who says no new mac pro this year?

Discussion in 'Mac Pro' started by jb502, May 4, 2011.

  1. derbothaus macrumors 601

    derbothaus

    Joined:
    Jul 17, 2010
    #2
  2. CaptainChunk, May 4, 2011
    Last edited: May 4, 2011

    CaptainChunk macrumors 68020

    CaptainChunk

    Joined:
    Apr 16, 2008
    Location:
    Phoenix, AZ
    #3
    This still tells us practically nothing - it's only speculation.

    Fact: SB-based Xeons don't ship until at least Q4 2011 (and possibly later, like Q1 2012), according to Intel.

    Based on this fact, it's pretty safe to assume that after adding in a couple of months for manufacturing, you're realistically looking at new Mac Pros shipping somewhere in Q1 2012 at the very earliest. SB-based desktop and mobile chips have been available for months before they finally shipped on new MBPs and iMacs.

    It wouldn't make a whole lot of sense to do an interim refresh on a socket that's for all intents and purposes dead. SB Xeons will run on the new LGA-2011 socket. Besides, what would Apple really want to add on the departing LGA-1366 socket? Slightly faster ones (which all have higher TDPs)? I highly doubt that.

    There has only been ONE major interim release in the history of the Mac Pro and that was simply to add an 8-core processing option in 2007. Look at the timeline:

    2006/2007: Woodcrest (Core-based) Xeons; interim 8-core made available in 2007 with Clovertown CPUs - LGA-771

    Early 2008: Harpertown (Penryn-based) Xeons - LGA-771

    Early 2009: Nehalem Xeons - LGA-1366

    Early 2010: Westmere (Nehalem die shrink) Xeons; 6- and 12-core options added - LGA-1366


    This isn't about being a "know-it-all." Rather, it's about being realistic about product launches based on factual information. In the consumer segment, products refresh a lot more often; that's not always the case with the workstation segment.
     
  3. nanofrog macrumors G4

    Joined:
    May 6, 2008
    #4
    When I focused on the part it mentions about the last MP, it really only states that it was last updated July of 2010.

    Now considering that recent history shows they wait for new processors, this can only happen as soon as they get their hands on the SB parts that will be used. So unless Apple can manage to get CPU's early, the fastest they could possibly get a system out, is Jan 2012 (actually ship a system; announce it in December 2011, and this is assuming the parts hit very early in Q4).
     
  4. Umbongo macrumors 601

    Umbongo

    Joined:
    Sep 14, 2006
    Location:
    England
  5. gpzjock macrumors 6502a

    Joined:
    May 4, 2009
    #6
    I have to agree with the no new model till end of the year concept and an upgrade in the meantime will be a partial one at best. Sandy Bridge is the quantum leap for memory and CPU performance here and Apple would want that in the box.

    Wu has a target price of $445 for Apple shares? Hmm, I thought $350 was high especially as Steve Jobs could "retire" permanently at any moment.
     
  6. Vylen macrumors 65816

    Joined:
    Jun 3, 2010
    Location:
    Sydney, Australia
    #7
  7. ugru macrumors 6502

    ugru

    Joined:
    Sep 8, 2002
    Location:
    Caput Mundi
    #8
    Impossible unless Apple is willing to abandon the use of Xeon Processors in the "normal" Mac Pro...

    They may want to differenziate even more their offer...

    Base MacPro--> Multi Processor Sandy Bridge i7 (if possible) - the infamous Xmac
    High End MacPro and MacPro Server--> Multi Processor Sandy Bidge Xeon
     
  8. johnnymg macrumors 65816

    johnnymg

    Joined:
    Nov 16, 2008
    #9
    In the investor community his nikname is 'Shaw Wrong'.

    Not only is he a terrible analyst but his "predictions" on Apple products is equally terrible.

