Shop OBEX P1 Docs P2 Docs Learn Events
What limits the maximum overclocking speed? — Parallax Forums

What limits the maximum overclocking speed?

MicrocontrolledMicrocontrolled Posts: 2,461
edited 2011-07-12 16:29 in Propeller 1
Assuming that the board was built for high frequency, and you were able to consistently cool the Propeller to a safe operating temperature, what limits the Propeller's overclocking capabilities?
Just something I was thinking about.

Thanks,
Microcontrolled

Comments

  • localrogerlocalroger Posts: 3,452
    edited 2011-07-12 06:00
    The primary limit is heat dissipation, although some of the Propeller's features (PLL's, I'm looking at you) will have capacitance issues.

    It is interesting to speculate about what you could get away with if you, say, immersed the Propeller in FlourInert and chilled it to 40 below...
  • Dave HeinDave Hein Posts: 6,347
    edited 2011-07-12 06:12
    As the clock frequency is increased the propagation delay plus the setup times of the internal logic will exceed the clock period.
  • AleAle Posts: 2,363
    edited 2011-07-12 06:33
    There are several threads on this topic, and the consensus seems to be that you get < 128 MHz, more in the 120 range, being heat not an issue due to the low power of the propeller.
  • mindrobotsmindrobots Posts: 6,506
    edited 2011-07-12 07:01
    Back in the day, we had a field upgrade for one of the UNIVAC mainframes that consisted of an 36" wire with female connectors on each end and instructions to remove a 12" wire wrap wire and replace it with the 36" wire. A cache invalidate signal was getting to the cache too early and the 24" (~2ns) delay was needed.

    Speed kills!! :smile:
  • ColeyColey Posts: 1,110
    edited 2011-07-12 10:22
    One of the things limiting Propeller overclocking is the core voltage.
    I have proven that the Propeller hits a ceiling of ~120Mhz on 3V3 operation.
    However if you increase the core voltage you can push it further, as seen in this video.

    I'll be doing some more tests soon (that's before you ask me again Baggers!)

    Coley
  • MicrocontrolledMicrocontrolled Posts: 2,461
    edited 2011-07-12 11:29
    @Coley: Wow!!! What keeps the Propeller from frying itself at 4.5V, good cooling?
  • Oldbitcollector (Jeff)Oldbitcollector (Jeff) Posts: 8,091
    edited 2011-07-12 11:40
    I kept looking for the date on that video suspecting it to be April 1st. Apparently, it wasn't the joke I thought it was. How did you achieve that Coley?

    OBC
  • SapiehaSapieha Posts: 2,964
    edited 2011-07-12 11:48
    Hi Oldbitcollector.

    It is NOT JOKE on one of my PCB's I run Linus Demo with 15MHz PLL8 and it clocks correctly as faster in Seconds as % of increasing Frequency.
    And I run it with standard Voltages (Maybe some 0.x V more that 3.3V)

    Oldbitcollector (Jeff) wrote: »
    I kept looking for the date on that video suspecting it to be April 1st. Apparently, it wasn't the joke I thought it was. How did you achieve that Coley?

    OBC


    Ps. I have even Movie to it -- But it is to big to post on forum
  • ColeyColey Posts: 1,110
    edited 2011-07-12 12:12
    Microcontrolled wrote: »
    @Coley: Wow!!! What keeps the Propeller from frying itself at 4.5V, good cooling?

    No cooling was required, all I did was modify a Hybrid PCB and used a DIP Prop.
    I know Sapieha has had success with 120MHz with surface mount Props, I have found the tolerance of these to be much tighter.
    Long story short, you need a DIP Package and a modified power supply to do this.

    Oh BTW, I ran this for several weeks non stop and there was no apparent sign of degradation even though I'm sure there will have been some.
    Temperature wise it didn't even get warm which was quite a surprise.

    One thing I did note is that different instructions fail at different times, for example I can push further than 141MHz and still get a image but the code chokes at the double buffer copy stage so all you get is a static image.

    Like I said, once I get some free time (when PropGFX demos are done) I will be doing a more definitive study.

    Coley
  • Toby SeckshundToby Seckshund Posts: 2,027
    edited 2011-07-12 16:29
    I found that at 9 Volts the Prop had a problem.

    Strangly( thankfully) the SD card was the only thing to survive, on a copy of a "Blade 2" (once it cooled down !)
Sign In or Register to comment.