search

GRIFFINCollaboration / GRSISort

A lean, mean, sorting machine.
MIT License
21 stars 54 forks source link

Sorting test data with new GRSISort #781

Closed MikeBowry closed 8 years ago

MikeBowry commented 8 years ago

I had a go at sorting a run (#7847 in /tig/daq08_data/Tests/) that we took with a single HPGe clover in the lab with a freshly downloaded GRSISort. At the moment I see nothing in the FragmentTree. I'm almost certain that I'm missing something (some flags etc.). Image attached.

This run sorts successfully on grifuser (i.e. where we skip the piled-up hits). We used a Bi-207 source. I think this run used firmware version grif16_30Sep16.sof, writing data in the GRF4 format. The data we will take in the array will use a different version, but the data format should be the same. Has anyone else tried sorting these lab runs?

07oct2016

MikeBowry commented 8 years ago

Just noticed that runinfo has 0 channels, perhaps this is an issue. With the test daq, we often omit some of the information that is present in the MIDAS files taken with the array daq.

r3dunlop commented 8 years ago

0 channels in RunInfo is just a bug that is in the new code. There are still 84 Channels in the TChannel from the look of it.

It’s doing something weird because there is no sub run number. I don’t think this is the issue, but it is one issue.

I personally won’t get a chance to look at this data until I’m in Vancouver on Monday.

MikeBowry commented 8 years ago

Ok. Yup I think with these test runs I always needed to provide a .cal file so that TChannel behaves properly. Rather naively I tried changing the name of the midas file from run07847.mid to run07847_000.mid but this didn't help with the sort.

There is some chance that we will take data this evening with the new cards. If anyone has time to take a look either this evening or on Saturday it would help a lot with preparations.

r3dunlop commented 8 years ago

Are you sure the data words are coming thorough properly? Have you tried running without errors suppressed and seeing where it’s failing?

MikeBowry commented 8 years ago

It might be that it is failing all the time. I ran the command above with no suppress-error flag. Is it suppressed automatically?

r3dunlop commented 8 years ago

I don’t believe so. This makes me think there is something up with the MIDAS event such that it isn’t being recognized as GRIFFIN.

VinzenzBildstein commented 8 years ago

Yeah, this is because the merging of the new version discarded the check for the bank type GRF4. I've fixed this and am checking it right now. Once I've verified it's sorting now, I'll push the fix.

VinzenzBildstein commented 8 years ago

Fixed by PR #782.

VinzenzBildstein commented 8 years ago

BTW, we might want to fix the code so that it gives an error if the bank name is unknown. I started this in PR #783, but right now it's just an error message with the midas event serial number, but no bank name, maybe we can add that later.

MikeBowry commented 8 years ago

Great, I can check this on the S1601 account. Ok, so now I see data: spectra_run7847

MikeBowry commented 8 years ago

There is a brand new run taken in the array with two revision 1 GRIF-16s, grsmid00:/data3/griffin/Tests/run07768_000.mid. The source is 152Eu. Fragment sorts fine (from what I can tell) but I am having some issues sorting into analysis files, which I would like to use to look at the CFD time difference (probably not optimized at the moment anyway). If I do:

grsisort -a fragment07768_000.root

It appears to sort OK (showing ~1.9 million events) but hangs up here: capture3 Unless I am just being impatient.

r3dunlop commented 8 years ago

Can you try something like adding the command --sort-depth 100000000 . Add less zeros if it crashes

On Oct 9, 2016, at 5:42 PM, Michael Bowry notifications@github.com<mailto:notifications@github.com> wrote:

There is a brand new run taken in the array with two revision 1 GRIF-16s, grsmid00:/data3/griffin/Tests/run07768_000.mid. The source is 152Eu. Fragment sorts fine (from what I can tell) but I am having some issues sorting into analysis files, which I would like to use to look at the CFD time difference (probably not optimized at the moment anyway). If I do:

grsisort -a fragment07768_000.root

It appears to sort OK (showing ~1.9 million events) but hangs up here: [capture3]https://cloud.githubusercontent.com/assets/14794662/19225042/8b6e47ce-8e47-11e6-84a7-5acedb605402.PNG Unless I am just being impatient.

You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://github.com/GRIFFINCollaboration/GRSISort/issues/781#issuecomment-252524254, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AFLTwAup1fXqjT_nBCOcremlzn9no3A8ks5qyYnzgaJpZM4KRiCJ.

r3dunlop commented 8 years ago

We tried sorting here this morning and it worked. It might just be getting stuck on a lock somewhere. The computers at TRIUMF had problems with the old (independent) version as well. Not sure what is going on exactly.

r3dunlop commented 8 years ago

We think we've found the problem. There is something strange going on when you sort with -a with no -q option. It gets stuck in the command prompt somehow. We will look into this some more but for now the solution it to sort with the -aq.

MikeBowry commented 8 years ago

The -aq flag works for me, even without the --sort-depth flag.

AdamGarnsworthy commented 8 years ago

Have all the data features from the new firmware been examined for consistency with the specification document, GRF4 format is Table 5.9 of https://rawgit.com/wiki/GriffinCollaboration/GRSISort/technical-docs/GRIFFIN_Event_Format.pdf? And check that the results look correct? If any bugs need to be fixed in the firmware we need to identify these as soon as possible. Thanks.

r3dunlop commented 8 years ago

We will get right on it.

pcbend commented 8 years ago

Just a quick comment, when simply giving the -a command alone, the program assumes you want to look at the tree, so it stays open. It is also why the "normal" status isn't displayed that comes with the minus q command. Particularly if you are using the new histogramming feature one can examine the histogram live from the terminal while the file sorts in the background.

Pressing enter would make the prompt reappear saying it is ready for commands.

r3dunlop commented 8 years ago

I don’t believe it was making the prompt appear. It was just creating more whitespace.

pcbend commented 8 years ago

well that is new. I had a prompt locally, but I am still using root 5...

On Wed, Oct 19, 2016 at 11:32 AM, Ryan Dunlop notifications@github.com wrote:

I don’t believe it was making the prompt appear. It was just creating more whitespace.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/GRIFFINCollaboration/GRSISort/issues/781#issuecomment-254849833, or mute the thread https://github.com/notifications/unsubscribe-auth/AAacI6YL2a1_V9kyc7_lTiJrRnnqQ3Bpks5q1jgqgaJpZM4KRiCJ .

VinzenzBildstein commented 8 years ago

I think this is done.