0
\$\begingroup\$

Video timing standards for example DMT, CVT provide information about the timing requirements of various video frame components like vertical/horizontal blank/active/total etc.

DisplayPort operates in terms of link clock, which is higher than pixel clock and to compensate for this gap, fill symbols are added according to the calculated value of active symbols per transfer unit.

Since number of active symbols in a transfer unit comes out to be a fractional value, it is likely that we either end up sending slightly more or less data thus consuming slightly more or less time.

So, what is that timing error margin upto which image generation won't be erratic?

\$\endgroup\$
1
  • 1
    \$\begingroup\$ Does the spec say anything about this? Finding an actual error threshold is hard and will differ based on hardware (including the cable), but there should be a specified maximum jitter below which errors are guaranteed not to happen. \$\endgroup\$
    – Hearth
    Commented Jan 20, 2023 at 18:32

1 Answer 1

1
\$\begingroup\$

The standard video signal formats are defined with pixel accuracy. Assuming standard video format is sent in to DP link, identical data stream must come out, as DP is just a fixed rate transport link for a video stream.

And, since there is a fixed ratio between video stream clock and link clock, there will always be an integer amount of pixels per transfer unit. Depending on the ratio, there will in theory always be X or X+1 pixels accumulated to transfer in a single transfer unit packet which can be 32 or 64 symbols per lane, and so in theory the average value will be exact.

The variability is given in standard as -8 symbols from target symbol count to actual symbol count.

This allows to send out a start of line with up to 8 symbols earlier than waiting for the actual average symbol count.

\$\endgroup\$
1
  • \$\begingroup\$ The ratio between video stream clock and pixel clock is "fixed" but it usually isn't a 4:16 type of a ratio, it many times is a 7:13 kind of a ratio. Transfer units are defined in terms of bytes/symbols and not in terms of pixels. Based on resolution, a pixel could be 2.25 or 3 or 6 bytes/symbols. \$\endgroup\$
    – lousycoder
    Commented Jan 23, 2023 at 5:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.