No Broadwell Desktop CPUs

Posted:
in Future Apple Hardware edited January 2014

I found this story that there will be no Broadwell CPUs for the desktop, just for tablets, laptops, and other mobile devices.

 

The Story link:  http://www.fudzilla.com/home/item/32524-broadwell-won%E2%80%99t-make-it-to-desktop

 

What do you guys think of the implacations, especially for the iMac/Apple desktops.  I think, that the Haswell-E will still come out next year, along with a Haswell refresh for desktops.

Comments

  • Reply 1 of 9

    That’s complete and utter idiocy. I refuse to believe that.

  • Reply 2 of 9
    wizard69wizard69 Posts: 13,377member
    That’s complete and utter idiocy. I refuse to believe that.

    I don't know but I tend to believe it. This isn't the first time I've heard of Haswell refresh. The reality is this, Intel is loosing big time in mobile. So I could see them stressing mobile heavily over the next year. They need design ins.

    Look at it this way, Apple will be shipping an enhanced A7 for its tablets soon. The current A7 comes close to and sometimes betters Intels best Atom effort. If Apple follows past behavior A7X will be running a bit faster, maybe a lot faster than anything Intel has to offer and does so at a lower power point. That should be embarrassing for Intel but more importantly it gives reason for Apples completion to pause and contemplate how they will compete. Intel is no longer the go to supplier of CPUs for the hottest part of the electronics market, they need to address that.
  • Reply 3 of 9
    MarvinMarvin Posts: 15,322moderator
    When you compare the likes of the i7-3770k and i7-4770k, the performance difference is very little (5-15%) because they put the effort into the IGP:


    [VIDEO]


    When desktops are coming with dedicated GPUs, it means the CPUs aren't worth upgrading to. Intel would be just as well bumping the clock speed or something else minor to get the 10% boost. I could eventually see Intel stop production of desktop chips altogether.

    This wouldn't necessarily be neglecting desktops because even with Haswell, desktop chips go right down to 35W now. That's low enough to fit in a laptop. If you compare the top 84W i7-4770k to the mobile 47W i7-4800MQ, the difference really isn't that much:

    http://ark.intel.com/products/75123/ - $350, 84W
    http://ark.intel.com/products/75128/Intel-Core-i7-4800MQ-Processor-6M-Cache-up-to-3_70-GHz - $380, 47W

    In Cinebench, the i7-4770K (not overclocked) scores 7.62, the 4800MQ is 7.42:

    http://www.notebookcheck.net/Review-Intel-Haswell-Processors.93189.0.html

    Both have 4600 graphics. Why even bother making the i7-4770K? They'd be better using the same mobile designs for desktop and switch them to 6-cores for high-end desktops. Whatever route they go down, it's about time we started seeing 50% performance jumps in a major CPU upgrade. It's forgivable to an extent because they do show 50% on the IGP but it's not much good when you have a dedicated GPU anyway. If they could add another 2 cores to a mobile CPU, that would bump a mobile CPU up 50% and make it more worthwhile than a much higher TDP desktop chip that is around the same speed.
  • Reply 4 of 9
    hmmhmm Posts: 3,405member

    Haswell-E is a different market with minor overlap. Of course you'll see that, given their margins in that segment. It's not necessarily on a 12 month cycle, but I've pointed that out before. We have the very first ivy bridge E machines being announced at this point. I'm a little curious if this is a misinterpretation of leaks and datamining. I recall a rumor of intel going to BGA solutions on non - E/EP desktop chips as of Broadwell, although I cannot remember if it was ever confirmed. This might be tangential to that data. As for a 2 year desktop chip cycle, it wouldn't surprise me. In EP chips we see core count bumps each tick/tock. Notebook chips see IGP gains. Mainstream desktop chips stay fairly consistent. As Marvin points out there is some convergence. It's just not that big of a deal. If that extra 5-10% was crucial, you wouldn't buy the aforementioned chips at all. You would go straight to EP solutions. Even the "watercooling crowd" may not take issue with this, as other restrictions sometimes limit overclocking of one generation compared to the prior one.

  • Reply 5 of 9
    Quote:
    Originally Posted by hmm View Post

     

    Haswell-E is a different market with minor overlap. 


     

    I didn't think Haswell-E would be going in a Mac anyway, I just included it for completeness of the CPUs coming next year for desktops.

