Skip to content

Why SNES -hardware works faster than expected – and why it is a problem

    A sample result of the DSP test program.

    Credit: Allan Cecil

    A sample result of the DSP test program.


    Credit: Allan Cecil

    However, these heat effects are fading in different consoles compared to the natural clock variation. The slowest and fastest DSPs in the Cecil monster showed a clock difference of 234 Hz, or approximately 0.7 percent of the 32,000 Hz specification.

    That difference is small enough that human players would probably not notice it immediately; TOTBOT team member Totaal estimated that it could be “for a maximum of one second or two [of difference] More than hours of gameplay. “Competable speed runners can notice small differences, but if different CPU and APU lines cause” remembering enemy pattern with the carefully changes into something else “between runs, Cecil said.

    For a frame-perfect tool-assisted speedrun, however, the clock variations between consoles can cause countless headaches. As Tasbot team member Undisbelder explained in his detailed analysis: “On one console this can take 0.126 frames to process the music ticker, on another console it can take 0.127 frames. It may not seem much, but it is enough to delay the start of the loading, laying and game-code).”

    Cecil's research found that variation between consoles was much higher than the effects of heat on a single console.

    Cecil's research found that variation between consoles was much higher than the effects of heat on a single console.


    Credit: SNES SMP Speed ​​Test Survey

    Cecil also said that the DSP clock speeds reported by research were also slightly higher than it expected, with an average speed of 32,076 Hz at room temperature. That is a bit higher than both the 32,000 Hz -Spec set by Nintendo and the 32.040 Hz speed that emulator developers have established after sampling the actual hardware in 2003.

    For some observers, this is proof that SNES APUS was originally produced in the 90s, accelerated somewhat as they get older and could stay faster in the coming years and decades. But Cecil says that the historical data they have is too indirect to make such a claim for sure.

    “We are all a bunch of different skilled geeks and nerds, and it is in our nature to argue about what the results mean, what is fine,” said Cecil. “The only thing that we can say with certainty is the statistical significance of the answers that show that the current average DSP stake speed is 32,076 Hz, on average faster than the original specification.