W
walala
Guest
Dear all,
I am design with Silicon Ensemble P&R:
The floorplan is of size 2400x1800, flip + abut, I am changing row spacing:
Can experts take a look at these changes, and give me some comments:
particularly, I don't know why row spacing play such important row, and for
case 1), row spacing = 0 track, along with flip + abut, it should be very
good, but why it has so many violations after WROUTE? Then I increased row
space = 1 track in case 2), and the violations drastically reduced to 0 in
WROUTE, but still there is one open in CONNECT RING; so I think increasing
row space is good, but then I try case 3), row spacing increases to 2
tracks, now the result is even worse, there are 37 violations remaining
after 29 search and repair pass in WROUTE; and the opens in CONNECT RING
increased...
You see, it's so strange to try out those combinations of numbers,... there
are so many things need to try, I have been doing these blindly in the past
one month. My best result is 0 violations, 1 open in CONNECT RING, and about
40% utilization ratio. Can you solve that OPEN problem? Can I do better in
terms of utilization?
Can any experts here give me some help?
Thank you very much,
-Walala
Case 1) row spacing = 0 track:
FINIT FLOOR rowu 0.85 rowsp 0 blockhalo 2000 f x 240000 y 180000 abut xio
300
0 yio 3000 ;
FINIT FLOORPLAN rowu 0.85 rowsp 0 blockhalo 2000 f x 240000 y 180000 abut
xio
3000 yio 3000 ;
Results:
core row utilization = 35.72%
core utilization = 85.00%
201 rows
2 opens when connect ring
WROUTE first pass 16797 violations remained, auto stopped
More strangely, if I arbitrarily change the rowu 0.85 to 0.5 in the above
command, the core row utilization = 35.72% doesnot change, instead, the core
utilziation changed from 85% to 50%....(which is against my belief, because
I thought "rowu" means row utilization, and not the core utilization...)
Case 2) row spacing = 1 track:
FINIT FLOOR rowu 0.85 rowsp 96 blockhalo 2000 f x 240000 y 180000 abut xio
30
00 yio 3000 ;
FINIT FLOORPLAN rowu 0.85 rowsp 96 blockhalo 2000 f x 240000 y 180000 abut
xi
o 3000 yio 3000 ;
Results:
core row utilization = 39.67%
core utilization = 76.57%
181 rows
1 opens when connect ring
WROUTE 0 violations remaining, after 14 search & repair, first pass 6486
viola
tions,
Case 3) row spacing = 2 track:
FINIT FLOOR rowu 0.85 rowsp 192 blockhalo 2000 f x 240000 y 180000 abut xio
3
000 yio 3000 ;
FINIT FLOORPLAN rowu 0.85 rowsp 192 blockhalo 2000 f x 240000 y 180000 abut
x
io 3000 yio 3000 ;
Results:
core row utilization = 43.78%
core utilization = 69.66%
165 rows
3 opens when connect ring
WROUTE 37 violations remaining, after 29 search & repair, first pass 6385
viol
ations,
I am design with Silicon Ensemble P&R:
The floorplan is of size 2400x1800, flip + abut, I am changing row spacing:
Can experts take a look at these changes, and give me some comments:
particularly, I don't know why row spacing play such important row, and for
case 1), row spacing = 0 track, along with flip + abut, it should be very
good, but why it has so many violations after WROUTE? Then I increased row
space = 1 track in case 2), and the violations drastically reduced to 0 in
WROUTE, but still there is one open in CONNECT RING; so I think increasing
row space is good, but then I try case 3), row spacing increases to 2
tracks, now the result is even worse, there are 37 violations remaining
after 29 search and repair pass in WROUTE; and the opens in CONNECT RING
increased...
You see, it's so strange to try out those combinations of numbers,... there
are so many things need to try, I have been doing these blindly in the past
one month. My best result is 0 violations, 1 open in CONNECT RING, and about
40% utilization ratio. Can you solve that OPEN problem? Can I do better in
terms of utilization?
Can any experts here give me some help?
Thank you very much,
-Walala
Case 1) row spacing = 0 track:
FINIT FLOOR rowu 0.85 rowsp 0 blockhalo 2000 f x 240000 y 180000 abut xio
300
0 yio 3000 ;
FINIT FLOORPLAN rowu 0.85 rowsp 0 blockhalo 2000 f x 240000 y 180000 abut
xio
3000 yio 3000 ;
Results:
core row utilization = 35.72%
core utilization = 85.00%
201 rows
2 opens when connect ring
WROUTE first pass 16797 violations remained, auto stopped
More strangely, if I arbitrarily change the rowu 0.85 to 0.5 in the above
command, the core row utilization = 35.72% doesnot change, instead, the core
utilziation changed from 85% to 50%....(which is against my belief, because
I thought "rowu" means row utilization, and not the core utilization...)
Case 2) row spacing = 1 track:
FINIT FLOOR rowu 0.85 rowsp 96 blockhalo 2000 f x 240000 y 180000 abut xio
30
00 yio 3000 ;
FINIT FLOORPLAN rowu 0.85 rowsp 96 blockhalo 2000 f x 240000 y 180000 abut
xi
o 3000 yio 3000 ;
Results:
core row utilization = 39.67%
core utilization = 76.57%
181 rows
1 opens when connect ring
WROUTE 0 violations remaining, after 14 search & repair, first pass 6486
viola
tions,
Case 3) row spacing = 2 track:
FINIT FLOOR rowu 0.85 rowsp 192 blockhalo 2000 f x 240000 y 180000 abut xio
3
000 yio 3000 ;
FINIT FLOORPLAN rowu 0.85 rowsp 192 blockhalo 2000 f x 240000 y 180000 abut
x
io 3000 yio 3000 ;
Results:
core row utilization = 43.78%
core utilization = 69.66%
165 rows
3 opens when connect ring
WROUTE 37 violations remaining, after 29 search & repair, first pass 6385
viol
ations,