platform/upstream/mesa.git
8 years agonvc0: get rid of NVE4_COMPUTE_MP_PM_{A,B}_SIGSEL_XXX
Samuel Pitoiset [Sat, 5 Nov 2016 16:56:02 +0000 (17:56 +0100)]
nvc0: get rid of NVE4_COMPUTE_MP_PM_{A,B}_SIGSEL_XXX

Instead, hardcode group sigsel because there are a bunch of unknown
groups, especially on SM50/SM52.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
8 years agogm107/ir: emit RED instead of ATOM when no dst
Samuel Pitoiset [Fri, 4 Nov 2016 19:08:57 +0000 (20:08 +0100)]
gm107/ir: emit RED instead of ATOM when no dst

This is similar to NVC0 and GK110 emitters where we emit
reduction operations instead of atomic operations when the
destination is not used.

Found after writing some tests which check if performance counters
return the expected value. In that case, gred_count returned 0
on gm107 while at least gk106 returned the correct value.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agost/mesa: initialize members of glsl_to_tgsi_instruction in emit_asm()
Brian Paul [Sat, 5 Nov 2016 17:00:10 +0000 (11:00 -0600)]
st/mesa: initialize members of glsl_to_tgsi_instruction in emit_asm()

This fixes random crashes with MSVC release builds.  It seems the
members are implicitly initialized to zero with gcc, but not MSVC.
In particular, the tex_offset_num_offset field was non-zero causing
a loop over the NULL tex_offsets array to crash.

Zero-init those fields and a few others to be safe.

The regression began with acc23b04cfd64e "ralloc: remove memset from
ralloc_size".

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoandroid: amd/common: add support for libmesa_amd_common
Mauro Rossi [Fri, 4 Nov 2016 23:00:29 +0000 (00:00 +0100)]
android: amd/common: add support for libmesa_amd_common

Fixes the following building error introduced with commit 7115e56
and related amd/common dependencies:

external/mesa/src/gallium/drivers/radeonsi/si_shader.c:6861: error: undefined reference to 'ac_is_sgpr_param'
external/mesa/src/gallium/drivers/radeonsi/si_shader.c:6951: error: undefined reference to 'ac_is_sgpr_param'
clang++: error: linker command failed with exit code 1 (use -v to see invocation)

ninja: build stopped: subcommand failed.
build/core/ninja.mk:148: recipe for target 'ninja_wrapper' failed
make: *** [ninja_wrapper] Error 1

Signed-off-by: Marek Olšák <marek.olsak@amd.com>
8 years agowinsys/radeon: don't call surface_best for FMASK
Marek Olšák [Fri, 4 Nov 2016 11:30:08 +0000 (12:30 +0100)]
winsys/radeon: don't call surface_best for FMASK

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98518

Acked-by: Edward O'Callaghan <funfunctor@folklore1984.net>
8 years agomesa: Add linear ETC2/EAC to the compressed format list with ES3 compat.
Kenneth Graunke [Thu, 3 Nov 2016 08:58:03 +0000 (01:58 -0700)]
mesa: Add linear ETC2/EAC to the compressed format list with ES3 compat.

GL_ARB_ES3_compatibility brings ETC2/EAC formats to desktop GL.

The meaning of the GL compressed format list is pretty vague - it's
supposed to return formats for "general-purpose usage".  (GL 4.2
deprecates the list because of this.)  Basically everyone interprets
this as "linear RGB/RGBA".

ETC2/EAC meets that criteria, so while we shouldn't be required to add
it to the list, there's also little harm in doing so, at least on
platforms with native support.  I doubt anyone is using this list for
much anyway, so even on platforms without native support, it's probably
not a big deal.

Makes the following GL45-CTS.gtf43 tests pass:

* GL3Tests.eac_compression_r11.gl_compressed_r11_eac
* GL3Tests.eac_compression_rg11.gl_compressed_rg11_eac
* GL3Tests.eac_compression_signed_r11.gl_compressed_signed_r11_eac
* GL3Tests.eac_compression_signed_rg11.gl_compressed_signed_rg11_eac
* GL3Tests.etc2_compression_rgb8.gl_compressed_rgb8_etc2
* GL3Tests.etc2_compression_rgb8_pt_alpha1.gl_compressed_rgb8_pt_alpha1_etc2
* GL3Tests.etc2_compression_rgba8.gl_compressed_rgba8_etc2

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
8 years agovc4: Use Newton-Raphson on the 1/W write to fix glmark2 terrain.
Eric Anholt [Fri, 4 Nov 2016 20:41:20 +0000 (13:41 -0700)]
vc4: Use Newton-Raphson on the 1/W write to fix glmark2 terrain.

The 1/W was apparently not accurate enough, and we were getting sparklies
in the distance.  The closed driver also did a N-R step here.

Cc: <mesa-stable@lists.freedesktop.org>
8 years agovc4: Make sure that vertex shader texture2D() calls use LOD 0.
Eric Anholt [Fri, 4 Nov 2016 19:04:15 +0000 (12:04 -0700)]
vc4: Make sure that vertex shader texture2D() calls use LOD 0.

I noticed this while trying to debug glmark2 terrain (which does vertex
shader texturing, but no mipmaps on its textures sampled from the VS).

8 years agoradeonsi: fix vertex fetches for 2_10_10_10 formats
Nicolai Hähnle [Wed, 2 Nov 2016 18:07:40 +0000 (19:07 +0100)]
radeonsi: fix vertex fetches for 2_10_10_10 formats

The hardware always treats the alpha channel as unsigned, so add a shader
workaround. This is rare enough that we'll just build a monolithic vertex
shader.

The SINT case cannot actually happen in OpenGL, but I've included it for
completeness since it's just a mix of the other cases.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agost/mesa: fix the layer of VDPAU surface samplers
Nicolai Hähnle [Thu, 3 Nov 2016 20:49:40 +0000 (21:49 +0100)]
st/mesa: fix the layer of VDPAU surface samplers

A (latent) bug in VDPAU interop was exposed by commit
e5cc84dd43be066c1dd418e32f5ad258e31a150a.

Before that commit, the st_vdpau code created samplers with
first_layer == last_layer == 1 that the general texture handling code
would immediately delete and re-create, because the layer does not match
the information in the GL texture object.

This was correct behavior at least in the DMABUF case, because the imported
resource is supposed to have the correct offset already applied.  In the
non-DMABUF case, this was just plain wrong but apparently nobody noticed.

After that commit, the state tracker assumes that an existing sampler is
correct at all times.  Existing samplers are supposed to be deleted when
they may become invalid, and they will be created on-demand.  This meant
that the sampler with first_layer == last_layer == 1 stuck around, leading
to rendering artefacts (on radeonsi), command stream failures (on r600), and
assertions (in debug builds everywhere).

This patch fixes the problem by simply not creating a sampler at all in
st_vdpau_map_surface.  We rely on the generic texture code to do the right
thing, adding the layer_override to make the non-DMABUF case work.

v2: add the layer_override

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98512
Cc: 13.0 <mesa-stable@lists.freedesktop.org>
Cc: Christian König <deathsimple@vodafone.de>
Cc: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com> (v1)
Reviewed-by: Christian König <christian.koenig@amd.com>
8 years agoRevert "st/vdpau: use linear layout for output surfaces"
Dave Airlie [Thu, 15 Sep 2016 03:58:33 +0000 (13:58 +1000)]
Revert "st/vdpau: use linear layout for output surfaces"

This reverts commit d180de35320eafa3df3d76f0e82b332656530126.

This is a radeon specific hack that causes problems on nouveau
when combined with the SHARED flag later. If radeonsi needs a fix
for this, please fix it in the driver.

[chk]
Using linear surfaces for this makes sense because tilling isn't
beneficial and the surfaces can potentially be shared with other GPUs
using the VDPAU OpenGL interop.

[airlied]
I think we need a flag that isn't SHARED/LINEAR that is more
SHARED_OTHER_GPU.

[mareko]
Does radeonsi need PIPE_BIND_VIDEO_DECODE_OUTPUT that it would translate
into linear ?

[mareko]
My only concern is decoding performance. If the decoder works in 64x1
blocks, tiling will hurt. That's the theory. I don't know how the
decoder works.

Cc: 12.0 13.0 <mesa-stable@lists.freedesktop.org>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Tested-by: Ilia Mirkin <imirkin@alum.mit.edu>
Tested-by: Nayan Deshmukh <nayan26deshmukh@gmail.com> (I+A)
8 years agoradeonsi: fix an assertion failure in si_decompress_sampler_color_textures
Marek Olšák [Thu, 3 Nov 2016 18:16:51 +0000 (19:16 +0100)]
radeonsi: fix an assertion failure in si_decompress_sampler_color_textures

This fixes a crash in Deus Ex: Mankind Divided. Release builds were
unaffected, so it's not too serious.

