I still feel like intel will shuffle most desktop users to the enthusiast platform beginning with Haswell-E. That's not necessarily a bad thing. Octo core CPUs will be the new baseline for that platform along with other goodies - DDR4 and SATA-E.

Normal joe consumer doesn't buy a desktop, this is obvious - but for existing enthusiasts and PC gamers, that niche of a market isn't really declining either. There's still money to be made there and i'm sure intel will create chips for it, even if the progress is slowed. PC gamers tend to spend a lot of money in the ecosystem, even if the "average user" doesn't buy a desktop PC anymore. As mentioned earlier I believe intel will move most of these users to an E platform which isn't necessarily bad.

 

[Arachnotronic]

Eh. I still feel like intel will shuffle most desktop users to the enthusiast platform beginning with Haswell-E. That's not necessarily a bad thing. Octo core CPUs will be the new baseline for that platform along with other goodies - DDR4 and SATA-E.

I just moved over to LGA 2011. Water's fine.

 

[mohit9206]

2 year desktop refresh is good thing.Mobile and tablets is where the money is.I would say even yearly desktop graphics refreshes are now becoming unnecessary.They should also move to two year refresh period like AMD did with 7000 series.Mobile and laptop graphics need catching up to do.

 

[cbn]

Eh. I still feel like intel will shuffle most desktop users to the enthusiast platform beginning with Haswell-E.

I think that would be a good thing, but I bet the price will still remain high.

What I really hope Intel does with Haswell Refresh is to allow BCLK overclocking (for the mainstream SKUs) using 125 Mhz and 167 Mhz straps. To me that would make up for the lack of 14nm on desktop.

 

[seitur]

Eh. I still feel like intel will shuffle most desktop users to the enthusiast platform beginning with Haswell-E. That's not necessarily a bad thing. Octo core CPUs will be the new baseline for that platform along with other goodies - DDR4 and SATA-E.

Normal joe consumer doesn't buy a desktop, this is obvious - but for existing enthusiasts and PC gamers, that niche of a market isn't really declining either. There's still money to be made there and i'm sure intel will create chips for it, even if the progress is slowed. PC gamers tend to spend a lot of money in the ecosystem, even if the "average user" doesn't buy a desktop PC anymore. As mentioned earlier I believe intel will move most of these users to an E platform which isn't necessarily bad.

For absolute huge majority of desktop users, even for majority of gamers. Intel Extreme is WAY too expensive.

If Extreme is only desktop left, then I won't have Intel desktop anymore. I know many people who have desktops and still uppgrade / replace them every now and then and no single one of them have Intel Extreme and no even a signle person would be willing to spend that kind of money for it.

Remember most desktop gamers are not people who go for Intel E & GTX 780 / Titans or SLI setups.
Most spend for whole unit (case, motherboard, memory, GPU, etc) less than price of i7 Extreme chip itself.

Even on enthusiast forums like this one and on pure PC gaming forums Intel Extreme is very rare.

 

[SPBHM]

they could still be selling their Q1-Q2 2011 CPUs only, without much of a different outcome, so... yes...

 

[blackened23]

For absolute huge majority of desktop users, even for majority of gamers. Intel Extreme is WAY too expensive.

If Extreme is only desktop left, then I won't have Intel desktop anymore. I know many people who have desktops and still uppgrade / replace them every now and then and no single one of them have Intel Extreme and no even a signle person would be willing to spend that kind of money for it.

Remember most desktop gamers are not people who go for Intel E & GTX 780 / Titans or SLI setups.
Most spend for whole unit (case, motherboard, memory, GPU, etc) less than price of i7 Extreme chip itself.

Even on enthusiast forums like this one and on pure PC gaming forums Intel Extreme is very rare.

It's not too expensive unless you're the impression that the 4960X is the only E CPU. There are a lot of E CPUs. Quad cores are around 300$. Quad cores on mainstream are also 300$. Starting with the 4820K, the E quad cores are now fully unlocked which wasn't the case in the past. Rumor has it with Haswell, that will be replaced by a 6 core fully unlocked CPU at 300$. Anyway, I don't feel that's expensive unless you exclusively buy low end stuff like celerons.

 

[Fjodor2001]

So Intel is going all in on mobile with 14 nm to begin with. Then they better start getting some design wins and start selling some CPUs to tablet and mobile phone manufacturers. Because so far ARM rules this segment. Otherwise Intel will be sitting on unused 14 nm capacity...

 

[seitur]

It's not too expensive unless you're the impression that the 4960X is the only E CPU. There are a lot of E CPUs. Quad cores are around 300$. Quad cores on mainstream are also 300$. Starting with the 4820K, the E quad cores are now fully unlocked which wasn't the case in the past. Rumor has it with Haswell, that will be replaced by a 6 core fully unlocked CPU at 300$. Anyway, I don't feel that's expensive unless you exclusively buy low end stuff like celerons.

You spend too much on enthusiast forum.

Most desktop gamers don't use i7, any i7 on whichever platofrm. Most desktop users don't buy GTX 780 price point GPUs. Most desktop gamers don't buy 250$ + motherboards.

You are sorely mistaken if any but small niche group is gonna accept premium price point of FCLGA2011, even at it lowest current price point.

 

[Arachnotronic]

So Intel is going all in on mobile with 14 nm to begin with. Then they better start getting some design wins and start selling some CPUs to tablet and mobile phone manufacturers. Because so far ARM rules this segment. Otherwise Intel will be sitting on unused 14 nm capacity...

