XBitlabs: Intel Cans Rockwell Processors for Desktops - Spy-Shot From Intel’s Roadmap

[AtenRa]

Not at all: http://wccftech.com/intel-haswell-desktop-cpu-prices-leaked-flagship-core-i74770k-cost-327/

Look at the 4770S and 4570S. They cost $280 and $189 respectively. The clocks are similar to 4770R and 4670R. The 4770R and 4670R would be just sold at $340 and $240 price ranges.

They are look like direct replacement to those parts, with similar clocks and identical TDP. The S parts are used in some AIO systems. Rather than using a discrete part with an S CPU, you'd use the R part with the integrated HD 5200 graphics. The OEMs decide whether the extra cost is worth it.

Mobile Haswell H series: So it looks like the H series mobile Haswell with HD Graphics 5200 is a replacement to 3615QM, 3612QM, and 3610QM.

The 4770S and 4570S has the GT2(20 EUs) HD4600 small die, the R series will have the GT3 (40 EUs) 5200 big die. If they will sell them at the same price, margins will fall. GT3e will even cost more because of the added Memory and die size.

 

[IntelUser2000]

The 4770S and 4570S has the GT2(20 EUs) HD4600 small die, the R series will have the GT3 (40 EUs) 5200 big die. If they will sell them at the same price, margins will fall. GT3e will even cost more because of the added Memory and die size.

They won't sell them at same price. They'll sell them at $340 and $240 price ranges which is approximately $50 higher.

Also, there's only one version of GT3 on the quad core chips. The non High Speed Memory integrated parts are only in the Ultrabooks.

HD 5200-GT3e in quad cores
HD 5100-GT3 in ULT
HD 5000-GT3 in ULT lower performance
HD 4600-GT2 in SV dual and quad
HD 4400-GT2 in ULT and Y series

 

[AtenRa]

They won't sell them at same price. They'll sell them at $340 and $240 price ranges which is approximately $50 higher.

And that's exactly my point, the R-series(GT3) which have lower CPU performance than GT2 parts, will be priced higher than the GT2 parts. Since the GT2 parts are already in a higher price category than Kaveri's sub $150.00, how exactly will higher priced Haswell R-series will compete against Kaveri ??

 

[lagokc]

And that's exactly my point, the R-series(GT3) which have lower CPU performance than GT2 parts, will be priced higher than the GT2 parts. Since the GT2 parts are already in a higher price category than Kaveri's sub $150.00, how exactly will higher priced Haswell R-series will compete against Kaveri ??

Different customers:

Haswell quad-core GT2: people that want a long lasting powerful computer but don't play games

Kaveri: gamers on a budget / eccentrics building ITX HTPCs that can play some games and want the smallest case possible

Haswell quad-core GT2 plus standalone GPU: gamers that have a bit more money

Haswell quad-core GT3e: idiots who listened to the sales guy at Best Buy

 

[Phynaz]

Intel socket 2011 has no competition from AMD and yet it counts for only one(1) percent of Intel CPU sales, does that make me wrong as well ??

Having no competition in a given segment and CPU sales in general is two different things.

Nice attempt at deflecting. Do you have any evidence that socket 2011 isn't a financial success for Intel? I didn't think so.

So let's use your new argument, since you say having no competition in a segment doesn't guarantee sales success, why did you bring it up to begin with? Because it certainly implies the you believe Kaveri will be successful because Intel can't compete with it.

You are saying that lack of competition in a market segment can only be an advantage for AMD but not for Intel.

And you never answered the question why the name given to a process node has anything to do with cpu sales.

All in all, your arguments fall apart.

 

[Phynaz]

And that's exactly my point, the R-series(GT3) which have lower CPU performance than GT2 parts, will be priced higher than the GT2 parts. Since the GT2 parts are already in a higher price category than Kaveri's sub $150.00, how exactly will higher priced Haswell R-series will compete against Kaveri ??

But wait, how do Intel's lower performing (according to you) parts compete against Kaveri today? Pretty well considering AMD continues to lose market share.

 

[tynopik]

I don't see how it will be any different than buying a laptop. Not saying it will be better, or even good, but surely no worse?

oh great, make the desktop buying experience just as 'wonderful' as the laptop buying experience, super

 

[Idontcare]

oh great, make the desktop buying experience just as 'wonderful' as the laptop buying experience, super

Personally I am not in favor of it, it will limit enthusiasts in ways that are not desirable.

But the rest of the market, the other 98%, probably won't even notice or care. And since that is where Intel gets the bulk of its profits from, can we really be all that surprised if they pursue it?

 

[IntelUser2000]

And that's exactly my point, the R-series(GT3) which have lower CPU performance than GT2 parts, will be priced higher than the GT2 parts. Since the GT2 parts are already in a higher price category than Kaveri's sub $150.00, how exactly will higher priced Haswell R-series will compete against Kaveri ??

R-series is IMO a strategic move in part by Intel. The R series have a lower 65W TDP just like the S series which go into AIOs. It's still a relatively small market, but a fastest growing one for desktops. Of course I never said it'll be a direct competitor to theoretically dirt cheap AMD competition.

