Shop OBEX P1 Docs P2 Docs Learn Events
Cog layout on P8X32A die — Parallax Forums

Cog layout on P8X32A die

TubularTubular Posts: 4,717
edited 2011-06-16 23:10 in Propeller 1
Does anyone know the physical order of cogs on the P8X32 chip die?

Do they go in sequence 0..7, or odds then evens, etc?

The closest image I have found so far is this one, but it doesn't detail the cog locations.

PropDieDiagram.jpg

Comments

  • User NameUser Name Posts: 1,451
    edited 2011-06-15 16:59
    It seems like Kuroneko knows. He has traced certain subtle behavioral differences down to the location of individual cogs.
  • kuronekokuroneko Posts: 3,623
    edited 2011-06-15 18:41
    All I have is an informed guestimate. cog N >> pin P >> cog M round trips suggest that timing is somehow symmetrical (for want of a better word). cogs 0..3 are better used with pin groups 0 and 1, cogs 4..7 favour groups 2 and 3. This ties in with the OR-chain schematic in the datasheet (9.5 Cog and I/O Pin Relationship). Looking at a QFP prop, pin 0 is located in the middle of a pin group (cw from chip marker). If you align that location - big assumption here - with the top/middle pin of the die then you'll end up with cog 0 left and cog 7 right (in a linear fashion).

    Note, the delay effect is only visible above 52MHz (I didn't have any other crystal to get closer to 80MHz, most likely 64MHz will still be OK). At standard operating frequency (80MHz) [thread=127653]pin synchronisation[/thread] is a no-go.
  • TubularTubular Posts: 4,717
    edited 2011-06-15 20:25
    Thanks Kuroneko & UserName.

    I'm seeing differences between the cogs, which I think is explained mostly by the OR chain difference, but partially due to other factors too.

    With regard to pins, I gather the double bond pad connections are Vss or Vdd. The left and right sides of the die image each have two single bond pads in between the Vss and Vdd pairs (BOEn/Reset, and XO/XI). I was assuming the left pin pair, which are more closely spaced and somewhat "symmetric" in their metal connection (on the Flylogic image, wrt the right hand pair), might be the XO and XI, but this would go against your suggestion above. I think I'll go with your suggestion and see how far it gets me, but keep in the back of my mind it may be 180 degrees out.

    Thanks for the pin syncronisation link, very interesting reading.
  • TubularTubular Posts: 4,717
    edited 2011-06-16 02:50
    Ok. I think I more or less verified the cogs are laid down in a linear fashion, 0 thru 7. This diagram is an averaged response of cogs 1 through 6 (labels A through F) to exactly the same pin (P12) being toggled, though it gets a bit murky in the middle with C D and E (cogs 3-5) being harder to distinguish.

    6cogs.jpg
    1024 x 531 - 35K
  • Dave HeinDave Hein Posts: 6,347
    edited 2011-06-16 08:09
    So how much variation are we talking about? A few nano-seconds?
  • TubularTubular Posts: 4,717
    edited 2011-06-16 18:01
    Dave, it was really a qualitative test averaged statistically over a large number of cycles. I would hate to have to work back to work out what it means in quantitative terms.
  • Cluso99Cluso99 Posts: 18,069
    edited 2011-06-16 23:10
    Now who said you could examine the heat from the die to detect instructions being executed???

    Just look for the silkscreen on the die Haha! Right beside his signature!
Sign In or Register to comment.