Diagonally Mounted Processors

Posted:
in General Discussion edited January 2014
The eMac, and the new PowerMac have processors that are diagonally mounted. Please don't link to the pictures of the prototype G4 though. If someone can dig up the pic of the eMac mobo and post it then please do so.



I was just wondering...what are the pros/cons of doing this what does it mean for next gen iMacs/Portables?



Perhaps utilizing this method daughtercards/mobos can be smaller? Maybe this tech will aid in the development of smaller notebooks, tablets, PDA/Phone hybrids, etc.

Comments

  • Reply 1 of 16
    cdhostagecdhostage Posts: 1,038member
    Diagonal to what? Which direction?



    Yes, indeed, let's get an image here.



    I can't imagine how they can fit stuff in a square container with stuff going slantwise
  • Reply 2 of 16
    bartobarto Posts: 2,246member
    [quote]Originally posted by Spart:

    <strong>

    I was just wondering...what are the pros/cons of doing this what does it mean for next gen iMacs/Portables?

    </strong><hr></blockquote>



    Any PCB designers here?



    The 7200 had a diagonally mounted 601 btw.



    Barto
  • Reply 3 of 16
    woozlewoozle Posts: 64member
    Mounting chips diagonaly to some reference is supposed to make it easier to route traces.



    A bit of simple examination reveals the following:



    Consider a chip with an array of 5x5 pins under it ( not many i know ).

    All the pins are to be routed to a slot.

    Aligned to the slot, the mean number of pins that a single trace must pass by is 46/25.

    It is hard to get lots of traces past the pins, and so either the board must have more layers, or traces must route around the chip.



    When aligned at 45 degrees, the mean number of pins a trace must pass is 30/25.

    Therefore there is less need for additional steps to be taken to route the traces.
  • Reply 4 of 16
    powerdocpowerdoc Posts: 8,123member
    speakinf of the mobo of the new G4, i did not recocnize the presence of a daughter card. did somebody clearly see it , or does Apple remove the daughter card in the new design ?



    [ 07-26-2002: Message edited by: Powerdoc ]</p>
  • Reply 5 of 16
    spartspart Posts: 2,060member
    ...There was a pic of the daughtercard itself and you could see part of it behind the GeForce 4 MX in the mobo photo.
  • Reply 6 of 16
    woozlewoozle Posts: 64member
    The daughter card was not _clearly_ visible, and at first I thought the diagonal chip was the controller.

    However, if you look at the components around the cpu, you will see that they match those around the chip on the board, then you will be able to see the board itself.
  • Reply 7 of 16
    krassykrassy Posts: 595member
    [quote]Originally posted by woozle:

    <strong>The daughter card was not _clearly_ visible, and at first I thought the diagonal chip was the controller.

    However, if you look at the components around the cpu, you will see that they match those around the chip on the board, then you will be able to see the board itself.</strong><hr></blockquote>



    i agree here. the daughtercard is there - seems to be a little bit bigger than the older single-proc-G4-card... what makes me wonder is that there's no pic of the heatsink?



    [ 07-26-2002: Message edited by: Krassy ]</p>
  • Reply 8 of 16
    powerdocpowerdoc Posts: 8,123member
    Thanks for the info, i was scared at the idea that in the new mobo the chip was directly implemented (it's means : no dual ...)



    [ 07-26-2002: Message edited by: Powerdoc ]</p>
  • Reply 9 of 16
    slackerslacker Posts: 127member
    I don't design PCB's but I've been in the industry of building them (for all types of applications) for about 7 years. I don't think mounting a chip diagonally makes it easier to route traces than if it were mounted normally. It's all going to depend on the layout of the rest of the board.



    I don't work in our surface mount area, so I don't know what issues arise from placing at an angle either. Maybe I'll have to call them up and see if there are any issues to placing parts at an angle offset from the boards sides.
  • Reply 10 of 16
    rogue27rogue27 Posts: 607member
    Diagonal mounting is more aerodynamic, so your computer can go faster!
  • Reply 11 of 16
    wheewhee Posts: 46member
    Usually there's a trace length reduction when chips are mounted diagonally, which can lead to greater performance, less power usage, etc. There's actually a <a href="http://www.xinitiative.org/"; target="_blank">web page</a> promoting the use of this layout technique.
  • Reply 12 of 16
    slackerslacker Posts: 127member
    [quote]Originally posted by whee:

    <strong>Usually there's a trace length reduction when chips are mounted diagonally, which can lead to greater performance, less power usage, etc. There's actually a <a href="http://www.xinitiative.org/"; target="_blank">web page</a> promoting the use of this layout technique.</strong><hr></blockquote>



    If I'm not mistaken you are talking about physical chip design (inside the IC), whereas this thread was discussing mounting a chip on a Printed Circuit Board at a 45 degree angle.



    Someone correct me if I'm wrong.
  • Reply 13 of 16
    eskimoeskimo Posts: 474member
    [quote]Originally posted by Slacker:

    <strong>



    If I'm not mistaken you are talking about physical chip design (inside the IC), whereas this thread was discussing mounting a chip on a Printed Circuit Board at a 45 degree angle.



    Someone correct me if I'm wrong.</strong><hr></blockquote>



    You are correct about the x initiative. It's aim is to allow for 45 degree routing of interconnects inside of chips.



    However, woozle is correct about the the diagonal mounting having the benefit of shorter traces. If you notice Nvidia and SiS have both used diagonal mounting on their recent chipsets. It is especially helpful with timing requirements for high speed memory.
  • Reply 14 of 16
    woozlewoozle Posts: 64member
    IANAEE.



    My knowledgable postition is based totaly on here say, but giving the concept a couple of minutes thought gave me the results about mean pins passed by a single trace.



    Certainly real boards are more complex, but the basic idea still holds.



    By making it easier to route traces, it is easier to keep them within their timing ( length ) and impedance requirements.



    By reducing the complexity of the routing it may be possible to prevent a board from using additional layers to meet those requirements, reducing costs.
  • Reply 15 of 16
    curiousuburbcuriousuburb Posts: 3,325member
    [quote]Originally posted by rogue27:

    <strong>Diagonal mounting is more aerodynamic, so your computer can go faster!</strong><hr></blockquote>



    if you assume they're tilted forwards... but with a 45 degree tilt, the glass-half-empty folks will insist it's tilted backwards



    now if the daughtercard had a leading edge curve...

    and the venturi holes were actually engine intakes...

    and the secret partner wasn't nVidia but Rocketdyne...



    this might be the promised iFlying car prototype :eek:
  • Reply 16 of 16
    Since this is more a general tech thread, moving to General Discussion.
Sign In or Register to comment.