6 core mainstream Haswell?

[BTRY B 529th FA BN]

Will there be mainstream 6 core Haswell chips or will that still be for the high end market?

 

[ShintaiDK]

On LGA1150? Nein.

 

[cmdrdredd]

One of the reasons I bought Ivy was because there was no roadmap for consumer level hexacore cpus. I was trying to hold out for that, but it seems it won't happen for quite a long time (AMD doesn't count right now).

 

[BTRY B 529th FA BN]

Daaang!

 

[BrightCandle]

No point really, it would be a waste of transistors. Lots of users haven't worked out what to do with 1 core let alone the 4 they currently get given.

 

[Red Hawk]

I don't think so, and the main reason is lack of competition. Intel's processors grind AMD into the ground at single-threaded performance, and its hyperthreading technology keeps it on an even playing field with AMD's dual-integer unit strategy. Intel has no reason to design and sell a mainstream hexacore processor as long as its quad core processors are sufficient for mainstream tasks and superior to AMD processors overall.

Were AMD to actually start effectively competing with Intel, things would change. One of Intel's logical responses would be to start selling mainstream hexacore processors. I mean, that's what AMD did in response to Intel's Core resurgence, right? Until that day (and hopefully, it will happen) Intel will happily continue selling quad-core processors at the mainstream level.

 

[BTRY B 529th FA BN]

Well hopefully thermal temps while overclocking will be a bit better on Haswell.

 

[ShintaiDK]

I don't think so, and the main reason is lack of competition. Intel's processors grind AMD into the ground at single-threaded performance, and its hyperthreading technology keeps it on an even playing field with AMD's dual-integer unit strategy. Intel has no reason to design and sell a mainstream hexacore processor as long as its quad core processors are sufficient for mainstream tasks and superior to AMD processors overall.

Were AMD to actually start effectively competing with Intel, things would change. One of Intel's logical responses would be to start selling mainstream hexacore processors. I mean, that's what AMD did in response to Intel's Core resurgence, right? Until that day (and hopefully, it will happen) Intel will happily continue selling quad-core processors at the mainstream level.

If you noticed AMD stays with 4 threads to in all APUs. Including the nextgen after Trinity.

It just make no sense to give people cores to idle with. The transistors are better used elsewhere. Just because a handful want cheap hexcores doesnt mean the hundres of millions other buyers do.

And you can always buy HW-E if you need a Haswell based hexcore.

 

[ShintaiDK]

Well hopefully thermal temps while overclocking will be a bit better on Haswell.

Maybe. But the chance is high.

Haswell is specificly designed for 22nm HKMG Trigates unlike Ivy. But you gonna see the same with Broadwell as you do with Ivy today I bet.

 

[Red Hawk]

If you noticed AMD stays with 4 threads to in all APUs. Including the nextgen after Trinity.

It just make no sense to give people cores to idle with. The transistors are better used elsewhere. Just because a handful want cheap hexcores doesnt mean the hundres of millions other buyers do.

And you can always buy HW-E if you need a Haswell based hexcore.

AMD is probably banking on APU/GPU acceleration like OpenCL becoming more widespread, allowing them to use their superior GPU component to make up for their lackluster CPU performance. It remains to be seen whether or not that strategy will ultimately pay off. Increasing core count does have a tangible benefit now, however. Encoding and other productivity tasks are now generally well multithreaded, games are increasingly more multithreaded, and even modern web browsers like IE9 and Google Chrome are multithreaded nowadays.

You could buy Haswell-E (or whatever the Sandy Bridge-E successor is called), but the OP was wondering if Intel will release a mainstream hexacore processor. Haswell-E would not be a mainstream processor; it would likely be much like Sandy Bridge-E, a stripped down server processor adapted for desktop use. Part of Sandy Bridge-E's expense (aside from Intel setting whatever price they want because AMD can't compete at that level) is its nature as a server part. Several somewhat costly provisions are made that wouldn't be necessary or desirable if it was designed as desktop chip. That's the way things are, and Intel has no reason to change that unless AMD steps up its game.

 

[lopri]

Probably not. I'm almost tempt to pick up another 1045T from MicroCenter for $99 so that I can make profit in near future.

 

[BTRY B 529th FA BN]

Probably not. I'm almost tempt to pick up another 1045T from MicroCenter for $99 so that I can make profit in near future.

How so?

 

[moonbogg]

6 core mainstream? No way in hell. Not for another 3-4 years I bet. I think I am screwed now though as I won't want to "settle" for less than 6 cores since thats what i have now. Its a psychological thing.

 

[T_Yamamoto]

6 core mainstream? No way in hell. Not for another 3-4 years I bet. I think I am screwed now though as I won't want to "settle" for less than 6 cores since thats what i have now. Its a psychological thing.

