Produce very large test patterns

[dinyar]

Given some very rare bugs (i.e. only O(1) errors in the large testfile) it would be useful to have a significantly larger testfile. I believe the current file is ~10 MB large in compressed format, it may be worth having two distinct files -- one ~100 MB for regular testing with testbenches and one between 500 MB and 1 GB for overnight testing of new firmware releases.

[jlingema]

ok -- this should be trivial to produce with the L1Ntuples we have now that contain O(1 M) events.

[jlingema]

I'm guessing these will not be very helpful for the testbench and should only be made for hw?

[dinyar]

I was thinking that the smaller of the two new ones could actually be used
for testbench testing too. I'm not sure how large the ttbar testbench file
is currently, but from memory I remember something ~70 MB. A x10 larger one
would run for about an hour which should be acceptable for semi-regular
running.

On Wed, Oct 28, 2015 at 9:23 AM Joschka Lingemann [email protected]
wrote:

I'm guessing these will not be very helpful for the testbench and should
only be made for hw?

—
Reply to this email directly or view it on GitHub
#51 (comment)
.