ARM doesn't sell a single chip into tablets and phones.

It licenses IP.

 

[cbn]

For absolute huge majority of desktop users, even for majority of gamers. Intel Extreme is WAY too expensive.

If Extreme is only desktop left, then I won't have Intel desktop anymore. I know many people who have desktops and still uppgrade / replace them every now and then and no single one of them have Intel Extreme and no even a signle person would be willing to spend that kind of money for it.

Remember most desktop gamers are not people who go for Intel E & GTX 780 / Titans or SLI setups.
Most spend for whole unit (case, motherboard, memory, GPU, etc) less than price of i7 Extreme chip itself.

Even on enthusiast forums like this one and on pure PC gaming forums Intel Extreme is very rare.

Yes, I agree about the price.

I think Intel would need to include some lower end SKUs for the Extreme Desktops socket in order to make that kind of transition work.

For example on LGA 2011 there are hexcore cpus and a i7 quad core available. For future Extreme desktops I think Intel would also have to include a i5 quad core at the minimum.

Even then I think it would still be a very expensive platform.....probably too expensive for most. (There would definitely need to be some cheaper motherboards thrown in to help out.)

 

[Puppies04]

I can only hope that decoupling the desktop and mobile chip release will enable intel to stop prioritizing per/watt and give us enthusiasts what we want.

I can only hope...

 

[ShintaiDK]

I can only hope that decoupling the desktop and mobile chip release will enable intel to stop prioritizing per/watt and give us enthusiasts what we want.

I can only hope...

Since servers want performance/watt. There is no way its gonna change. The only difference is simply that desktop will be a year behind node wise and only release every second year.

 

[Fjodor2001]

ARM doesn't sell a single chip into tablets and phones.

It licenses IP.

Correct. You get the point though, right? It doesn't matter if they license IP or sell CPUs. If they're dominating that segment it means Intel doesn't. Unless that changes Intel could end up sitting on unused 14 nm capacity if they initially are going all in on mobile with 14 nm.

 

[ShintaiDK]

Correct. You get the point though, right? It doesn't matter if they license IP or sell CPUs. If they're dominating that segment it means Intel doesn't. Unless that changes Intel could end up sitting on unused 14 nm capacity if they initially are going all in on mobile with 14 nm.

Even if we imagine Airmont fails, even tho Silvermont is already a huge success with OEMs. The 14nm capacity would simply create desktop Broadwells to fit in 9 series boards. Intel is not skipping 14nm for desktop because they got excess capacity, its because they dont have enough capacity.

 

[Arachnotronic]

Even if we imagine Airmont fails, even tho Silvermont is already a huge success with OEMs. The 14nm capacity would simply create desktop Broadwells to fit in 9 series boards. Intel is not skipping 14nm for desktop because they got excess capacity, its because they dont have enough capacity.

Ding, ding, ding. Nobody needs an LGA Broadwell. CPU improvements will likely be minimal, clock speed headroom won't really be there for the OC'ers, and the LGA version probably wouldn't get the best IGP.

Why not just keep selling Haswells for mainstream, then HSW-E for top end?

 

[Fjodor2001]

Even if we imagine Airmont fails, even tho Silvermont is already a huge success with OEMs. The 14nm capacity would simply create desktop Broadwells to fit in 9 series boards. Intel is not skipping 14nm for desktop because they got excess capacity, its because they dont have enough capacity.

But still, they must have reserved 14 nm capacity for CPUs intended for mobile phones and tablets. Now if that push into the tablet / mobile phone space fails Intel will be sitting on unused 14 nm capacity, right? Or do you think they suddenly will re-arrange their launch plans completely and introduce a so far unplanned line of Broadwell desktop CPUs in a couple of months time to make use of such unused 14 nm capacity?

 

[cbn]

Ding, ding, ding. Nobody needs an LGA Broadwell. CPU improvements will likely be minimal, clock speed headroom won't really be there for the OC'ers, and the LGA version probably wouldn't get the best IGP.

Why not just keep selling Haswells for mainstream, then HSW-E for top end?

Yeah, it sounds like Intel will proabably optimize the FinFET for low leakage on 14nm even more than they did for Haswell.

Which brings us to the 22nm process on Haswell Refresh? Will this refresh 22nm bring us back the overclocking headroom we lost from Ivy Bridge? (Will the Refresh Haswell have a FinFET design optimized more towards drive current rather than low leakage?)

 

[Insert_Nickname]

Ding, ding, ding. Nobody needs an LGA Broadwell. CPU improvements will likely be minimal, clock speed headroom won't really be there for the OC'ers, and the LGA version probably wouldn't get the best IGP.

^^This.

If the only improvement is ~5% IPC, no OC headroom and a slightly better IGP, then its something I can live without.

Wasn't really planning on upgrading until Skylake anyway.

 

[ShintaiDK]

But still, they must have reserved 14 nm capacity for CPUs intended for mobile phones and tablets. Now if that push into the tablet / mobile phone space fails Intel will be sitting on unused 14 nm capacity, right? Or do you think they suddenly will re-arrange their launch plans completely and introduce a so far unplanned line of Broadwell desktop CPUs in a couple of months time to make use of such unused 14 nm capacity?

OEMs order CPUs in advance, not day to day like you do in a shop.