  • Reply 6 of 9
    hmmhmm Posts: 3,405member
    Quote:
    Originally Posted by Kendog52404 View Post

     

     

    I didn't think Haswell-E would be going in a Mac anyway, I just included it for completeness of the CPUs coming next year for desktops.


    EP would be a likely mac pro candidate. E would not. EP versions typically slide out second. My prediction for Haswell E/EP actually shipping from oems is possibly early 2015. It may or may not be next year. E typically represents the set branded as i7. Those are typically out slightly earlier than the Xeons. We won't see E versions in imacs. I said the same thing when people predicted a 6 core imac. There was no way Apple would generate another complete logic board design to accommodate that single processor option. I would have expected it if intel had released a 6 core chip in an LGA 1155 socket.

  • Reply 7 of 9
    winterwinter Posts: 1,238member
    Marvin wrote: »
    When you compare the likes of the i7-3770k and i7-4770k, the performance difference is very little (5-15%) because they put the effort into the IGP:

    When desktops are coming with dedicated GPUs, it means the CPUs aren't worth upgrading to. Intel would be just as well bumping the clock speed or something else minor to get the 10% boost. I could eventually see Intel stop production of desktop chips altogether.

    This wouldn't necessarily be neglecting desktops because even with Haswell, desktop chips go right down to 35W now. That's low enough to fit in a laptop. If you compare the top 84W i7-4770k to the mobile 47W i7-4800MQ, the difference really isn't that much:

    In Cinebench, the i7-4770K (not overclocked) scores 7.62, the 4800MQ is 7.42:

    Both have 4600 graphics. Why even bother making the i7-4770K? They'd be better using the same mobile designs for desktop and switch them to 6-cores for high-end desktops. Whatever route they go down, it's about time we started seeing 50% performance jumps in a major CPU upgrade. It's forgivable to an extent because they do show 50% on the IGP but it's not much good when you have a dedicated GPU anyway. If they could add another 2 cores to a mobile CPU, that would bump a mobile CPU up 50% and make it more worthwhile than a much higher TDP desktop chip that is around the same speed.

    I am quoting most of your post but did not want to quote the entire thing including the links however my first reaction was "Holy shit!" as I am just trying to imagine having a Mac mini with the i7-4800MQ and trying to exploit it for all it's worth.

    It clearly tells me that if I do want to invest in a machine I'll have for a lot of years, definitely go for a quad core mini over the base (most likely) dual core.
  • Reply 8 of 9
    wizard69wizard69 Posts: 13,377member
    winter wrote: »
    I am quoting most of your post but did not want to quote the entire thing including the links however my first reaction was "Holy shit!" as I am just trying to imagine having a Mac mini with the i7-4800MQ and trying to exploit it for all it's worth.
    It just gets better and better each year as process technology catches up with the Minis case. This is why I was posting for a very long time about a smaller Mac Pro even if what we got wasn't imagined. Also consider this, Mini uses laptop chips so Broadwell isn't out of the question in a future Mini. At least some of the Broadwell chips are rather complete SoC which could make for another round of performance increases in the Mini. These increases won't come just from the SoC buy also from the freed up board space in the Mini. If the Mini doesn't get refactored for a Haswell, Broadwell may give Apple the incentive it needs.
    It clearly tells me that if I do want to invest in a machine I'll have for a lot of years, definitely go for a quad core mini over the base (most likely) dual core.

    I remember back to when dual processors started to become common, this was before dual cores. Everybody would say why do you need that and then sneer when you said the machine works better. When dual core chips became mainstream people began to realize that the early dual processor guys where onto something as the machines work all around better. Mind you this was before operating systems really had the technologies in place to leverage more than one core well.

    In any event, on the desktop, we are at the same point with quad cores, they are a far better investment if you are thinking long term. So much so I advise most people to buy quad core if they have a usage profile that at all suggests a benefit. Even then it is hard to guess how an individual will benefit as you really don't know how, when or where any given app will benefit from Grand Central Dispatch or other technology that leverages the cores in your machine.
  • Reply 9 of 9
    winterwinter Posts: 1,238member
    I honestly don't see myself waiting until a Broadwell mini comes out for me to upgrade my hardware but I probably will not buy a Haswell mini (quad or dual) on launch.
Sign In or Register to comment.