PSRDADA block updates

[telegraphic]

These updates are from https://github.com/Molonglo/bifrost, which is used for UTMOST-2D.

An alternative to this merge is to move PSRDADA functionality into a plugin.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 9.66543% with 243 lines in your changes missing coverage. Please review.

Project coverage is 67.09%. Comparing base (898e4a8) to head (60e7a92).
Report is 554 commits behind head on master.

Files with missing lines	Patch %	Lines
python/bifrost/blocks/dada_file.py	0.00%	112 Missing ⚠️
python/bifrost/blocks/psrdada.py	16.66%	110 Missing ⚠️
python/bifrost/psrdada.py	16.00%	21 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #194      +/-   ##
==========================================
- Coverage   68.19%   67.09%   -1.11%     
==========================================
  Files          65       67       +2     
  Lines        5515     6026     +511     
==========================================
+ Hits         3761     4043     +282     
- Misses       1754     1983     +229     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[jaycedowell]

Is any encoding needed here for Py2 vs Py3?

[telegraphic]

very good chance there is :/. I guess I should write a unit test

[jaycedowell]

I don't think we have anything that covers psrdata now. We probably don't even install psrdata to test the wrapping.

[jaycedowell]

I'm looking back through the code here and I'm not convinced that encoding is needed here.

[jaycedowell]

The CI now installs PSRDADA and the Bifrost wrappers get generated. @telegraphic do you have a .dada file I can test with?

[ledatelescope]

Tensor core arithmetic also accelerates 16 bit type data.

…

On 2/14/23 9:49 AM, jaycedowell wrote: In python/bifrost/blocks/psrdada.py <#194 (comment)>: > + Returns: + s (str): DADA header string with padding to hdrlen + """ + s = "HDR_VERSION 1.0\n" + s+= "HDR_SIZE %i\n" % hdrlen + keys_to_skip = ('HDR_VERSION', 'HDR_SIZE') + + # update parameters from bifrost tensor + if '_tensor' in hdr_dict.keys(): + print(hdr_dict['_tensor']) + dtype = hdr_dict['_tensor']['dtype'] + dtype_vals = { + 'cf32': { 'NBIT': '32', 'NDIM': '2' }, + 'f32': { 'NBIT': '32', 'NDIM': '1' }, + 'ci8': { 'NBIT': '8', 'NDIM': '2' }, + 'i8': { 'NBIT': '8', 'NDIM': '1' } } Are additional data types expected/allowed? --------------------------------------------------------------------------------

[ledatelescope]

I raised the notion of 16-bit  to accommodate directly time-series output from many-bit ADCs.  i16 would do that.  Quick examination suggests that uint16 is recognized by CUDA7(?) but using it may not result in speed advantage over 4-byte integer math.  Proper research required.

…

On 2/15/23 1:58 AM, Danny Price wrote: ***@***.**** commented on this pull request. -------------------------------------------------------------------------------- In python/bifrost/blocks/psrdada.py <#194 (comment)>: > + Returns: + s (str): DADA header string with padding to hdrlen + """ + s = "HDR_VERSION 1.0\n" + s+= "HDR_SIZE %i\n" % hdrlen + keys_to_skip = ('HDR_VERSION', 'HDR_SIZE') + + # update parameters from bifrost tensor + if '_tensor' in hdr_dict.keys(): + print(hdr_dict['_tensor']) + dtype = hdr_dict['_tensor']['dtype'] + dtype_vals = { + 'cf32': { 'NBIT': '32', 'NDIM': '2' }, + 'f32': { 'NBIT': '32', 'NDIM': '1' }, + 'ci8': { 'NBIT': '8', 'NDIM': '2' }, + 'i8': { 'NBIT': '8', 'NDIM': '1' } } There isn't really a strict definition AFAIK. There is no header keyword for float/integer, but I think it's reasonable to add cf64/f64. Would need to decide between i16/ci16 and f16/cf16 though — Reply to this email directly, view it on GitHub <#194 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACL54MQIQ7KCUU3K6EQ33PDWXR5CFANCNFSM6AAAAAAS7I5RJ4>. You are receiving this because you commented.Message ID: ***@***.***>