On 23 March 2011 23:50, Adam Martinson wrote:
> On 03/22/2011 01:33 PM, Henri Verbeet wrote:
>>
>> How much of a real difference does this patch make, after you
>> avoid the redundant lookups? (I.e., after the second patch in this
>> series.)
>
> In that case the additional reduction in context_ap
On 03/22/2011 01:33 PM, Henri Verbeet wrote:
How much of a real difference does this patch make, after you
avoid the redundant lookups? (I.e., after the second patch in this
series.)
In that case the additional reduction in context_apply_fbo_state() CPU
time from the 1st patch is only about 10%.
On 03/22/2011 05:02 AM, Henri Verbeet wrote:
On 21 March 2011 21:56, Adam Martinson wrote:
Cuts CPU time in context_apply_fbo_state() in half.
This is meaningless. Which applications, and how much time of the
total is spent in context_apply_fbo_state()?
In the 3DMark06 batch size 8 test, as of
On 22 March 2011 18:12, Adam Martinson wrote:
> On 03/22/2011 05:02 AM, Henri Verbeet wrote:
>> This is meaningless. Which applications, and how much time of the
>> total is spent in context_apply_fbo_state()?
>
> In the 3DMark06 batch size 8 test, as of Friday's tip, it was 5.5% of
> wined3d; 1.2
On 21 March 2011 21:56, Adam Martinson wrote:
>
> Cuts CPU time in context_apply_fbo_state() in half.
This is meaningless. Which applications, and how much time of the
total is spent in context_apply_fbo_state()? How does this translate
into concrete thing like frame time? More importantly, why do
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Am 21.03.2011 um 21:56 schrieb Adam Martinson:
>
> Cuts CPU time in context_apply_fbo_state() in half.
What is the performance impact on real apps?
-BEGIN PGP SIGNATURE-
Version: GnuPG/MacGPG2 v2.0.17 (Darwin)
iQIcBAEBAgAGBQJNiHM5AAoJEN0/
Adam Martinson wrote:
> Cuts CPU time in context_apply_fbo_state() in half.
> ---
> dlls/wined3d/context.c | 65
> +--
> dlls/wined3d/wined3d_private.h |1 +
> 2 files changed, 36 insertions(+), 30 deletions(-)
It looks like you are not subs