tran MC sim result size too big

This is almost certainly due to the fact that with normal APS runs there is some collapsing of parallel devices which can lead to fewer devices and nodes/terminals in the output database. That parallel simplification is not performed when running Monte Carlo mismatch simulation (to allow for mismatch between separate devices), and that could lead to a larger result database.

That said, I'm rather surprised that the difference is so significant - that seems a surprisingly large change - you should contact customer support.

Of course, the simplest approach is not to save all signals! Just save what you're interested in.

Andrew

Hi Andrew,

My environment is

CDS: IC618.500.24

Spectre: 21.1.0.460, ISR10

And my simulator is spectreX, errpreset = vx, NOT APS. 

Also i have included some big pex view. 

As I mentioned, the standard save option is "selected", but NOT "all" NOR "allpub". 

Do you got any extra clue?

Thanks a lot for your help. 

Very hard to know without seeing it (which is why I suggested customer support). Maybe doing an "ls -l" on the psf directory would give a clue (you might need to poke around in the maestro results structure to see).

Andrew

Very hard to know without seeing it (which is why I suggested customer support). Maybe doing an "ls -l" on the psf directory would give a clue (you might need to poke around in the maestro results structure to see).

Andrew

Dear sjwprcker,

sjwprcker said:

Also i have included some big pex view. 

As I mentioned, the standard save option is "selected", but NOT "all" NOR "allpub". 

Do you got any extra clue?

Two other thoughts came to mind that I thought I should pass by you.

1. You mentioned you were using a "big pex view" - which I assume suggests the netlist was composed using an extracted view of the circuit under test. I do not know if you are using an RC or C only basd extracted view (i.e., extracted view contains both parasitic resistances and capacitances or just parasitic capacitances). I assume since it appears so large, it may be an RC based extracted view netlist. If your selected nodes to save are based on a schematic view, the resulting saved signals will include all the fragmented nets associated with that node in the extracted view. Hence, suppose you intended to save a single node, say V1, that connects I1 and I2. With an extracted view based netlist, it will save every intermediate - or fragmented - net from the output of I1 to the input of I2. This can be a few or many hundreds of nets for a long trace between I1 and I2. I have seen this result in very large databases. There are a couple of ways around this.

2. You also noted your save option was set to "selected". Is the save option set to "selected" in both the Transient GUI "Outputs" tab as well as your ADE Outputs GUI?

Shawn

I was wondering whether maybe it wasn't the transient database that was large, but the instances, operating point or something else that was big. Very hard to know without some details as to what's in the psf directory as I said.

Andrew