Ok, but then USEMALLOC is at least a workaround to get program compiled.
As I understand you right there is not a big performance difference in using either the old MT Core or the new MT Core with USEMALLOC.
Or is the performance then even worse?
Matthias
I have not measured. But the new allocator is much better with MT (it has much reduced mutex locking, using that TLS thing).
Frankly, using MingW to do any serious stuff does not seem as a good idea anymore....