stom / stomna on a large balanced panel hangs indefinitely in EViews 14 (instant in EViews 12)
Posted: Fri Jun 05, 2026 1:47 am
Hi,
I think I've hit a performance regression in series-to-matrix conversion (stom / stomna) between EViews 12 and EViews 14.
I run a model stochastic solve and save the successful repetitions to a new page (the p= option on Model::stochastic). That creates a balanced panel page: 70,000 cross-sections (repetitions) x 20 quarters = 1,400,000 observations. Each tracked endogenous variable is an ordinary numeric series in that page, with valid data.
A single line:
stom(timeseries, target)
never returns. EViews goes to "Not Responding", the busy spinner stays up no error and no crash for hours. The identical operation on the identical data completes in well under a minute in EViews 12.
Is this a known regression in 13/14? I couldn't find any mention of stom/stomna in the 13 or 14 patch notes.
Is the slowdown specific to converting a panel-structured series - i.e. is unstructuring the page first the recommended workaround, or is there a better way to pull a 20 x 70,000 matrix out of a balanced-panel page?
Thanks.
I think I've hit a performance regression in series-to-matrix conversion (stom / stomna) between EViews 12 and EViews 14.
I run a model stochastic solve and save the successful repetitions to a new page (the p= option on Model::stochastic). That creates a balanced panel page: 70,000 cross-sections (repetitions) x 20 quarters = 1,400,000 observations. Each tracked endogenous variable is an ordinary numeric series in that page, with valid data.
A single line:
stom(timeseries, target)
never returns. EViews goes to "Not Responding", the busy spinner stays up no error and no crash for hours. The identical operation on the identical data completes in well under a minute in EViews 12.
Is this a known regression in 13/14? I couldn't find any mention of stom/stomna in the 13 or 14 patch notes.
Is the slowdown specific to converting a panel-structured series - i.e. is unstructuring the page first the recommended workaround, or is there a better way to pull a 20 x 70,000 matrix out of a balanced-panel page?
Thanks.