timespec | Functions for working with timespec structures | Genomics library

You are looking for a function, which has O(log n) performance. I can tell you that finding the maximum won't be a good example, because of those two reasons:

• Either your list is not sorted, so you'll need to investigate all items in your list => the performance is at least O(n).
• Either your list is sorted, then you just take the last value => the performance is O(1).

Is there anything you can do with a performance O(log n)? Yes, there is, let me give you the following example: you have a sorted list and you want to insert a new item, making sure that the list stays sorted. You can then use a binary search algoritm for finding the place where you need to insert that new item.

Python cannot run more than one CPU-intensive thread. The cause of it the the GIL. Basically Python threads are not good for anything but waiting for I/O.

If you want a CPU-intensive part, try either rewriting the intensive part using Cython, or use multiprocessing, but then the time to send data back and forth could be significant.

Is this what your trying to achieve?

I tried logging events using your code. the events are logged in the local timezone, not the specified one. I think the problem is with this code

futimens requires an array of two struct timespec. That's why you get an invalid argument error. The zeroth element is the access time, and the first is the modification time. So you could set the modification time like this:

Reduce the floating point math.

OP's code is doing excessive FP math in what should be a scale and lookup.

Scale the radians per a pre-computed factor into an index.

The number of entries in the lookup table should be an unsigned power-of-2 so the mod is a simple &.

At first, let us simplify and have [0 ... 2*pi) map to indexes [0 ... number_of_entries) to demo the idea.

SIGSEGV means SIGnal SEGmentation Violation, meaning that the code accesses memory areas that it is not allowed to access, in this case it smashes the stack ('Address boundary error'), see Nate Eldredge's and kaylum's comments under the question. The program is terminated by this signal (SIGSEGV) from the operating system. When using dynamic allocation the memory for the arrays is allocated on the heap (What and where are the stack and heap?) and so you will not have this issue (malloc(), calloc(),... : https://www.programiz.com/c-programming/c-dynamic-memory-allocation )

Continuing from my comment above. Any time you read text input in C (as opposed to binary input from a file), you are reading characters not numeric values. When the user enters a number, it is a string of digits, not a numeric value. It is up to you, the programmer, to perform the conversion from the string of digits to the numeric value using either the strtol() family of functions which provides full error detection, or at minimum, use sscanf() to validate the conversion from a succeed/fail standpoint.

In your case, you can simply use strtoul() to provide the greatest range for input values. You simply convert argv[1] and argv[2] to numeric values using strtoul(). A minimum validation would be, e.g.

I am trying to run the producer-consumer problem while using pthread_cond_signal() instead of pthread_cond_broadcast(), however, I attempted a lot of things and can't seem to do it (if I choose n producers and one consumer then the consumer finishes but not all producers finish, some are stuck on full queue so the problem never finishes executing.

Well that sounds eminently plausible. If you have multiple threads blocked on a CV, then one signal will wake one of them. The rest will remain blocked.

I am generally inclined to go the other way. If you use your CV correctly, then it is safe to always broadcast to it instead of signaling it, but doing the opposite exposes more area for possible bugs, especially when more than two threads are involved.

For the shutdown scenario in particular, I would recommend just using a broadcast. You need to wake potentially multiple threads, and that's exactly what pthread_cond_broadcast() is for. You could have the main thread do that instead of either consumer or producer if you wish. But if you insist on using only pthread_cond_signal() then you must be sure to call that function enough times to wake all threads that may be blocked on the CV. Again, some or all of those calls could be performed by the main thread.

Notwithstanding my above recommendation to broadcast, a relatively good way to obtain clean shutdown with signaling only would be for each producer to signal the notFull CV before terminating. There are a couple of places you could put that, but I would probably do this, myself: