NI488.2 drivers for Intel-based Mac?

[costap]

It just dawned on me that the reason that regressing to v2.0 does not work for me is due to VISA not working. Is that correct? Could 488 version 2.0 work on an Intel Mac if you are using VISA in Labview or are you dead in the water if you are using VISA on an Intel Mac for GPIB?

[JasonSmith]

Hi costap,

I spent the morning working an a Mac Pro so that I would be able to answer your question.

Our initial testing to see how NI-488.2 would behave on the Intel-based Macs was done before LabVIEW 8.2 was released, and had focused on the use of direct NI-488.2 calls rather than NI-VISA.  In my further testing today I found that from LabVIEW I was only able to communicate if using LabVIEW 8.0 with the 488 API (I did not try LabVIEW 7.x, but I find it likely that it would work like LabVIEW 8.0).  I was unsuccessful communicating with the GPIB-ENET/100 at all from LabVIEW 8.2, or using NI-VISA from either version of LabVIEW.

So it appears that for now the only GPIB option on Intel-based Macs is to use the following:

• LabVIEW 8.0 (7.x likely to work as well)
• NI-488.2 Version 2.0 for Mac OS X
• GPIB-ENET/100
• 488 API from LabVIEW (No VISA support)

I hope that is helpful.  Please let me know if I can help answer any additional questions.

Jason S.
GPIB & Serial
National Instruments

[sth]

Thanks Jason for the quick list of compatible systems. It is a rather limited list since we changed all our software to VISA calls awhile ago. I would like to see the availability improve since I just had someone asking about this for buying a system. This would be my order for priority.

get VISA working for ENET-GPIB for LV 8.20 for intel based macs
get traditional GPIB calls for ENET-GPIB working for LV 8.20 for intel based macs
(theoretically all the traditonal calls are actually emulated VISA?)
get PCIe-GPIB VISA and traditonal 488 working for LV 8.20 for intel based macs.

The first two should not be all that difficult since it is written in a higher level language.

[UrsL]

Jason, Scott et. al.

I am primarily interested in the GPIB-ENET/100 solution to drive instruments remotely in combination of conventional data acquisition made by a PCIe-6152 from LV 8.20/DAQmxBase 2.1. In contrast to Scott my own GPIB experience with LV is sparse and I am not sure if talking GPIB through the ENET-adapters is completely done with VISA. I am aware that the most recent VISA driver for MacOSX is 4.0 now. For time being I am still working with PPC-Macs but this will of course change in the future and I want to be sure that my solutions will hold up for quite some time from now.

Do you have any comments or ideas about what hardware to get for my software versions which will be fully upgradable to MacIntel native performance?

Thanks

Urs

Urs Lauterbrg
Physics demonstrator
Physikalisches Institut
University of Bern
Switzerland

[sth]

Urs, Thanks!

I try to keep up but these upgrades are a little stealthy? I am still at version 3.5. Now the readme file for 4.0 (is dated July 2006, and the version was finalized on 8/7/2006 so there could even be later revisions?).

NI-VISA 4.0
Added support for Mac OS X on the x86 architecture, aka Mactel.
Added support for 64-bit data transfers in register-based operations, to support the VISA 4.0 specification.
Added support for DTR/DSR flow control. The Agilent 34401A requires DTR/DSR flow control when used via Serial. Note that not all Serial port drivers on the Mac support DTR/DSR. We have tested the Agilent 34401A with NI-VISA 4.0 using the Keyspan USA-19HS and verified that it works properly.
The I/O control in LabVIEW 8.0 would not always honor the parameters you specified in the filtering dialog, depending on the order in which you made changes. This is fixed.
viScanf would crash when reading indefinite-length block data. This is fixed.
For ENET-Serial INSTR sessions, framing errors were occasionally not detected. This is fixed.
Improved performance of NI Spy.
For TCP/IP SOCKET sessions, re-enabling the termination character between read operations would incorrectly flush the internal buffer. This is fixed.
For TCP/IP SOCKET sessions, reading less than the entire amount of data that the device sent would not work correctly for asynchronous read operations. This is fixed.

This might imply that VISA will work with 8.20 AND with Macintel? Jason, what version of VISA did you test? This is encouraging and hopefully will solve the first 2 of my requests. Kudos to the VISA team!

BTW: link to driver http://digital.ni.com/softlib.nsf/websearch/1F9681F90A20F3BF862571C3005500F0

And download ftp://ftp.ni.com/support/visa/drivers/macosx/4.0/

Message Edited by sth on 10-08-2006 11:54 AM

[UrsL]

Thanks Scott

for pitching in your thought to this. If NI-VISA 4.0 does indeed support GPIB-ENET/100 with LV 8.20 and Macintel then I guess I am all set. Can you confirm our optimistic assumptions Jason?

