🚀 go-pugleaf

Might be interesting to add

GlobalFree( GlobalAlloc( GMEM_FIXED, 0x2000 ) );
GlobalFree( GlobalAlloc( GMEM_FIXED|GMEM_ZEROINIT, 0x2000 ) );
GlobalFree( GlobalAlloc( GMEM_MOVEABLE, 0x2000 ) );
GlobalFree( GlobalAlloc( GMEM_MOVEABLE|GMEM_ZEROINIT, 0x2000 ) );

 to your tests.
And then, you could add the GMEM_NOCOMPACT flag to the above to see if
they go any faster.

But, if you want to get closer to the speed of malloc, then this sort
of thing should help:

(Allocates 81,920,000 bytes)
char*array[10000];
void*memory[10000/16];
for( int i=0; i<10000/16; ++i )
{	memory[i] = GlobalAlloc( GMEM_FIXED,16*0x2000);
	for( int j=0; j<16; ++j )
		array[i*16+j] = (char*)memory[i];
};

As i understand it, this is what malloc does, only bigger and more
complicated.

You might rewrite your test for a more real-world experience.
Inside a 2,000 timing loop,
{	/* Maximum garbage generation */
   {Allocate 1000 random size allocations}
   {Randomly free the allocations}
}
Random Free might be simple, like
	for( int i=0; i<13; ++i )
		for( int k=i; k<1000; k+=13 )
			free( array[k] );