On Wed, 2011-06-29 at 13:50 -0400, Dominic Lachowicz wrote:
> LARGE_INTEGER is required by that Win32 API. I'd argue for using that
> in platform-specific code.
Ah sorry about that. Then only 1) would be nice to have.
Marc
> On Wed, Jun 29, 2011 at 1:46 PM, J.M. Maurer <uwog@uwog.net> wrote:
> >
> > Just two small remarks about this commit:
> >
> > 1. Could you wrap all timing related stuff into an if #DEBUG block? That
> > will disable it for release builds.
> >
> > 2. We tend to use our own integer types, so we are consistent across all
> > platforms. So instead of LARGE_INTEGER, we'd use UT_sint64 .
> >
> > Happy hacking!
> > Marc
> >
> > On Tue, 2011-06-28 at 10:45 +0200, cvs@abisource.com wrote:
> >> + LARGE_INTEGER t1, t2, freq;
> >> +
> >> + QueryPerformanceCounter(&t1);
> >> +
> >> // copy the screen to the buffer
> >> BitBlt(m_bufferHdc, 0, 0, width, height, m_hdc, 0, 0,
> >> SRCCOPY);
> >> +
> >> + QueryPerformanceCounter(&t2);
> >> +
> >> + QueryPerformanceFrequency(&freq);
> >> + double blitSpeed = ((double)(t2.QuadPart - t1.QuadPart)) /
> >> ((double)freq.QuadPart);
> >>
> >> + UT_DEBUGMSG(("ASFRENT: measured BitBlt speed: %lfs [client
> >> rectangle W = %d, H = %d]\n",
> >> + blitSpeed, width, height));
> >> +
> >
> >
>
>
>
Received on Wed Jun 29 20:57:27 2011
This archive was generated by hypermail 2.1.8 : Wed Jun 29 2011 - 20:57:27 CEST