I have done a simple matrix vector product with sparse matrix, and it works but the problem is that the first time I executed it takes much more time than the following times, for example, with a dimension of 4993 the first execution takes 500 miliseconds and the following are between 70 and 90 miliseconds!! By the way I measure time with getTimeOfDay() after reading the matrix and vector from a file, as the idea is to compare this version to a cpu version, and in both versions I have to read them from and there is no difference in the way I do it so that time is irrelevant.
One more thing, these results are from a 1201n netbook with a 9400m gpu, but the code was tested in a much better machine and it also had a big difference.
I guess this my be caused by my code but I´m not sure. Any help would be apreciated!!
Thanks in advance!!