Never the less a big hand to the VISA team. They have always provided great Mac support, consistently from the early days until the present time. They did this silently without making a big fuzz about it and I really appreciate their splendid work! Kudos to whoever is responsible and who does the NI-VISA for the Mac job.

Urs

Urs Lauterburg
Physics demonstrator
Physikalisches Institut
University of Bern
Switzerland

[JasonSmith]

Hi Scott,

While NI-VISA 4.0 does work with with Intel-based Macs and LabVIEW 8.20, it does not support every interfaces which is supported on PowerPC-based Macs, as can be seen by this feature table.  GPIB support is not supported due to the lack of an appropriate NI-488.2 driver on the Intel-based Mac platform.

With regards to the palettes in LabVIEW, the GPIB palette actually does not utilize NI-VISA, and communicates directly through the NI-488.2 driver, and the NI-VISA API requires NI-488.2 in order to work with with GPIB interfaces.  Due to this, there will be no support for GPIB devices in LV 8.2 on Intel-based Macs until NI-488.2 is available for that platform.  It may be possible to utilize the ENET/100 if you force LV 8.2 to run in Rosetta, but I have not tested this and do not think this will be a satisfactory solution for most of you.

Please continue your discussion in this forum.  I am monitoring this closely, and will let you know when I have any more information.

Jason S.
GPIB & Serial
National Instruments

[JoshuaP]

To clarify, I would also like to add that another reason that LabVIEW and NI-VISA can't communicate with the GPIB ENET driver is, because the GPIB driver is running as a Rosetta process.

When universal binaries (drivers and applications like NI-VISA 4.0 and LV 8.20) are running natively on intel-based Mac they can not communicate with Rosetta applications or drivers. In addition, applications and drivers running under Rosetta (like the current GPIB drivers) can not communicate with any natively running binaries.

Hopefully this clears things up a little bit.

-Josh

Message Edited by JoshuaP on 10-09-2006 09:43 AM

[sth]

Well, it is unclear on a couple of counts....
Version LV 8.0 works as a rosetta process communicating with the enet-gpib device using traditional GPIB calls. On Mactel.

The enet-gpib is NOT a kernel driver. It uses standard tcp-ip driver calls so should be platform independent, in fact it should not know what platform it is on since every thing is network byte ordered which is platform independent.

Drivers do NOT run under rosetta. The are kernel processes and are way lower level than the rosetta functionality.

The simple question remains, what version VISA did Jason S. use in the compatibility tests?

Can the VISA version 4.0 communicate with a VISA-GPIB Passport that is running in rosetta. This is a userland process not kernel and has a good chance of running under rosetta. I understand the "feature table" which means that it was not tested. The version that Jason shows working is not documented in any NI documentation either. We are trying to get outside the box here to use an over used cliche.

We seem to be back at the same wall we had when there was a switch to OS X. NI has no intention of supporting it but will consider it. VISA does not support every interface that was supported on PPC Macs. And why? Because like serial (which are standard OS calls to the standard OS driver) just like calls to the standard TCP/IP stack is outside the driver for the ENET-GPIB. In fact that is why ENET-Serial was so easy to add as an interface.

I have looked at the communication with the ENET-GPIB. It is just a tcp/ip string, in plain text ascii. I am willing to take a bet that the communication protocol for the USB-GPIB and 1394 GPIB are identical strings as well. It is odd that support for the PCI-GPIB and ENET-GPIB are lumped together when they are so different in details of how to support them.

Would NI be willing to release the protocol for communicating with the ENET-GPIB device such that we could write a driver? How about for the USB-GPIB and/or the 1394-GPIB? If NI isn't going to do it then it should not be a big deal. How about opening up how to write a VISA Passport? It is time for NI to fish or cut bait. And if they won't step up because it is not worth it then they should get out of the way and let someone else do it.

Cheers!

[JasonSmith]

Hi Scott,

I tested with both Intel-native NI-VISA 4.0 and PowerPC native NI-VISA 3.5.  The issue is that an x86 application cannot communicate with a PowerPC application running in Rosetta, so at some point communication will break down between LabVIEW and the GPIB driver.  Depending on the versions used, this could break down between LabVIEW and VISA, or between VISA and NI-488.2.

As for NI-488.2 support on Intel Macs, it is currently an open issue and under consideration. We are evaluating all options available to see what support we can provide and what the time frame is going to be, but nothing concrete has yet been determined.

Jason S.
GPIB & Serial
National Instruments

Message Edited by JasonS on 10-09-2006 05:18 PM