DFT violation
时间:10-02
整理:3721RD
点击:
最近在整DFT,scan chain插好了,但是跑TMAX跑出来很多violation,不知道是不是scan chain插的有问题,还是可以忽略这些violation。请大神赐教:
总结了下,主要有这些Warning:
1、这类我觉得应该没啥问题:
Warning: Rule B7 (undriven module output pin) was violated 17088 times.
Warning: Rule B8 (unconnected module input pin) was violated 32147 times.
Warning: Rule B9 (undriven module internal net) was violated 5378 times.
Warning: Rule B10 (unconnected module internal net) was violated 23785 times.
2、觉得需要看一下,但不确定怎么整的:
Warning: Rule S19 (nonscan cell disturb) was violated 5 times.
这一类是我看了下是负沿FF的clock gating,由于库里面没有integrated的ICG,工具用OR和LATCH做了一个,但这个LATCH在插scan的是找不到可以替换的SCAN单元。不知道需不需要清。
Warning: Rule S22 (multiply clocked scan chain) was violated 1 times.
这一类应该是我设置了set_scan_configuration -clock_mixing mix_clocks的原因,应该问题不大吧?
Warning: Rule C3 (no latch transparency when clocks off) was violated 5 times.
这一类报错的单元与S19一样,不知道怎么整。
Warning: Rule C5 (LS port captured data affected by new capture) was violated 11853 times.
这一类很多,都是综合的时候加入的clock gating单元。看了下,像是ICG的E输入受扫描链上的FF的输出控制,在capture的时候,由于TE输入是低的,E的状态不确定,时钟就不一定过得去了。不知道这样理解的对不对?
Warning: Rule C6 (TE port captured data affected by new capture) was violated 146 times.
这一类感觉是caputure的时候,存在正沿到负沿的电路,这个是不是可以不管?
Warning: Rule C8 (LS port clock path affected by new capture) was violated 12889 times.
这一类也很多,也是clock gating cell,好像和C5差不多,不过这个是clock path。
Warning: Rule C9 (TE port clock path affected by new capture) was violated 144 times.
这一类感觉和C6差不多,这个是clock path。
写的有点多啊,主要后面几个不太明白怎么弄,希望有大神耐心给我讲讲,需不需要清,要清的话有什么办法清。多谢!
总结了下,主要有这些Warning:
1、这类我觉得应该没啥问题:
Warning: Rule B7 (undriven module output pin) was violated 17088 times.
Warning: Rule B8 (unconnected module input pin) was violated 32147 times.
Warning: Rule B9 (undriven module internal net) was violated 5378 times.
Warning: Rule B10 (unconnected module internal net) was violated 23785 times.
2、觉得需要看一下,但不确定怎么整的:
Warning: Rule S19 (nonscan cell disturb) was violated 5 times.
这一类是我看了下是负沿FF的clock gating,由于库里面没有integrated的ICG,工具用OR和LATCH做了一个,但这个LATCH在插scan的是找不到可以替换的SCAN单元。不知道需不需要清。
Warning: Rule S22 (multiply clocked scan chain) was violated 1 times.
这一类应该是我设置了set_scan_configuration -clock_mixing mix_clocks的原因,应该问题不大吧?
Warning: Rule C3 (no latch transparency when clocks off) was violated 5 times.
这一类报错的单元与S19一样,不知道怎么整。
Warning: Rule C5 (LS port captured data affected by new capture) was violated 11853 times.
这一类很多,都是综合的时候加入的clock gating单元。看了下,像是ICG的E输入受扫描链上的FF的输出控制,在capture的时候,由于TE输入是低的,E的状态不确定,时钟就不一定过得去了。不知道这样理解的对不对?
Warning: Rule C6 (TE port captured data affected by new capture) was violated 146 times.
这一类感觉是caputure的时候,存在正沿到负沿的电路,这个是不是可以不管?
Warning: Rule C8 (LS port clock path affected by new capture) was violated 12889 times.
这一类也很多,也是clock gating cell,好像和C5差不多,不过这个是clock path。
Warning: Rule C9 (TE port clock path affected by new capture) was violated 144 times.
这一类感觉和C6差不多,这个是clock path。
写的有点多啊,主要后面几个不太明白怎么弄,希望有大神耐心给我讲讲,需不需要清,要清的话有什么办法清。多谢!
有大神么
都算比较正常的设计,
tmax没报 Error , coverage 又满足要求, 就先不管这些warning.
多谢回复,coverage 比较低,仿真没跑通。不知道啥原因。这种warning报出来,TMAX需要做什么特殊的设定么?谢谢
coverage 比较低,
---
先提高coverage ,这些warning能在RTL解决最好。
仿真没跑通。
---
这个情况就多了,SDF ,pattern都有可能。
RTL不能修改,还能提高coverage么?
不改设计,是否 所有FFs都串到 chain 里面了?
DFT 模式下应该有很多 tie逻辑,无法翻转,造成coverage 低。
谢谢,我再研究研究
先找出没cover到的的原因,这些警告一般没问题
整理的不错,点个赞
B8有问题吧,input没驱动?