    JohnG
     
  9. deconstruct60, May 5, 2011
    Last edited: May 5, 2011

    deconstruct60 macrumors 604

    Joined:
    Mar 10, 2009
    #10
    There is practically zero evidence that you need to add more months for manufacturing. A previous MacPro edition shipped with new Xeons from Intel before released them to other vendors. The Mac Pro ships a relatively (to the other Macs) low volume so there is zero need to compose a stockpile over months to meet with demand on day one. Apple may be off a couple of weeks from the release date perhaps because can't get enough out of the gate ( since buy so few relative to the Dell' and HP ) or the release of a more high volume mac that conflicted with Intel's Xeon E5 date. The latter is become less likely over time as the other 2011 Macs make it out the door.


    Furthermore, Xeon chips are released to the manufacturers months before the general release so they can do their own extensive server/workstation testing.

    A common reason for Apple to delay is because the new chips arrive in a window that is too close to when Apple did its last model refresh. Since the 2010 refresh came in August, any Q4 release by Intel is late relative to that timeframe. I'm sure the drumbeat of the "Mac Pro is dead" will be quite loud 2 months after the August date passes. The delay last year was exactly to mitigate the delay this year. Slide 4-5 months last year. Slide 2-3 months this year. As opposed to sliding 7-8 months this year.



    The MP may slide but that would more likely be to there being bugs either in the Xeon or the supporting chipset (e.g., PCI-e v3.0 bugs or the previously exposed SATA bug). That seems less likely now since Intel's E5 date has slid from Q3 to Q4.

    I can also see a slide from Q4 to Q1 if the Q4 release date is inside of December. On December 15th they probably aren't going to start something new. October or November are not a problem.

    If it is Q1 it is because the E5's needed for the full line up ( single and dual package models ) didn't get released until Q1.




    Again you are ignoring two factors. Availability of other major components ( GPUs ) and dates of last release (they are not trying to release multiple changes of each model per year.)

    For example, the MBP got hung up one year because the 320M weren't out and shipping in volume yet. The Xeon's come in yearly intervals so normally it is not some other component that the MPs have to wait on.
     
  10. sjordan macrumors member

    Joined:
    Apr 20, 2011
    #11
    http://www.anandtech.com/show/4314/apples-imac-the-first-z68-for-sale

    According to the article the new Imacs use the Z68 which is ahead of Intels Official Launch date.

    So to say that Apple will release the new MP with SB is not to far fetched.

    My guess is they will release the MP a month or 2 before the official launch date of new Xeon SB. I won't put money on it though :)
     
  11. philipma1957 macrumors 603

    philipma1957

    Joined:
    Apr 13, 2010
    Location:
    Howell, New Jersey
    #12
    So at best October. I would think they want it out no later then dec. my reasoning is it is a pro. That means for many buyers it is a legit business exp/deduction. So a dec purchase allows it to be deducted this year.
     
  12. deconstruct60, May 5, 2011
    Last edited: May 5, 2011

    deconstruct60 macrumors 604

    Joined:
    Mar 10, 2009
    #13
    No. It is possible if Apple is abandoning the dual CPU package model Mac Pro. If 4 highly clocked cores is all their users really need then it is possible. It is just not very probable.

    The new Xeon E3s are currently shipping. The E3-1280 is 3.5GHz and turbos up to 3.9GHz. That would make many high clock, low core folks wet their pants and Apple could probably do it for $3,000-3,200 by shaving some costs in the box. If you couple the new "slim fast" Mac Pro case rumor with an E3 powered machine you could release in next couple of months. However, they are aimed at lower end workstations ( exactly the space that overlaps in price with the iMac line up).

    With the XServe dropped perhaps there is a small chance that Apple will split the Mac Pro. A Mac Pro mini/slim and a Mac Pro classic. It is possible but I wouldn't not put a more that 50% probability on it. They could use the R&D budget of the XServe to "pay for" the increased effort for the split Mac Pro models.

    However, the Mac Pro already has volume problems. Increasing the R&D for multiple cases, boards, releases will only make that worse. The Mac mini and mini server is largely the same except for a slot cut in the case. The Mac Pro is highly likely to be in the same boat. The MBP ships in substantially higher numbers so there are more units to distribute the multiple case/board R&D to. Furthermore, it would represent a 180 degree shift by Apple into believing in the mini-tower is more important that product pricing separation. I don't think they do. The billions in the bank and the increasing stock price are strong arguments to overcome.

