Flipping the array

[rpr]

---

Since a 100x100 array  has 10000 elements, how does placing the next element at position 1100 make any sense?

---

I meant 1 comma 100 and not 1100.Sorry abt the confusion

[rpr]

JoelabVIEW u r right.A picture is worth more than 100x100 words.Please find the attachment

[GerdW]

Hi rpr,

have you tried corys example from message 6? You need just one for loop less - it's your turn to find out which one to omit

[altenbach]

---

rpr wrote:
JoelabVIEW u r right.A picture is worth more than 100x100 words.Please find the attachment

---

Here are two simple possibilities.

Message Edited by altenbach on 07-15-2008 12:35 AM

[rpr]

Thanx a ton altenbach.Just a question,I am using this to flip aan image.The image dimension is 2000x1500.Is this the best possible way altenbach.Will revert back with the exact time taken.

Thanx for the help guys

PS: I suppose altenbach wouldn't mind me using the icon he designed


Message Edited by rpr on 07-15-2008 02:44 AM

[altenbach]

No problem with the icon. 🙂

Here are 4 more versions. I would code up all version and run a quick benchmark with typical data. I would guess the versions that operate "in place" are best.

Message Edited by altenbach on 07-15-2008 01:03 AM

[rpr]

Thanx altenbach.The time taken is 3 seconds approx.

[altenbach]

OK, here are some benchmarks.

Using the numbering of the output indicator, you want to go with #5 or #6. They are about equivalent.

Version #1 is > 2x slower. The two transpositions are not exactly free.

Version #2 is > 3x slower. Operating on columns is inefficient because the elements are not adjacent in memory.

Version #3 is ~1.5x slower. Pretty efficient but twice as many iterations as #5/#6.

Version #4 builds an array in a loop and is thus several orders of magnitude slower (>1000x for 1500x2000!). Definitely not recommended.

Here's the final version of #5, which is what I probably would use. Note that the earlier image has a slight bug, so use the following:

Please verify correct operation. This is not fully tested. 😄

Message Edited by altenbach on 07-15-2008 02:07 AM

[altenbach]

OK, here's a quick benchmark version comparing some of the algorithms. (LV 8.5.1)

(The numberings is a bit different and I left out the "built array" version. It is too slow. :))

[rpr]

What else can i ask for thanx a lot altenbach