TBB malloc proxy vs include file

[norswap]

I'm using tbb malloc in a project and I wondered what the difference was between using the tbb malloc proxy library and using a header file to redefine the standard memory allocation functions. Something like:

void* operator new(size_t sz)
{
void *res = scalable_malloc(sz);

if (res == NULL)
throw std::bad_alloc();

return res;
}

void operator delete(void* ptr) throw()
{
scalable_free(ptr);
}

What are the advantages/drawbacks to both approaches ? Is one superior to the other ?

[SergeyKostrov]

Quoting norswap

...What are the advantages/drawbacks to both approaches ?

Please take a look at a thread:

'Difference between scalable_memory and classic memory allocation'

http://software.intel.com/en-us/forums/showthread.php?t=105401&o=a&s=lr

[Vladimir_P_Intel2 · Accepted solution]

Hellonorswap

In your case you will have more control on what should go to scalable allocator and what shouldn't. tbbmalloc proxy redirects _all_ application calls to scallable_malloc. But our implementation of new/delete in proxy is pretty as the same as above one.

tbbmalloc proxy is a great advantage for big projects that you can add one line to one source and get everything working.

--Vladimir