Re: [Mesa-dev] [PATCH v2] i915/aa: fixing anti-aliasing bug for thinnest width lines

[Predut, Marius]

> -Original Message-
> From: mesa-dev [mailto:mesa-dev-boun...@lists.freedesktop.org] On Behalf Of
> Ian Romanick
> Sent: Wednesday, September 09, 2015 9:11 PM
> To: Predut, Marius; mesa-dev@lists.freedesktop.org
> Subject: Re: [Mesa-dev] [PATCH v2] i915/aa: fixing anti-aliasing bug for
> thinnest width lines
> 
> On 09/09/2015 06:59 AM, Marius Predut wrote:
> > On PNV platform, for 1 pixel line thickness or less,
> > the general anti-aliasing algorithm gives up, and a garbage line is
> generated.
> > Setting a Line Width of 0.0 specifies the rasterization
> > of the "thinnest" (one-pixel-wide), non-antialiased lines.
> > Lines rendered with zero Line Width are rasterized using
> > Grid Intersection Quantization rules as specified by bspec G45: Volume 2:
> 3D/Media,
> > 7.3.13.1 Zero-Width (Cosmetic) Line Rasterization section.
> >
> > This patch follow the same rules as patches fixing the
> > https://bugs.freedesktop.org/show_bug.cgi?id=28832
> > bug.
> >
> > v1: Eduardo Lima Mitev:  Wrong indentation inside the if clause.
> >
> > Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90367
> >
> > Signed-off-by: Marius Predut <marius.pre...@intel.com>
> > ---
> >  src/mesa/drivers/dri/i915/i915_state.c | 15 +++
> >  1 file changed, 15 insertions(+)
> >
> > diff --git a/src/mesa/drivers/dri/i915/i915_state.c
> b/src/mesa/drivers/dri/i915/i915_state.c
> > index 4c83073..ebb4e9a 100644
> > --- a/src/mesa/drivers/dri/i915/i915_state.c
> > +++ b/src/mesa/drivers/dri/i915/i915_state.c
> > @@ -599,6 +599,21 @@ i915LineWidth(struct gl_context * ctx, GLfloat widthf)
> >
> > width = (int) (widthf * 2);
> > width = CLAMP(width, 1, 0xf);
> > +
> > +   if (ctx->Line.Width < 1.5 || widthf < 1.5) {
> > + /* For 1 pixel line thickness or less, the general
> > +  * anti-aliasing algorithm gives up, and a garbage line is
> > +  * generated.  Setting a Line Width of 0.0 specifies the
> > +  * rasterization of the "thinnest" (one-pixel-wide),
> > +  * non-antialiased lines.
> > +  *
> > +  * Lines rendered with zero Line Width are rasterized using
> > +  * Grid Intersection Quantization rules as specified by
> > +  * bspec G45: Volume 2: 3D/Media,
> 
> Eh... Isn't G45 actually i965-class hardware?  GEN4.5, right?  I don't
> see how a GEN4 bspec reference is useful (or correct?) in the GEN3
> driver.  I think you want "2.8.4.1 Zero-Width (Cosmetic) Line
> Rasterization" from volume 1f of the GEN3 docs.

I used this reference specs because I was unable to find dedicated specs for 
gen 3.
Send v2 patch.

> 
> I looked at section 2.8.4.3 Anti-aliased Line Rasterization of volume 1f
> of the GEN3 docs that I have.  I don't see any mention of this
> restriction.  Is it documented anywhere?

my knowledge says no 

> 
> It also seems like this will affect non-antialised wide lines.  Is that
> correct?

