Challenge 11 : Setup/Hold slack calculation
Challenge on Setup/hold slack calculation...
check below image.
- without crosstalk impact, setup/hold slack between FF1-->FF2 is 0ps.
- with crosstalk, crosstalk of 10ps is introduced on below 3 nets.
what will be setup/hold slack considering this crosstalk effect?
#STA #Setup #Hold #Challenge
Type your answer in Comment below.
No effect it will be as it is
ReplyDeleteCould you please explain why it will be no effect? (I.e slack 0ps )
DeleteI wrote equation and it was cancelling from both side ,is it correct by the way ?
DeletePls show the calculation here so will check
DeleteSorry I meant I wrote equation verbally
DeleteYou can type verbal one also here...
DeleteI would appreciate that atleast you have tried to answer first!!
In slack there is RT and AT and when I calculate it's common in both so it cancels
DeleteShouldnt we know whether its positive or negative crosstalk?
DeleteYou can consider the approach Timing engine uses.
DeleteConsider pessimistic scenario.
Okk so -20 ps for both
DeleteIf we introduce crosstalk on 10 ps of each 3 nets we get impact on setup vio.where hold checks on launch edge, adding delay on data path it becomes hold faster . So skew skew reduce .
ReplyDeleteIn setup worst case --> adding delay becomes more worsen in capture clk path.
Pls share final setup and hold slack value.
Delete-40ps setup slack and -20ps hold slack due to Cross talk delay
ReplyDelete-20,-20ps is it correct id yes then will share process?
ReplyDeleteOther members also presented their logic.
DeleteSo you can also share it even though if you feel your answer is not right.
-40ps setup slack and -20ps hold slack due to Cross talk delay.
ReplyDeleteSetup slack: capture path early by 20ps and launch path late by 10ps and data path delay increase by 10ps. Because of this Cross talk delay setup slack -40ps.
Hold slack : capture clock path late by 10ps and data delay early by 10ps. Because of this Cross talk delay -20ps.
I guess this is the correct answer
DeleteYes. This is correct answer!
DeleteWhat about crpr rhe first crosstalk how it is modelled early and late ?
DeleteFor setup analysis,
DeleteLauch and capture at different edges.
So you can use banifit of CRPR as timing window can be different for launch and capture edge.
So if used then setup slack will be -20ps right?
DeleteSorry.
DeleteThere is typo in my previous comment.
For setup analysis, you can use CRPR banifit as launch and capture edge are different.
For hold analysis, CRPR can be used as launch and capture edge are same.
Now calculate slack.
-20 ps for both Setup and Hold
ReplyDeleteHI Shubham,
DeletePlease explain your calculations for the same.
First crosstalk is on common path to launch and capture clock path, so will not be including its impact.
DeleteWorst-case Setup Analysis:
Data path should be max delayed (i.e., AT' = AT+10ps) and clock arrival should be at its earliest (i.e., RT' = RT-10ps).
Setup Slack = RT' - AT' = (RT-10ps) - (AT+10ps) = RT-AT-20ps = -20ps (as earlier RT-AT = 0)
Worst-case Hold Analysis:
Data path should be at earliest (i.e., AT' = AT-10ps) and clock arrival should be delayed (i.e., RT' = RT+10ps).
Hold Slack = AT' - RT' = (AT-10ps) - (RT+10ps) = AT-RT-20ps = -20ps (as earlier AT-RT = 0)
Consider CRPR.
DeleteSetup Slack = RT' - AT' = (RT-10ps-10ps) - (AT+10ps+10ps) = RT-AT-40ps = -40ps (as earlier RT-AT = 0)
DeleteHold Slack = AT' - RT' = (AT-10ps) - (RT+10ps) = AT-RT-20ps = -20ps (as earlier AT-RT = 0)
Hi Shubham,
DeleteNow it's correct!
Setup -40ps and hold -20 ps. Crosstalk on a common path will not impact on hold but it will be for setup. So
ReplyDeleteFor Setup( RT -10 -10) - (AT +10 +10) . Total 40ps is violated. For hold (AT-10 ) - (RT +10) so 20ps is violated
Yes Balaji M.
DeleteYou are totally correct here.
Why for hold it will not impact ?
DeleteFor setup analysis,
DeleteLauch and capture at different edges.
So you can't use banifit of CRPR as timing window can be different for launch and capture edge.
For hold analysis, launch and capture edge are same hence timing window is same only. Hence you can use CRPR banifit for hold.
But I see crpr for setup paths also in some blogs and reports
DeleteCRPR not only considers crosstalk.. there are other parameters as well....(Will try to cover it in upcoming posts!)
DeleteOkk thanks got it that is due to cell max min and this is for crosstalk analysis. So for cell delays whether it is. Setup or hold for both crpr benefit will be applied right ? Is that correct understanding?
DeleteThe setup slack will be -40ps and the Hold slack will be -20ps.
ReplyDeleteNote - The below calculation is considering the CRPR effect. It will have effect on setup but not on hold.
Worst Setup Slack = Rt - At = (Rt -10 -10) - (At +10 +10) = Rt - At - 40 = -40ps
Worst Hold Slack = At - Rt = (At - 10) - (Rt + 10) = -20ps
for setup, worst case would be capture -10 ps, data +10ps resulting -20. common path wont be a considered since same cell cant have different delay(cppr concept).
ReplyDeletehold wise, also same. data would be -10ps, capture would be +10ps resulting in -20ps vio
Hi choisun,
DeletePlease refer previous comments for correct answer and explanation.
Hi Sir,
DeleteIs my answer not correct?
Yes Afroz
DeleteYour answer is correct.