 

[dastral]

Haswell quad-core GT3e: idiots who listened to the sales guy at Best Buy

This would actually be quite nice for balanced high end classy laptops (aka Macbook).
Not having a GPU could be a bonus (TDP/Chasis Design).

I am the only one seeing Intel merging a 25W CPU and a 25W GPU and selling them as a 35W package ?

I dont see why GT3 should have a lower CPU perf than GT2....
At max load of course, but you'll pay for the potential to have the best of both worlds
They must have an idea we havent thought of...

 

[IntelUser2000]

What's hard to understand? TDP is a worst case scenario. If you have a more powerful iGPU, you need to cut the CPU performance down a bit.

 

[NTMBK]

The R series is probably ideal for the next gen iMacs. Apple is always looking for a way to slim them down without killing performance, and replacing the discrete laptop GPU it currently has with Haswell iGPU would be a good way of doing that.

Also, Intel makes a shedload of money from Socket 2011. The entire Xeon E5 series is socket 2011.

 

[dastral]

What's hard to understand? TDP is a worst case scenario. If you have a more powerful iGPU, you need to cut the CPU performance down a bit.

Of course, however i see this the other way around : you can have either but not both.
I see Intel offering a i5-3570 (77W) and a 7750 (75W) in a 75W Package.

You have an incredibly high ceiling (for CPU or GPU perf) depending on your need (game, flash, photoshop, browsing, productivity, encoding) with an well managed power consumption (power gating).

I expect this type of integration to be more efficient (power wise) than 2 chips which do not communicate directly (or do so via the OS/Drivers).

 

[Pilum]

I can't believe that we'll see no LGA Broadwells at all. This would mean no Xeon E3 v4 refresh. The 1290v2 sells at $885, and I don't see how Intel can not produce successors to these kind of SKUs. The profits are just too immense considering that it's a ~160mm² die.

So they'll probably be running a LGA production line, at least for the high-end die. I don't think it would be much of an effort to also run K-series i7/i5 off that line as well. So I don't really see a reason not to produce at least high-end 14nm LGA SKUs. It just doesn't hurt anywhere, and profitability will be higher due to smaller die sizes.

What may happen to the lower SKUs is a different matter. But this will depend on volume planning for the different products. Intel has to ramp 14nm quickly, that's a fundamental part of the whole tick-tock paradigm. And if 14nm volume doesn't go into a desktop switchover, they must have plans on a really huge push into mobile in 2014/2015. Or they start serious foundry work @14nm which will soak up volume.

 

[NTMBK]

I can't believe that we'll see no LGA Broadwells at all. This would mean no Xeon E3 v4 refresh. The 1290v2 sells at $885, and I don't see how Intel can not produce successors to these kind of SKUs. The profits are just too immense considering that it's a ~160mm² die.

So they'll probably be running a LGA production line, at least for the high-end die. I don't think it would be much of an effort to also run K-series i7/i5 off that line as well. So I don't really see a reason not to produce at least high-end 14nm LGA SKUs. It just doesn't hurt anywhere, and profitability will be higher due to smaller die sizes.

What may happen to the lower SKUs is a different matter. But this will depend on volume planning for the different products. Intel has to ramp 14nm quickly, that's a fundamental part of the whole tick-tock paradigm. And if 14nm volume doesn't go into a desktop switchover, they must have plans on a really huge push into mobile in 2014/2015. Or they start serious foundry work @14nm which will soak up volume.

Look at Westmere. They did 4 core Xeons, but no 4 core i7s.

 

[Pilum]

Look at Westmere. They did 4 core Xeons, but no 4 core i7s.

Yep, but it was a different SKU makeup back then. I don't think there was something like the E3s, which are dual-use between desktop and low-end server. And with WST they converted much of Xeon production to 32nm, including 6c, which accounts for some serious wafer volume. But this simply isn't the case here, as E5 gets moved to newer processes only after one year.

Well, it is possible they'll do a 2c 14nm BGA push into low-end desktop, and use the rest of the volume for attacks into mobile, while everything else stays on 22nm. This may just be the beginning of a "new Intel" which goes for mobile/low-end first, instead of putting the priority on the mainstream.

 

[Ajay]

Look at Westmere. They did 4 core Xeons, but no 4 core i7s.

Westmere-EP was Gulftown, which came out before the desktop version in an odd twist.

 

[uribag]

But wait, how do Intel's lower performing (according to you) parts compete against Kaveri today? Pretty well considering AMD continues to lose market share.

Bang & Olufsen might have the best Smart TV... What´s their marketshare? I know, and you also know that product quality is not the only factor of marketshare distribution.

Speaking for myself, as i´m no CPU a enthusiast, I think Kaveri will be a nice product. My daily use of a computer (and maybe more than 90% of people): internet, word, excel, autocad (just reading and light modifications on projects) and other "light" programs. It´s good to have a product that will do all of that (because even a Pentium D does) and play some games in free time too.