Tool/software:
Hi, TI Expert
we have several group of exams as follows:
case 1: pipeline depth = 1 model A running on C7x-1, model B running on C7x-2
[C7x_1 ] 10542.613267 s: =================tidl process exhausted total=95 ms //A model
[C7x_2 ] 10542.635209 s: =================tidl process exhausted total=101 ms //B model
[C7x_1 ] 10545.013252 s: =================tidl process exhausted total=92 ms //A model
[C7x_2 ] 10545.015831 s: =================tidl process exhausted total=96 ms //B model
case 2: Pipeline depth=2,model A running C7x-2,model B running on C7x-1
[C7x_2 ] 9541.888283 s: =================tidl process exhausted total=95 ms //A model
[C7x_1 ] 9541.910465 s: =================tidl process exhausted total=101 ms //B model
[C7x_2 ] 9544.168862 s: =================tidl process exhausted total=93 ms //A model
[C7x_1 ] 9544.187310 s: =================tidl process exhausted total=100 ms //B model
case 3: pipeline depth=1 only model B running on C7x-2
[C7x_2 ] 11459.889934 s: =================tidl process exhausted total=69 ms
case 4: pipeline depth=1 model A and model B both running on C7x-2
[C7x_2 ] 12410.281604 s: =================tidl process exhausted total=85 ms //A model
[C7x_2 ] 12410.351151 s: =================tidl process exhausted total=69 ms //B model
[C7x_2 ] 12412.564401 s: =================tidl process exhausted total=83 ms //A model
[C7x_2 ] 12412.633824 s: =================tidl process exhausted total=69 ms //B model
so i have doubt on the test result above. why is the time exhaust when two models are deployed on different dsp more than that when those are deployed on single dsp?
looking forward to your replay!
Thanks