"aligned in memory"

[Ben]

I read this on page 82 of the LV 8 Upgrade notes;

"

In LabVIEW 7.0 and earlier, only 1D arrays are aligned in memory, which

aids performance for array operations. In LabVIEW 7.1 and later, 1D and

2D arrays are aligned in memory. This aids in performance for linear

algebra operations and operations involving the matrix data type.

"

So what does that mean, Long word alligned, quad word alligned.... ?

When would I have to remenber this?

Ben

[rolfk]

---

@Ben wrote:

I read this on page 82 of the LV 8 Upgrade notes;

"

In LabVIEW 7.0 and earlier, only 1D arrays are aligned in memory, which

aids performance for array operations. In LabVIEW 7.1 and later, 1D and

2D arrays are aligned in memory. This aids in performance for linear

algebra operations and operations involving the matrix data type.

"

So what does that mean, Long word alligned, quad word alligned.... ?

When would I have to remenber this?

Ben

---

Hmm, I think that this would only apply to double arrays. I'm pretty sure that the memory allocater does already return at least long aligned pointers, at least I never saw anything else when debugging my DLLs and CINs. So for smaller datatypes any LabVIEW array will always be aligned already. For double arrays (and complex, and 64bit in LabVIEW 8.0) this would be different so I guess LabVIEW now makes sure that the actual data portion of an array will start at the proper alignment of 8 and adjust the header with the dim sizes accordingly, possibly wasting 4 bytes.

Rolf Kalbermatter