intel: Add marketing names for CHV
PATCH 1/3 intel: Add marketing names for CHV
From: Ville Syrjälä <[email protected]>
All CHV devices will be branded as "Intel(r) HD Graphics".
--- src/intel_module.c | 6 ++++++ 1 file changed, 6 insertions(+)
src/intel_module.c +6 -0
@@ -259,6 +259,12 @@ static const SymTabRec intel_chipsets[] = { {0x163D, "HD graphics"}, {0x163E, "HD graphics"}, +/* Cherryview (Cherrytrail/Braswell) */ + {0x22b0, "HD Graphics"}, + {0x22b1, "HD Graphics"}, + {0x22b2, "HD Graphics"}, + {0x22b3, "HD Graphics"}, +/* When adding new identifiers, also update: * 1. intel_identify() * 2. man/intel.man
Signature · 6 lines
2.0.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 2/3 sna/gen8: Reduce CHV VS URB entries to 640
From: Ville Syrjälä <[email protected]>
According to BSpec the max number of VS URB entries for CHV is 640.
--- src/sna/gen8_render.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)
src/sna/gen8_render.c +6 -1
@@ -205,6 +205,11 @@ static const struct blendinfo {#define OUT_VERTEX (x,y) vertex_emit_2s(sna, x,y)#define OUT_VERTEX_F (v) vertex_emit(sna, v) +inline static bool is_chv (struct sna *sna) +{ +return sna->kgem.gen ==0101 ; +} +static inline bool too_large (int width,int height) {return width >GEN8_MAX_SIZE || height >GEN8_MAX_SIZE ;
@@ -462,7 +467,7 @@ gen8_emit_urb(struct sna *sna) {/* num of VS entries must be divisible by 8 if size < 9 */ OUT_BATCH (GEN8_3DSTATE_URB_VS | (2 -2 )); -OUT_BATCH (960 <<URB_ENTRY_NUMBER_SHIFT | +OUT_BATCH ((is_chv (sna) ?640 :960 ) <<URB_ENTRY_NUMBER_SHIFT | (2 -1 ) <<URB_ENTRY_SIZE_SHIFT |4 <<URB_STARTING_ADDRESS_SHIFT );
Signature · 6 lines
2.0.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 3/3 gen8: Fix the YUV->RGB shader
In reply to [email protected]From: Ville Syrjälä <[email protected]>
Use the correct register (Yn_01) with first half of the Y samples instead of using the register (Yn_23) with the second half twice when computing the green channel.
Also use the Yn_01 register name instead of Yn for the red channel as well, just for a bit of extra consistency.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89807
--- src/render_program/exa_wm_yuv_rgb.g8a | 4 ++-- src/render_program/exa_wm_yuv_rgb.g8b | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-)
src/render_program/exa_wm_yuv_rgb.g8a +2 -2
@@ -76,7 +76,7 @@ add (16) Cbn<1>F Cb<8,8,1>F -0.501961F { compr align1 }; /* * R = Y + Cr * 1.596 */ -mov (8) acc0<1>F Yn<8,8,1>F { compr align1 }; +mov (8) acc0<1>F Yn_01<8,8,1>F { compr align1 }; mac.sat(8) src_sample_r_01<1>F Crn_01<8,8,1>F 1.596F { compr align1 }; mov (8) acc0<1>F Yn_23<8,8,1>F { compr align1 };
@@ -84,7 +84,7 @@ mac.sat(8) src_sample_r_23<1>F Crn_23<8,8,1>F 1.596F { compr align1 }; /* * G = Crn * -0.813 + Cbn * -0.392 + Y */ -mov (8) acc0<1>F Yn_23<8,8,1>F { compr align1 }; +mov (8) acc0<1>F Yn_01<8,8,1>F { compr align1 }; mac (8) acc0<1>F Crn_01<8,8,1>F -0.813F { compr align1 }; mac.sat(8) src_sample_g_01<1>F Cbn_01<8,8,1>F -0.392F { compr align1 };
src/render_program/exa_wm_yuv_rgb.g8b +1 -1
@@ -6,7 +6,7 @@ { 0x80600048, 0x21c03ae8, 0x3e8d02c0, 0x3fcc49ba }, { 0x00600001, 0x24003ae0, 0x008d0320, 0x00000000 }, { 0x80600048, 0x21e03ae8, 0x3e8d02e0, 0x3fcc49ba }, - { 0x00600001, 0x24003ae0, 0x008d0320, 0x00000000 }, + { 0x00600001, 0x24003ae0, 0x008d0300, 0x00000000 }, { 0x00600048, 0x24003ae0, 0x3e8d02c0, 0xbf5020c5 }, { 0x80600048, 0x22003ae8, 0x3e8d0340, 0xbec8b439 }, { 0x00600001, 0x24003ae0, 0x008d0320, 0x00000000 },
Signature · 6 lines
2.0.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH xf86-video-intel 3/3] gen8: Fix the YUV->RGB shader
Quoted 11 lines · On Thu, Apr 16, 2015 at 08:40:39PM +0300, [email protected] wrote:
From: Ville Syrjälä <[email protected]>
Use the correct register (
Yn_01) with first half of the Y samples instead of using the register (Yn_23) with the second half twice when computing the green channel.Also use the
Yn_01register name instead of Yn for the red channel as well, just for a bit of extra consistency.Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89807
Signed-off-by Ville Syrjälä <[email protected]>
My hero. I had decided that the problem lay in the shaders and was going to start rewriting them...
-Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 5 lines · On Thu, Apr 16, 2015 at 08:40:37PM +0300, [email protected] wrote:
From: Ville Syrjälä <[email protected]>
All CHV devices will be branded as "Intel(r) HD Graphics".
Signed-off-by Ville Syrjälä <[email protected]>
Tweaked patch 2 to start filling out a gt_info struct like gen7, for the moment it allows us to sneak the codename into the logs without putting into the product identifier.
Pushed with thanks,
-Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx