LabVIEW
array manipulation from fpga to tdms
Hi Labview community,
i'm working to optimize my code.
i acquire data from fpga using DMA
i read DMA in real time vi and i need to manipulate it before logging it usind advanced tdms write
data manipulation means that: reducing size of DMA (desampling) but having also option for averaging
i need help to understand which is the better way. i used to manipulate like:
I NEED HELP TO:
A) is there a vi similar to Decimate (continous).vi that allow me to pass direct the array interleaved data that come from FPGA without access it to retrive channels? there is something that accept data I32 or I16 instead of DBL (5 picture above)?
B) which is the best way (1, 2 , 3 or 4 picture above) to prepare FPGA array to TDMS advanced write?
C) data to TDMS advanced write could be interleaved or not, it depends by the way i prepare it. which is the better way?
in the real application that could be some real order of data:
PXI
20/30 channels at 500KHz logged continous
20/30 channels at 100KHz logged continous
100/150 channels at 1 KHz logged continous
CRIO
10 channels at 500KHz logged short time
50/100 channels at 1 KHZ logged continous
thank you for support
Michele
I do not know which one is faster. Below is an option where all manipulations are in-place. Should save memory but may be slower. Data channels are in columns, so you should be able to use interleaved TDMS data. Lastly, never used the decimate functions, but believe you need to keep double output. Decimate data should low pass filter the data then decimate to prevent aliasing. That low pass filter should introduce non-integer values.
EDIT: Reshape final array for TDMS input
I like mcduff's answer.
If you wanted to try to offload the decimation to the FPGA before it even gets to the host/rt, I believe there are a couple implementations floating around. The "Rational Resampler" on the FPGA Math and Analysis palette can support up to 32 channels interleaved (not sure why the limit is 32 channels):
