reading 64-bit integers from file

[dlbuhl]

I have a binary file that contains 64-bit integers (I guess what LV would consider U64 or I64) that I need to read into labview 7.1. I have read a great deal in the forums stating that there is no such way to do this in previous versions of Labview - messages from over two years ago. I hope that NI has finally programmed a way to deal with 64-bit integers considering the amount of discussion for them. Is there an easy way that I am missing to do this? I cannot read it as a double (although it is the same size).

[GerdW]

Hello dlbuhl,

I think LabView doesn't have an U64 or I64 datatype. You will have to program your own...
And reading should be easy: read in as U32 and combine the MSB and LSB to your datatype.

[Bob_Y.]

This link
also has some good ideas for accessing 64 bit data from a DLL. Much of the information would also apply to your situation. The reading of the data is simply to do it as GerdW suggests and using a U32 twice. The rest of handling it in LabVIEW could use the suggestions in this link.

Hope that this helps,
Bob

[dlbuhl]

this may seem a niave question... but how does one obtain the MSB and LSB of a number so they may be combined?

thanks, i will look into the other suggestions as well.

[Sheldon_Stokes]

First, you are talking about integers, so your problem is a lot easier; floats need a bit more work to combine (mantissa and exponent).

Your 64 bit integer is stored as 8 bytes, in either of the two following ways:

| MSB | | | | | | | LSB |

Or

| LSB | | | | | | | MSB |

This is dependant on what code wrote the data and whether the machine is big endian or little endian (insert religious war here).

Reading a 64 bit number into two 32 bit numbers results in the left half of the above bytes being in one location and the right half being in a second location. Here’s an example which will illustrate how to deal with this issue:

Assume that we have a two digit (base 10) number: 87
Lets say that I can only store one digit in a location, so I read in the 87 and store the 8 in variable “A” and the 7 in variable “B”.

A = 8
B = 7

To get the number back, you would have to multiply A by 10 (because this digit came from the 10’s column) and add it to “B”.

87 = A*10 + B

Your 64 bit problem is the same. Where “A” is the most significant bytes, and “B” is the least significant bytes. So:

64bit_value = A* (2^32) + B


Now, I’ve swept a bit under the rug, which is that you may have to byte swap to convert from big edian to little edian or visa versa.

Sheldon

[DFGray]

As the link Bob pointed out says, you really do want to read the 64-bit integer as a double. This method will work for arrays and structures (clusters) as well, since platform-specific problems such as byte alignment will automatically be done for you. When you get it into LabVIEW, use the Type Cast primitive to change it to a cluster of two 32-bit integers.

If you want a good example of this, check out Can I Edit and Create Hierarchical Data Format (HDF5) files in LabVIEW?. The VIs in the examples use this method extensively to interface to the 64-bit HDF5 DLL. There is also quite a bit of 64-bit infrastructure code you can probably use.

[dlbuhl]

So I have tried all of your suggestions and to no avail have failed to obtain the correct number. I have attached a zip file of one of my recordings. The data format is as follows:

16384 byte ascii header, followed by a continuous loop of the following:
int64 timestamp
int32 channel number
int32 sampling freq
int32 number of valid samples
int16 array of 512 elements

I need to extract the int64 timestamp value - it should begin at 16384 bytes within the file, and occur every 8212 bytes. If any of you have a few minutes to devote to showing me how I can obtain these values, I will be forever in your debt! (and it may help other users trying to do something similar as well!).

Thanks!

[GerdW]

Hello dlbuhl,

try this.
I changed the offset in the file to 1036 (512*int16 + 3*int32 is 1036 bytes to me...).
The vi gets all timestamps as 2 U32 and displays them in hexadecimal format.

[dlbuhl]

Dear GerdW,

Thanks for the vi, but I still have a slight problem. The first number that I get from my c-script, which I know is the correct number, is 196,235,189. How can I get this number from the Hex number that your vi outputs? Sorry, I'm a bit a novice on this topic.

Best,
Derek

[dlbuhl]

Oh yes, and of course you are correct on the 1036... I still need my morning coffee 😉