Re^7: Rosetta Code: Long List is Long (faster - vec

Very interesting! This will help me get started with OpenMP, which I've been meaning to do for a while now.

Unfortunately, strtok is not thread-safe e.g. strtok(NULL, "\n") causing segfault. So I factored out strtok.

Ha ha, that I used strtok at all is an indication of how desperate I was, given I singled this function out for a special dishonourable mention at On Interfaces and APIs. :)

To round out this section, notice that the ANSI C strtok function has a shocker of an interface: it surprises by writing NULLs into the input string being tokenized; the first call has different semantics to subsequent calls (distinguished by special-case logic on the value of its first parameter); and it stores state between calls so that only one sequence of calls can be active at a time (bad enough in a single-threaded environment, but so intolerable in multi-threaded and signal handler (reentrant) environments that the POSIX threads committee invented a replacement strtok_r function).

Comment on Re^7: Rosetta Code: Long List is Long (faster - vec - OpenMP) Select or Download Code

Replies are listed 'Best First'.
Re^8: Rosetta Code: Long List is Long (faster - vec - OpenMP) by marioroy (Prior) on Jan 11, 2023 at 22:07 UTC
It was weird experiencing the segfault due to strtok(NULL, "\n"). Great! What is strtok_r() function in C language? I compared strtok_r with context against strchr. After testing, I updated the OpenMP demonstration to find the tab character within the string. Now passing: limited length and variable length words, including OpenMP. Update 2 enables parallel sort. `# 15 input files, 1 thread, clang++ -std=c++20 strtok_r 9.655 seconds strchr 8.713 seconds` [download] `while ( ::fgets(line, MAX_LINE_LEN_L, fh) != NULL ) { found = ::strchr(line, '\t'); count = ::atoll( &line[found - line + 1] ); line[found - line] = '\0'; // word ... }` [download] Also, I replaced strlen(...) with found - line. `::memcpy( fixword.data(), line, found - line );` [download]	[reply] [d/l] [select]

Replies are listed 'Best First'.

Re^8: Rosetta Code: Long List is Long (faster - vec - OpenMP)
by marioroy (Prior) on Jan 11, 2023 at 22:07 UTC

It was weird experiencing the segfault due to strtok(NULL, "\n").

Great! What is strtok_r() function in C language? I compared strtok_r with context against strchr. After testing, I updated the OpenMP demonstration to find the tab character within the string. Now passing: limited length and variable length words, including OpenMP. Update 2 enables parallel sort.

# 15 input files, 1 thread, clang++ -std=c++20
strtok_r  9.655 seconds
strchr    8.713 seconds
[download]

while ( ::fgets(line, MAX_LINE_LEN_L, fh) != NULL ) {
   found = ::strchr(line, '\t');
   count = ::atoll( &line[found - line + 1] );
   line[found - line] = '\0';  // word
   ...
}
[download]

Also, I replaced strlen(...) with found - line.

::memcpy( fixword.data(), line, found - line );
[download]

[reply]
[d/l]
[select]