Hi, I am writing a converter for our file format (about 150 channels each of time/packet series of 30kS/s 2byte wide), if I can make it faster than it seems to be we want to support it as our new file format during recording.
Is there any hint how to select the chunk size? how the number of packet tables in one file may affect the performance (should I avoid creating 150 packet tables, and instead use a single table and index it on channel ID)? Does it matter in which order packet tables are created, i.e. if it is better to create all the packet tables at the beginning before any data is appended, or it does not matter if one is created when the first packet of that channel appears. How does using a variable length type in packet table affect the performance? Thanks for any information Regards, dashesy _______________________________________________ Hdf-forum is for HDF software users discussion. Hdf-forum@hdfgroup.org http://mail.hdfgroup.org/mailman/listinfo/hdf-forum_hdfgroup.org
