On 09/20/2012 12:55 PM, Chad Versace wrote:
> This patch reduces the time spent in glTexImage and glTexSubImage by
> over 5x on Sandybridge for the workload described below.
>
> It adds a new fast path for glTexImage2D and glTexSubImage2D,
> intel_texsubimage_tiled_memcpy, which is optimized for G
Chad Versace writes:
> This patch reduces the time spent in glTexImage and glTexSubImage by
> over 5x on Sandybridge for the workload described below.
>
> It adds a new fast path for glTexImage2D and glTexSubImage2D,
> intel_texsubimage_tiled_memcpy, which is optimized for Google Chrome's
> paint
This patch reduces the time spent in glTexImage and glTexSubImage by
over 5x on Sandybridge for the workload described below.
It adds a new fast path for glTexImage2D and glTexSubImage2D,
intel_texsubimage_tiled_memcpy, which is optimized for Google Chrome's
paint rectangles. The fast path is impl
On 09/14/2012 07:01 PM, Paul Berry wrote:
> On 11 September 2012 12:04, Chad Versace wrote:
>> + if (!image->mt ||
>> + image->mt->region->tiling != I915_TILING_X) {
>> + /* The algorithm below is written only for X-tiled memory.
>> + *
>> + * Memcpy'ing to Y-tiled memory
On 11 September 2012 12:04, Chad Versace wrote:
> This patch adds a new fast path for glTexImage2D and glTexSubImage2D,
> intel_texsubimage_tiled_memcpy, which is optimized for Google Chrome's
> paint rectangles. The fast path is implemented only for 2D GL_BGRA
> textures on gen >= 6.
>
> Reduces
This patch adds a new fast path for glTexImage2D and glTexSubImage2D,
intel_texsubimage_tiled_memcpy, which is optimized for Google Chrome's
paint rectangles. The fast path is implemented only for 2D GL_BGRA
textures on gen >= 6.
Reduces the time spent in glTexImage and glTexSubImage by roughly 3x
On Thu, Sep 13, 2012 at 04:22:20PM +0300, Chad Versace wrote:
> On 09/11/2012 10:40 PM, Daniel Vetter wrote:
> > Only quick read-through but I'd have expected a has_llc check in there
> > - if vlv is anything like the previous platforms wc gtt will be much
> > faster there.
>
> I'm not too familia
On 09/11/2012 10:40 PM, Daniel Vetter wrote:
> Only quick read-through but I'd have expected a has_llc check in there
> - if vlv is anything like the previous platforms wc gtt will be much
> faster there.
I'm not too familiar with why one would want to check has_llc here. I see that
has_llc is che
On 09/11/2012 10:29 PM, Matt Turner wrote:
> On Tue, Sep 11, 2012 at 12:04 PM, Chad Versace
> wrote:
>> + * This is a performance win over the conventional texture upload path
>> because
>> + * it avoids the performance penalty of writing throuh the write-combine
>
> "through"
>
> Very cool wor
Only quick read-through but I'd have expected a has_llc check in there
- if vlv is anything like the previous platforms wc gtt will be much
faster there.
-Daniel
On Tue, Sep 11, 2012 at 9:04 PM, Chad Versace
wrote:
> This patch adds a new fast path for glTexImage2D and glTexSubImage2D,
> intel_te
On Tue, Sep 11, 2012 at 12:04 PM, Chad Versace
wrote:
> + * This is a performance win over the conventional texture upload path
> because
> + * it avoids the performance penalty of writing throuh the write-combine
"through"
Very cool work.
___
mesa-de
This patch adds a new fast path for glTexImage2D and glTexSubImage2D,
intel_texsubimage_tiled_memcpy, which is optimized for Google Chrome's
paint rectangles. The fast path is implemented only for 2D GL_BGRA
textures on gen >= 6.
Reduces the time spent in glTexImage and glTexSubImage by roughly 3x
12 matches
Mail list logo