Yes, for Line.Width < 1.5,

But seems this is the normal behavior.(for width = 0 , antialiased is disabled)


> 
> > +  * 7.3.13.1 Zero-Width (Cosmetic) Line Rasterization section
> > +  */
> > +  width = 0;
> > +   }
> > lis4 |= width << S4_LINE_WIDTH_SHIFT;
> >
> > if (lis4 != i915->state.Ctx[I915_CTXREG_LIS4]) {
> 
> ___
> mesa-dev mailing list
> mesa-dev@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/mesa-dev
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH v2] i915/aa: fixing anti-aliasing bug for thinnest width lines

[Brian Paul]

On 09/09/2015 07:59 AM, Marius Predut wrote:

On PNV platform, for 1 pixel line thickness or less,
the general anti-aliasing algorithm gives up, and a garbage line is generated.
Setting a Line Width of 0.0 specifies the rasterization
of the "thinnest" (one-pixel-wide), non-antialiased lines.
Lines rendered with zero Line Width are rasterized using
Grid Intersection Quantization rules as specified by bspec G45: Volume 2: 
3D/Media,
7.3.13.1 Zero-Width (Cosmetic) Line Rasterization section.

This patch follow the same rules as patches fixing the
https://bugs.freedesktop.org/show_bug.cgi?id=28832
bug.

v1: Eduardo Lima Mitev:  Wrong indentation inside the if clause.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90367

Signed-off-by: Marius Predut 
---
  src/mesa/drivers/dri/i915/i915_state.c | 15 +++
  1 file changed, 15 insertions(+)

diff --git a/src/mesa/drivers/dri/i915/i915_state.c 
b/src/mesa/drivers/dri/i915/i915_state.c
index 4c83073..ebb4e9a 100644
--- a/src/mesa/drivers/dri/i915/i915_state.c
+++ b/src/mesa/drivers/dri/i915/i915_state.c
@@ -599,6 +599,21 @@ i915LineWidth(struct gl_context * ctx, GLfloat widthf)

 width = (int) (widthf * 2);
 width = CLAMP(width, 1, 0xf);
+
+   if (ctx->Line.Width < 1.5 || widthf < 1.5) {


By using "1.5f" we can avoid some float/double conversion.



+ /* For 1 pixel line thickness or less, the general
+  * anti-aliasing algorithm gives up, and a garbage line is
+  * generated.  Setting a Line Width of 0.0 specifies the
+  * rasterization of the "thinnest" (one-pixel-wide),
+  * non-antialiased lines.
+  *
+  * Lines rendered with zero Line Width are rasterized using
+  * Grid Intersection Quantization rules as specified by
+  * bspec G45: Volume 2: 3D/Media,


bspec?


+  * 7.3.13.1 Zero-Width (Cosmetic) Line Rasterization section
+  */
+  width = 0;
+   }
 lis4 |= width << S4_LINE_WIDTH_SHIFT;

 if (lis4 != i915->state.Ctx[I915_CTXREG_LIS4]) {



___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH v2] i915/aa: fixing anti-aliasing bug for thinnest width lines

[Marius Predut]

On PNV platform, for 1 pixel line thickness or less,
the general anti-aliasing algorithm gives up, and a garbage line is generated.
Setting a Line Width of 0.0 specifies the rasterization
of the "thinnest" (one-pixel-wide), non-antialiased lines.
Lines rendered with zero Line Width are rasterized using
Grid Intersection Quantization rules as specified by bspec G45: Volume 2: 
3D/Media,
7.3.13.1 Zero-Width (Cosmetic) Line Rasterization section.

This patch follow the same rules as patches fixing the
https://bugs.freedesktop.org/show_bug.cgi?id=28832
bug.

v1: Eduardo Lima Mitev:  Wrong indentation inside the if clause.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90367

Signed-off-by: Marius Predut 
---
 src/mesa/drivers/dri/i915/i915_state.c | 15 +++
 1 file changed, 15 insertions(+)

diff --git a/src/mesa/drivers/dri/i915/i915_state.c 
b/src/mesa/drivers/dri/i915/i915_state.c
index 4c83073..ebb4e9a 100644
--- a/src/mesa/drivers/dri/i915/i915_state.c
+++ b/src/mesa/drivers/dri/i915/i915_state.c
@@ -599,6 +599,21 @@ i915LineWidth(struct gl_context * ctx, GLfloat widthf)

