On 08/17/2010 10:09 AM, Tim Bird wrote: > Well, page faults are always inherently slower than reads,
It's embarrassing, but I have to correct myself here. An mmap and an access of a single page will likely take longer than a corresponding read. But accessing multiple pages can amortize the cost of setting up the page tables for the mmap, and avoids a data copy for each read, which the read syscall incurs. Whether mmap()/page-fault or read() is faster depends on a number of factors, but mmap often wins. -- Tim ============================= Tim Bird Architecture Group Chair, CE Linux Forum Senior Staff Engineer, Sony Network Entertainment ============================= -- unsubscribe: android-porting+unsubscr...@googlegroups.com website: http://groups.google.com/group/android-porting
