Re: [Linux-gpib-general] Spec doc for Agilent 82350b?

[Stanley A. <st...@ne...>]

Dave:
Thanks for responding.
We had a recurring problem with the agilent_82350b_accel_read() function. In almost all cases we are able to request and receive the correct number of bytes, but for one PAD/SAD pair we are requesting five bytes and receiving six.
What makes it strange is that we have multiple units under test (UUTs) which are identical except for the PAD. The SADs for each PAD are the same. We are querying SAD 1 (the "status register") on each UUT and are getting five bytes from each, except for this one.
We're migrating from the National Instruments GPIB PCIe card to the Keysight PCIe card. We've been using the NI driver with the NI card to talk to these UUTs for many years (20+) and have always received five bytes from each unit. So we don't believe the one unit itself is having a problem.
Debugging showed the problem was associated with the acceleration/fifo portion of the read() operation. So we modified read() to avoid the accel/fifo logic altogether & the problem went away. The modified read() calls the tms9914_read() in all cases; but now reading large blocks of data is fairly slow.
We thought we could review the spec to see if there was a problem with the accel/fifo logic. We do need to be concerned about throughput.
Regards,

Stanley