Good way to optimize or replace a nested for loop?

[NilssonCreative]

Hi. A friend of mine wrote this crazy function in VB and i have translated it to C++ in hopes that we could get a bit better performance. I guess the performance might be a bit better but it is still an enourmous amount of loops to go through. Do you have any suggestions on how to optimize this loop or perhaps replace it completely to be able to finish all iterations in a managable time? Is it hard to have it run on the GPU and would that improve anything? // Parre

int CountTheOffsetValids() {

int c0;
int c1;
int c2;
int c3;
int c4;
int c5;
int c6;
int c7;
int c8;
int c9;
int c10;
int c11;
int c12;
int c13;
int c14;
int c15;
int c16;
int c17;
int c18;
int c19;

int llim \[20\] = {1, 2, 5, 8, 11, 14, 17, 21, 24, 28, 31, 35, 38, 42, 45, 50, 53, 58, 61, 66};
int hlim \[20\] = {6, 10, 14, 19, 22, 26, 29, 34, 37, 41, 44, 48, 51, 54, 57, 61, 63, 67, 69, 70};


for (c0 = llim\[0\]; c0 <= hlim\[0\]; ++c0) {
        arrVals\[0\] = c0;
		for (c1 = llim\[0\]; c1 <= hlim\[1\]; ++c1) {
            arrVals\[1\] = c1;
            if (c1 > c0) {
                for (c2 = llim\[2\]; c2 <= hlim\[2\]; ++c2) {
                    arrVals\[2\] = c2;
                    if (c2 > c1) {
                        for (c3 = llim\[3\]; c3 <= hlim\[3\]; ++c3) {
                            arrVals\[3\] = c3;
                            if (c3 > c2) {
                                for (c4 = llim\[4\]; c4 <= hlim\[4\]; ++c4) {
                                    arrVals\[4\] = c4;
                                    if (c4 > c3) {
                                        for (c5 = llim\[5\]; c5 <= hlim\[5\]; ++c5) {
                                            arrVals\[5\] = c5;
                                            if (c5 > c4) {
                                                for (c6 = llim\[6\]; c6 <= hlim\[6\]; ++c6) {
                                                    arrVals\[6\] = c6;
                                                    if (c6 > c5) {
                                                        for (c7 = llim\[7\]; c7 <= hlim\[7\]; c7) {
                                                            arrVals\[7\] = c7;
                                                            if (c7 > c6) {
                                                                for (c8 = llim\[8\]; c8 &lt

[_Superman_]

The C++ standard template library (STL) comes with a lot of algorithms optimized for performance. So if you can tell me what the above code is trying to do, maybe I could suggest an STL algorithm that does the same. Not sure if the GPU will be able to do this, because I see a lot of data dependencies between the loop variables. AFAIK, GPU programming needs the loop variables to be independent of each other.

«_Superman_»  _I love work. It gives me something to do between weekends.

_Microsoft MVP (Visual C++) (October 2009 - September 2013)

Polymorphism in C

[NilssonCreative]

Hi and thanks for the answer. The code above is trying every combination of values 1-70 and is supposed to output them to an array of 20 values. Example output each iteration: arrVals[] = {3, 6, 12, 15, 22, 26, 29, 34, 37, 41, 44, 48, 51, 54, 57, 61, 63, 67, 69, 70}; Each position in the array has limitations to how big or small they can be as follows: int llim [20] = {1, 2, 5, 8, 11, 14, 17, 21, 24, 28, 31, 35, 38, 42, 45, 50, 53, 58, 61, 66}; int hlim [20] = {6, 10, 14, 19, 22, 26, 29, 34, 37, 41, 44, 48, 51, 54, 57, 61, 63, 67, 69, 70}; The array arrVals is sent to another function each iteration to decide if the current row of 20 is usable or not. I hope this is clear enough, please ask if not.

[_Superman_]

Not exactly sure what the condition logic is. But I believe you could use the std::copy_if[^] alogirthm. Other algorithms that could be possible candidates are - std::next_permutation[^] std::generate[^] std::transform[^] std::for_each[^]

«_Superman_»  _I love work. It gives me something to do between weekends.

_Microsoft MVP (Visual C++) (October 2009 - September 2013)

Polymorphism in C

[NilssonCreative]

I think the std::next_permutation would be perfect tough i need conditions for it so that it doesn't try to find every permutation for all 70^70, which would take forever. I'm actually not sure if it will be fast enough even with conditions. I'm obviously not good enough with C++ to manage this :p