    The XServe died in part because couldn't generate the volume to justify existence. If Apple split the dual package Mac Pro off from the single package one that would be putting it into a very similar situation. It would be the high priced option that increasingly shrinks in Mac market share over time. Within a couple of years the Mac Pro "classic" would die off just like the XServe did because placed it in the same position.

    It would be nice if Apple broke off the Mac Pro into a $2,000-3,200 to the "slim" model and a $3,500+ differentiated rackable workstation model, but I don't think they will. The overall workstation market is moving down over time in both price and relative volume. There are already a sizable number of $1,500-2,000 models and that is only going to get larger over time. ( computers get more affordable over time. a long term term for last 40 years. ).
     
  13. deconstruct60, May 5, 2011
    Last edited: May 5, 2011

    deconstruct60 macrumors 604

    Joined:
    Mar 10, 2009
    #14
    Again... a Z68 couple to Xeon's E3 doesn't have to wait for the E5 launch date. The E3's are already shipping. However, this is really about "overclocker, minitowers" which Apple has spurned repetitively for last 7-8 years.

    It is also a reach. The Z68 supports hybird HDD/SDD usage. Maybe Mac OS X 10.7 isn't so lame from a "under the hood" perspective and could enable this on the new iMacs once installed. Or 10.8 ( which these iMacs should get updates for also. ) I wouldn't draw conclusion about a MP from the iMac's support chip.
     
  14. anim8or macrumors 65816

    anim8or

    Joined:
    Aug 16, 2006
    Location:
    Scotland, UK
    #15
    How much trouble would it be for them to simply add TB to the lineup on the MP and possibly a spec bump somewhere.. or price drop... just in the interim until the SB chips are readily available...

    This would certainly keep me happy and would allow me to upgrade my MP 1,1 to a 12 core machine.... :)
     
  15. xgman macrumors 601

    xgman

    Joined:
    Aug 6, 2007
    #16

    anything is possible, but I am doubting that they will do anything before the new cpu's are ready, and quite frankly I hope they wait.
     
  16. derbothaus macrumors 601

    derbothaus

    Joined:
    Jul 17, 2010
    #17
    Apple is going to most likely use X78 not Z68.
     
  17. deconstruct60, May 5, 2011
    Last edited: May 5, 2011

    deconstruct60 macrumors 604

    Joined:
    Mar 10, 2009
    #18
    E3's come with GPUs. So if putting Thunderbolt on a box with "Mac Pro" in the label somewhere were an objective then it is much easier. Mac Pros where the video solely comes off a PCI-e card. ... not likely to happen. Neither Nvidia nor AMD are inclined to jump into bed with Thunderbolt and guarantee that Intel makes money off of every card they sell. Intel is king of the graphics market... but that solely riding on integrated graphics. It would be kind of silly to fall further behind by handing Intel even more money to beat them over the head with.


    Thunderbolt and maximize GPU performance are at conflicting goals right now. If you have PCI-e slots there is not as great a need for a PCI-e extender.

    It may help justify forking into a "MP slim" (with TB) and "MP classic" (where use PCI-e sockets) path but unless fork TB is somewhat at odds with existing MP added value.

    P.S. Before someone snaps on a MP with integrated graphics. The integrated graphics could be for a secondary or tertiary monitor that you optionally hang off the TB socket(s). The primary monitor(s) could be driven by the 1-2 slots in the "slim" model. The integrated graphics is just filling the checkbox that the TB controller "should be" hooked to graphics on a computer. Defacto folks use TB as the PCI-e lane multiplier/switch that is present in many boxes today (include MP.) and push the PCI-e connctivity out of the box to make it smaller.
     
  18. Inconsequential macrumors 68000

    Joined:
    Sep 12, 2007
    #19
    Not going to happen before X78 and when the LGA2011 chips are released.