width = (int) (widthf * 2);
width = CLAMP(width, 1, 0xf);
+
+   if (ctx->Line.Width < 1.5 || widthf < 1.5) {
+ /* For 1 pixel line thickness or less, the general
+  * anti-aliasing algorithm gives up, and a garbage line is
+  * generated.  Setting a Line Width of 0.0 specifies the
+  * rasterization of the "thinnest" (one-pixel-wide),
+  * non-antialiased lines.
+  *
+  * Lines rendered with zero Line Width are rasterized using
+  * Grid Intersection Quantization rules as specified by
+  * bspec G45: Volume 2: 3D/Media,
+  * 7.3.13.1 Zero-Width (Cosmetic) Line Rasterization section
+  */
+  width = 0;
+   }
lis4 |= width << S4_LINE_WIDTH_SHIFT;
 
if (lis4 != i915->state.Ctx[I915_CTXREG_LIS4]) {
-- 
1.9.1

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH v2] i915/aa: fixing anti-aliasing bug for thinnest width lines

[Ian Romanick]

On 09/09/2015 06:59 AM, Marius Predut wrote:
> On PNV platform, for 1 pixel line thickness or less,
> the general anti-aliasing algorithm gives up, and a garbage line is generated.
> Setting a Line Width of 0.0 specifies the rasterization
> of the "thinnest" (one-pixel-wide), non-antialiased lines.
> Lines rendered with zero Line Width are rasterized using
> Grid Intersection Quantization rules as specified by bspec G45: Volume 2: 
> 3D/Media,
> 7.3.13.1 Zero-Width (Cosmetic) Line Rasterization section.
> 
> This patch follow the same rules as patches fixing the
> https://bugs.freedesktop.org/show_bug.cgi?id=28832
> bug.
> 
> v1: Eduardo Lima Mitev:  Wrong indentation inside the if clause.
> 
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90367
> 
> Signed-off-by: Marius Predut 
> ---
>  src/mesa/drivers/dri/i915/i915_state.c | 15 +++
>  1 file changed, 15 insertions(+)
> 
> diff --git a/src/mesa/drivers/dri/i915/i915_state.c 
> b/src/mesa/drivers/dri/i915/i915_state.c
> index 4c83073..ebb4e9a 100644
> --- a/src/mesa/drivers/dri/i915/i915_state.c
> +++ b/src/mesa/drivers/dri/i915/i915_state.c
> @@ -599,6 +599,21 @@ i915LineWidth(struct gl_context * ctx, GLfloat widthf)
> 
> width = (int) (widthf * 2);
> width = CLAMP(width, 1, 0xf);
> +
> +   if (ctx->Line.Width < 1.5 || widthf < 1.5) {
> + /* For 1 pixel line thickness or less, the general
> +  * anti-aliasing algorithm gives up, and a garbage line is
> +  * generated.  Setting a Line Width of 0.0 specifies the
> +  * rasterization of the "thinnest" (one-pixel-wide),
> +  * non-antialiased lines.
> +  *
> +  * Lines rendered with zero Line Width are rasterized using
> +  * Grid Intersection Quantization rules as specified by
> +  * bspec G45: Volume 2: 3D/Media,

Eh... Isn't G45 actually i965-class hardware?  GEN4.5, right?  I don't
see how a GEN4 bspec reference is useful (or correct?) in the GEN3
driver.  I think you want "2.8.4.1 Zero-Width (Cosmetic) Line
Rasterization" from volume 1f of the GEN3 docs.

I looked at section 2.8.4.3 Anti-aliased Line Rasterization of volume 1f
of the GEN3 docs that I have.  I don't see any mention of this
restriction.  Is it documented anywhere?

It also seems like this will affect non-antialised wide lines.  Is that
correct?

> +  * 7.3.13.1 Zero-Width (Cosmetic) Line Rasterization section
> +  */
> +  width = 0;
> +   }
> lis4 |= width << S4_LINE_WIDTH_SHIFT;
>  
> if (lis4 != i915->state.Ctx[I915_CTXREG_LIS4]) {

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev