Trouble Preloading Data for timed digital IO in 6612

[AmauryD]

I want to simulate an encoder and I want to finely control distance and speed.

I have written a 4bit long pattern to the digital lines and I want to clock it out at some sample rate.

If my sample rate is too high I get an error while waiting for task done:

nidaqmx.errors.DaqError: Onboard device memory underflow. Because of system and/or bus-bandwidth limitations, the driver could not write data to the device fast enough to keep up with the device output rate.
Reduce your sample rate. If your data transfer method is interrupts, try using DMA or USB Bulk. You can also reduce the number of programs your computer is executing concurrently.
Task Name: tach_out2

I should be writting everything to the board ahead of start, but some how it seems the driver is leaving the data in memory to be fetched at start, any thoughts on how I can push the data down to onboard? 

cpm = 48609
#   __
#__|  |__|
#  __    _
#_|  |__|
#1 0 2 3
#__    __
#  |__|  |
#  __    _
#_|  |__|
#2 3 1 0

fwd_bits = [0, 1, 3, 2]
rev_bits = [2, 3, 1, 0]
i=0

for i,d in enumerate(pattern):

     do_task = mx.Task(f'tach_out{i}')
     do_task.do_channels.add_do_chan(dev + "/port0")

     do_task.timing.cfg_samp_clk_timing(d.speed * cpm, sample_mode=AcquisitionType.FINITE, samps_per_chan=d.distance * cpm)
     do_task.out_stream.regen_mode = RegenerationMode.ALLOW_REGENERATION
     do_task.out_stream.output_buf_size=len(fwd_bits)
     do_task.out_stream.offset=0
     print(do_task.out_stream.output_onbrd_buf_size)
     do_task.write(rev_bits if d.rev else fwd_bits, auto_start=False)
     time.sleep(.1)
     do_task.start()
     print(f"{d.speed}m/s, {d.distance}, {'rev' if d.rev else 'fdw'} {d.distance/d.speed}")

     while not do_task.is_task_done():
          print(do_task.out_stream.curr_write_pos/(d.distance * cpm))
          time.sleep(.1)
     do_task.stop()
     do_task.close()

[Kevin_Price]

I only program in LabVIEW and don't know much about any text API syntax.  In LabVIEW there's a deeply-nested channel property that lets me set "Use Only Onboard Memory" to True.

Another way to approach this is to set up your PC-side task buffer to contain many consecutive copies of the bit pattern.  Heck, memory's cheap, replicate the 4-state pattern 1000 times to make a 4000-sample buffer.  Betcha that'll work too.   (Then you can back off to see how many replications you need.  I wouldn't be surprised at all if it still worked with only 10.)

The size of the task buffer and the sample rate determine how often the driver must transfer data down to the device.  Your system can't seem to keep up with your sample rate when it can only move 4 samples max at a time.  With a bigger buffer, it can move more samples less often and keep up.

-Kevin P