[SOLVED] conflict between slack equation and Timequest calculation

[kaz1]

There seems plenty contradictions between slack equations and Timequest calculations for setup slack as well as hold slack.

for example the equation below is documented for setup:
Data RequiredTime = Latch Edge + Clock Network Delay to Destination Register – µtSU – Setup Uncertainty

but Timequest shows the following:

Thus:
conflict A: clock uncertainty is -0.030 yet it is added as negative. I expect -(-0.030) to be adding +0.030
conflict B: uTsu is 0.242 yet it is added. I expect 0.242 to be subtracted

I wonder is this editing error or some mystery internal modelling by timequest or to do with sense of +

 

[kaz1]

Further analysis showed that TimeQuest has got careless sign issues.

conflict A: clock uncertainty of -0.030 is shown on data and waveform. I see this as meaningless since it shouldn't be negative. This value is then added (in effect is subtracted from accumulated total and so agrees with slack equation indirectly).

conflict B: uTsu is reported as +0.242 then added. However the waveforms shows it as -0.242 therefore it is subtracted from accumulated total -(-0.242) = + 0.242

Though I am surprised that uTsu is negative.

I have seen people reporting same issue in 2013. So after one decade it is still there.

 

[kaz1]

To be fair with Timequest, I noticed Vivado does same. It looks like the issue originates from Synopsys approach.

What I conclude is that:
The terms of equations are "normalised to be added always". So appear positive or negative as required for addition.