Does the hardware you care about not have a reg to read out the current
timestamp? Ours does, and it seems like a way better way to go about it
than hoping that the two clocks don't drift relative to each other.
pgpOwy0UJAI9h.pgp
Description: PGP signature
___
On Wed, 2012-06-27 at 17:16 +0200, Marek Olšák wrote:
> I am working in nanoseconds and precision is quite important. I don't
> think I can use CLOCK_MONOTONIC_COARSE. CLOCK_MONOTONIC might be fine.
My mistake, read the math the wrong way around.
- ajax
signature.asc
Description: This is a dig
On Wed, Jun 27, 2012 at 4:42 PM, Adam Jackson wrote:
> On Wed, 2012-06-27 at 09:29 +0200, Michel Dänzer wrote:
>> On Mit, 2012-06-27 at 03:49 +0200, Marek Olšák wrote:
>> >
>> > +GLuint64
>> > +_mesa_get_cpu_timestamp(void)
>> > +{
>> > +#ifndef _WIN32
>> > + struct timeval tv;
>> > +
>> > + g
On Wed, 2012-06-27 at 09:29 +0200, Michel Dänzer wrote:
> On Mit, 2012-06-27 at 03:49 +0200, Marek Olšák wrote:
> >
> > +GLuint64
> > +_mesa_get_cpu_timestamp(void)
> > +{
> > +#ifndef _WIN32
> > + struct timeval tv;
> > +
> > + gettimeofday(&tv, 0);
> > + return (GLuint64)tv.tv_sec * 1
On 06/26/2012 07:49 PM, Marek Olšák wrote:
---
src/mesa/main/get.c | 10
src/mesa/main/mtypes.h |1 +
src/mesa/main/queryobj.c | 61 ++
src/mesa/main/queryobj.h |5
4 files changed, 77 insertions(+)
diff --git a/s
On Mit, 2012-06-27 at 03:49 +0200, Marek Olšák wrote:
>
> +GLuint64
> +_mesa_get_cpu_timestamp(void)
> +{
> +#ifndef _WIN32
> + struct timeval tv;
> +
> + gettimeofday(&tv, 0);
> + return (GLuint64)tv.tv_sec * 10ull + (GLuint64)tv.tv_usec * 1000;
As in a recent discussion about DRM
---
src/mesa/main/get.c | 10
src/mesa/main/mtypes.h |1 +
src/mesa/main/queryobj.c | 61 ++
src/mesa/main/queryobj.h |5
4 files changed, 77 insertions(+)
diff --git a/src/mesa/main/get.c b/src/mesa/main/get.c
index 677