12-15-2005 03:47 PM
Hi Micaela
Here is a part of the error message, attached is the modified vi that runs at 1Mhz (not faster) on the Xeon system.
I even had to add the 2ms delay between starting the data write and starting the clock !
The answer we got and accepted was to wait for the next daqmx release.
Later on we found another "difficult" problem. When reading analog inputs at 1MHz (8 times 30 channels and that 240 times)
we got 240 plots of 240 datapoints. This happily works ok. But... drawing a plot of this awful lot of datapoints speeds the system up!
We accept that system behaviour is not completely predictable in a hyperthreaded system (2 real but 4 hyperthreaded cpu's) but that our system speeded up from 200ms to 170ms for such amount of data was unexpected much. But of course we keep the plotting for the moment.
Original message:(Reference#5178-EX7193) NI Mseries NI-6259 too slow in Xeon system
we were trying to dump digital output data at 10MHz.this runs OK on a normal machine (e.g. DELL GX280) but sucks in a DELL Precision 670.
The reason is probably the multiprocessor architecture but also disabling the second processor in the BIOS does not help.
We tried the latest version of the DAQMX drivers (release candidate driver October 2005) but no effect.
The error message is -200016
DAQmx Wait Until Done.vi<append>
<B>Task Name: </B>_unnamedTask<0>
Possible reason(s):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, alter the data transfer method (from interrupts to DMA),
use a product with more onboard memory, or reduce the number of programs your computer is executing concurrently.
We expect that this works on such a machine, at least has the same performance as a slower desktop.
The attached routine is not meant as an example how to write daqmx data but is the result of try and error to get it running.
The data normally sent is about 100k byte and is included in the example.
Modifying the amount of data (a lot less data) gave also strange results (data not sent out but no error message)
If you have suggestions we will try as soon as the owner of the machine has some spare time. The system now is doing experiments.
12-16-2005 10:45 AM
12-17-2005 02:34 PM
12-19-2005 05:54 PM