    There are plenty of threads on this already.

    IF, and ONLY IF, Intel allow Apple to have first dibs on the chips (AKA MP2009) then we MIGHT see one as early as Q3.

    Personally my money (and my *actual* money - im buying a SB MP) is on Q1 2012.

    Anyway, given the fact not even hardware enthusiasts and overclockers who get chips very early have SB-E means it's not going to be for quite a while yet!!!
     
  19. nanofrog macrumors G4

    Joined:
    May 6, 2008
    #20
    The LGA2011 parts will use the X79 chipset. There's articles out on it, which sparked the creation of a thread about it here in MR.

    If they expanded the MP line, I could see this (i.e. use the E3's as the entry level machine). But I don't expect the MP has the sales volume to justify this (add in yet another socket to the mix). And this isn't even taking any performance issues into account, such as the IGP and DMI 2.0.

    Granted, the IGP would be better used as a secondary GPU source, primarily to get the DisplayPort signal to the TB chip as you mention. But I see the DMI 2.0 interconnect between the CPU and chipset as a bigger issue when a user is pushing the system in terms of I/O.

    There are ways of doing it though, whether GPU makers work with Intel to create an open standard (i.e. edge connector on the board to get the DisplayPort data to the TB chip), or Apple orders custom card with that or the TB chip on it. Not as easy as either an IGP or embedded GPU, but at least possible.

    This would be the way to approach it to me as well.

    But like you, I just don't see this happening, unless:
    1. Apple either splits the SP and DP systems even further (but I don't think they've the sales volume to do this).
    2. They do away with DP systems (unless they do 2x SP systems to get past 4 cores, which is still a 2 system lineup - it should be cheaper though).
     
  20. JeepGuy, May 5, 2011
    Last edited: May 5, 2011

    JeepGuy macrumors regular

    Joined:
    Sep 24, 2008
    Location:
    Barrie
    #21
    The X79 is a DMI 2.0, or will the next mac pro have some other as of yet unannounced chipset with QPI. Also the E3 series is available with and without IGP, Dell is using the non-IGP versions in it's lower end T1600 workstations.

    The recent rumor of apple doing a case redesign maybe an equivalent to the T1600 (low cost), they need more choice with all the people migrating from PC's to Macs, just some food for thought.


    Edit:

    from the quote it looks like DMI 2.0 on the X79 is different from the Z68, even thou they share the same name.
     
  21. Umbongo, May 5, 2011
    Last edited: May 5, 2011

    Umbongo macrumors 601

    Umbongo

    Joined:
    Sep 14, 2006
    Location:
    England
    #22
    Okay I had a look around after originally posting my response to this.

    From what I gather QPI will be used just as it is on current X58 and 5520 chipsets. Intel / leaked sources have provided very little information on memory so far so that is probably why it isn't mentioned.
     
  22. nanofrog macrumors G4

    Joined:
    May 6, 2008
    #23
    The X79 uses 2x QPI links, not DMI 2.0 (DMI 2.0 is limited to the LGA1155 socket).

    Maybe, but I'll take the wait and see approach, as the LGA1155 socket running 4 cores makes for a limited system in terms of performance.

    DMI 2.0 /= QPI. There are similarities, such as the use of 4x physical links per internconnect, but they differ past that.

    Read here for a bit more information (read the last line/paragraph in th e Implementation section in particular). The DMI wiki also offers some additional information. Hopefully this can help clear things up a bit.

    They're continuing with QPI in it's current implementation (they have finessed a bit more band out of it from what I'm seeing).
     
  23. derbothaus macrumors 601

    derbothaus

    Joined:
    Jul 17, 2010
    #24
    Sorry, that's what I meant. X79 not X78. The X58 replacement. Their consistency is as good as the graphics card companies.
     
  24. nanofrog macrumors G4

    Joined:
    May 6, 2008
    #25
    Intel's naming schemes is definitely getting more complicated, due in large part to 3 sockets for Sandy Bridge CPU's.
     

Share This Page