Guest
Greetings all:
I am simulating (DC analysis) a simple digital circuit that has many
inputs (~20): there are 2^20 (~1000,000) DC runs using a data sweep.
DC runs are independent of each other. So +mt seems promissing to
speedup the simulation. What I observe though is that a 4 processor
machine and a single processor machine have very similar CPU times in
Spectre report.
Can Spectre +mt handle this? Is there a trick in netlisting to tell it
the DC runs are independent? Any way in general to exceute many DC
runs in parallel in a single Spectre call (provided a multiprocessor
host)?
Thanks for any hint in advance
Regards,
Hrh.
I am simulating (DC analysis) a simple digital circuit that has many
inputs (~20): there are 2^20 (~1000,000) DC runs using a data sweep.
DC runs are independent of each other. So +mt seems promissing to
speedup the simulation. What I observe though is that a 4 processor
machine and a single processor machine have very similar CPU times in
Spectre report.
Can Spectre +mt handle this? Is there a trick in netlisting to tell it
the DC runs are independent? Any way in general to exceute many DC
runs in parallel in a single Spectre call (provided a multiprocessor
host)?
Thanks for any hint in advance
Regards,
Hrh.