Yes exactly. If Intel did. They'd price gorge the hell out of it

 

[ShintaiDK]

Lets try it from a different angle.

People on the desktop/laptop wanting hexcores can be counted in the 100Ks. People wanting dual or quadcores in the 100Ms.

 

[lopri]

Lets try it from a different angle.

People on the desktop/laptop wanting hexcores can be counted in the 100Ks. People wanting dual or quadcores in the 100Ms.

That's indeed a good argument for the uslessness of HyperThreading.

 

[Smoblikat]

No you wont see any compettitive hexa cores for a while, the last was thuban from AMD and after that the only choice would be the overpriced SB-E.

 

[Red Hawk]

That's indeed a good argument for the uslessness of HyperThreading.

No, it's more like a good argument for hyperthreading. Hyperthreading does provide real benefits in heavily multithreaded applications. With hyperthreading, you add multithreading performance to a chip without actually making a more expensive chip with more transistors. You can sell the chip with hyperthreading disabled to people who don't need that multithreaded performance, and then charge a price premium to those who want the benefits of hyperthreading. It's not actually more expensive to make the chip, but you're making more profit. Those 100k people who want more multithreaded performance will pay more than if you only had the non-hyperthreading chip available.

And that's only regarding the quadcore/hexacore space. Down at the dual core level hyperthreading provides a pretty essential boost at minimal cost, and it's one of Intel's major advantages over AMD at the budget level.

 

[Smoblikat]

No, it's more like a good argument for hyperthreading. Hyperthreading does provide real benefits in heavily multithreaded applications. With hyperthreading, you add multithreading performance to a chip without actually making a more expensive chip with more transistors. You can sell the chip with hyperthreading disabled to people who don't need that multithreaded performance, and then charge a price premium to those who want the benefits of hyperthreading. It's not actually more expensive to make the chip, but you're making more profit. Those 100k people who want more multithreaded performance will pay more than if you only had the non-hyperthreading chip available.

And that's only regarding the quadcore/hexacore space. Down at the dual core level hyperthreading provides a pretty essential boost at minimal cost, and it's one of Intel's major advantages over AMD at the budget level.

You just prooved how greedy intel is, they disable HT on the I5's which costs MORE than leaving it alone and sell the I7's for 100$+ more for a "feature" that is inherent to the design from the get-go.

 

[Kristijonas]

You just prooved how greedy intel is, they disable HT on the I5's which costs MORE than leaving it alone and sell the I7's for 100$+ more for a "feature" that is inherent to the design from the get-go.

In capitalism, it's sugarcoated as 'marketing'

 

[Edrick]

No you wont see any compettitive hexa cores for a while, the last was thuban from AMD and after that the only choice would be the overpriced SB-E.

Overpriced to who? you?

They seem within reason to me based on the past 10 years of CPU pricing.

 

[greenhawk]

You just prooved how greedy intel is, they disable HT on the I5's which costs MORE than leaving it alone and sell the I7's for 100$+ more for a "feature" that is inherent to the design from the get-go.

except that ignores the market segmentation/pricing intel aim for, it ignores the ability for intel to sell failed chips at a lower price and ingnores the benifit of HT vs a real core.

 

[Homeles]

Maybe. But the chance is high.

Haswell is specificly designed for 22nm HKMG Trigates unlike Ivy. But you gonna see the same with Broadwell as you do with Ivy today I bet.

Matured process that's probably tweaked by the time Haswell rolls out, plus what you listed above. It's almost 100% that leakage will be cut significantly by then.

Overclocking may or not improve though -- Haswell promises "improved overclocking," but we have no idea what Intel means by that. Maybe they removed the multiplier only garbage? Maybe it'll reach higher clocks? It's a new architecture (although not totally new), so there's too many unknowns.

Anyways, this is why Haswell will be a quad core: http://www.anandtech.com/show/5174/why-ivy-bridge-is-still-quad-core

Basically, it's unlikely to see a mainstream 6 core until after Broadwell or Skylake. Intel's still focusing on moving to an SoC. Until they complete that transition, I wouldn't expect a 6 core, as they're still focusing on cramming more crap on the same die -- there's simply no die real estate for another 2 cores + L3 when you're already tacking on other junk. I wouldn't call it greed or anything... honestly Intel's focus is more important. If you need more than 4 cores for the time being, pay extra or go AMD. Hopefully Piledriver makes the latter option a sensible one.

 

[lopri]

Snip

Why are you quoting me for something that I have not said?

 

[Borealis7]

here's an idea!

IVBs are quite small and heat up quickly due to flux, so...cram in "2 moar coars" and increase the die size to improve heat flux! yey!