[QST] firfilter performance on Jetson

[gpeyton]

I'm running benchmarking tests on a Jetson Nano 4GB board, and am testing a number of filter algorithms. The performance of each (for N number of taps) is shown below.

firfilter is tested as follows:

    b_cp = cp.asarray(b)
    x_cp = cp.asarray(x)
    
    # warm up
    firfilter_result = firfilter(b_cp, x_cp, axis=0) 

    start = time.time()
    for i in range(20):
        firfilter_result = firfilter(b_cp, x_cp, axis=0)   # x is a 2D array
    stop = time.time()
    tot_time = (stop - start) / 20
    csig_time.append(tot_time)

In contrast, running the same code on a GeForce GTX 1050 Ti yields the following:

In this case, firfilter (GPU) performance exceeds CPU performance for the other algorithms, which is expected.

However, we do not expectfirfilter to be significantly slower than the other algorithms on Jetson . What potential performance optimizations could help solve this problem?

[github-actions]

This issue has been labeled inactive-90d due to no recent activity in the past 90 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed.

[github-actions]

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.

[guillaumedavidphd]

This issue still needs to be investigated.

[github-actions]

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.

[github-actions]

This issue has been labeled inactive-90d due to no recent activity in the past 90 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed.