On Nov 7, 2007 7:09 AM, Robin Garner <[EMAIL PROTECTED]> wrote: > > Aleksey Shipilev wrote: > > On 07 Nov 2007 00:27:25 +0300, Egor Pasko <[EMAIL PROTECTED]> wrote: > > > >> Vladimir, guess what? :) I actually mixed several things > >> altogether. > >> > > <giggly>(sigh) As usual, miracle does not happen :) I dreamed to see > > software sqrt() implementation that could be faster than hardware one. > > </giggly> > > > > > >> So, we are left with SSE asm that can be inlined by JIT and AFAI can > >> see it is not as fast as HotSpot? Weird :) > >> > > No, for now we have just the intrinsic in native code, so we also have > > overheads for JNI transition (and it is heavy!), parameter passing, > > chains of calls, etc. I believe that sqrt() magic will lead that NBody > > performance very close to RI. > > > > BTW, in my thought this benchmark is like the top of the iceberg > > called "FP performance problems". > > > > Thanks, > > Aleksey. > > > You might be interested to know that the JikesRVM implementation calls > out to glibc via a fast syscall (we have an @SysCall annotation for > native methods), and it is 1.5x slower than Sun 1.6 on a Core2/Linux for > Tim's (non-strict) benchmark. Maybe you can squeeze the overhead of > intrinsic calls down a bit further ?
I am also wondering why not to leverage the system support directly. This solution is by default portable. :-) > cheers > > cheers > -- http://xiao-feng.blogspot.com
