I'm currently trying to understand the pipeline but I'm puzzled by one small detail. The dec_iX_decode_d signal is used as the highest value of the iX_pipe_en signal which is then shifted down the pipe. From this value, we derive the enable signals (iX_eX_data_en and iX_eX_ctl_en) for the individual pipeline FFs. For the ...data_en signals, this makes sense to me, but the ...ctl_en signals stay active for one additional cycle. This leads to some "random" value being latched in the control FFs one cycle after dec_iX_decode_d changes to 0 (due to stalling etc.). So far, I don't think this "random" value matters, since the piped destination-packet contains a valid-decode flag that is not set for this cycle. But what was the design decision for this behavior?
Is this related to the comment that reads "// allow illegals to flow down the pipe"?
Hi,
I'm currently trying to understand the pipeline but I'm puzzled by one small detail. The
dec_iX_decode_dsignal is used as the highest value of theiX_pipe_ensignal which is then shifted down the pipe. From this value, we derive the enable signals (iX_eX_data_enandiX_eX_ctl_en) for the individual pipeline FFs. For the...data_ensignals, this makes sense to me, but the...ctl_ensignals stay active for one additional cycle. This leads to some "random" value being latched in the control FFs one cycle afterdec_iX_decode_dchanges to 0 (due to stalling etc.). So far, I don't think this "random" value matters, since the piped destination-packet contains a valid-decode flag that is not set for this cycle. But what was the design decision for this behavior? Is this related to the comment that reads "// allow illegals to flow down the pipe"?I would really appreciate some explanation.
Kind regards