Cc: 11.2 12.0 13.0 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoglx: make interop ABI visible again
Marek Olšák [Wed, 2 Nov 2016 17:59:22 +0000 (18:59 +0100)]
glx: make interop ABI visible again

This was broken when the GLAPI use was removed from mesa_glinterop.h.

Cc: 12.0 13.0 <mesa-stable@lists.freedesktop.org>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
8 years agoegl: make interop ABI visible again
Marek Olšák [Wed, 2 Nov 2016 17:59:22 +0000 (18:59 +0100)]
egl: make interop ABI visible again

This was broken when the GLAPI use was removed from mesa_glinterop.h.

Cc: 12.0 13.0 <mesa-stable@lists.freedesktop.org>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
8 years agoegl: use util/macros.h
Marek Olšák [Wed, 2 Nov 2016 17:56:39 +0000 (18:56 +0100)]
egl: use util/macros.h

I need the definition of PUBLIC.

Cc: 12.0 13.0 <mesa-stable@lists.freedesktop.org>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
8 years agoradeonsi: enable GLSL 4.50
Nicolai Hähnle [Fri, 7 Oct 2016 16:21:51 +0000 (18:21 +0200)]
radeonsi: enable GLSL 4.50

Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Reviewed-by: Dave Airlie <airlied@redhat.com>
8 years agost/glsl_to_tgsi: fix dvec[34] loads from SSBO
Nicolai Hähnle [Thu, 3 Nov 2016 10:00:36 +0000 (11:00 +0100)]
st/glsl_to_tgsi: fix dvec[34] loads from SSBO

When splitting up loads, we have to add 16 bytes to the offset for
the high components, just like already happens for stores.

Fixes arb_gpu_shader_fp64@shader_storage@layout-std140-fp64-shader.

Cc: 13.0 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoglsl/cache: correct asprintf error handling
Nicolai Hähnle [Thu, 3 Nov 2016 09:23:17 +0000 (10:23 +0100)]
glsl/cache: correct asprintf error handling

From the manpage of asprintf:

   "If memory allocation wasn't possible, or some other error occurs,
    these functions will return -1, and the contents of strp are
    undefined."

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>
8 years agogallium/radeon: Multiply bpe by nsamples in surf_winsys_to_drm
Michel Dänzer [Wed, 2 Nov 2016 09:54:44 +0000 (18:54 +0900)]
gallium/radeon: Multiply bpe by nsamples in surf_winsys_to_drm

For symmetry with surf_drm_to_winsys.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agogallium/radeon: Use flags parameter in radeon_winsys_surface_init
Michel Dänzer [Wed, 2 Nov 2016 10:48:35 +0000 (19:48 +0900)]
gallium/radeon: Use flags parameter in radeon_winsys_surface_init

Fixes valgrind warnings about surf_ws->flags being uninitialized while
starting X.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agogallium/radeon: Only convert stencil info if RADEON_SURF_SBUFFER is set
Michel Dänzer [Wed, 2 Nov 2016 10:09:06 +0000 (19:09 +0900)]
gallium/radeon: Only convert stencil info if RADEON_SURF_SBUFFER is set

Fixes valgrind warnings about using uninitialized memory when starting X.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agogallium/radeon: Only loop up to last_level for drm<->winsys conversion
Michel Dänzer [Wed, 2 Nov 2016 09:43:37 +0000 (18:43 +0900)]
gallium/radeon: Only loop up to last_level for drm<->winsys conversion

Fixes spurious assertion failure in surf_level_drm_to_winsys when
starting X, due to processing a miplevel which was never initialized.

Fixes: e9c76eeeaa67 ("gallium/radeon: remove radeon_surf_level::pitch_bytes")
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoanv: use limits.h instead of deprecated/obsolete values.h
Tapani Pälli [Thu, 3 Nov 2016 11:44:47 +0000 (13:44 +0200)]
anv: use limits.h instead of deprecated/obsolete values.h

Mesa uses limits.h elsewhere, and this makes is possible to
compile anv_allocator.c on Android.

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
8 years agovc4: Add miptree/texture state support for ETC1 compressed textures.
Eric Anholt [Tue, 6 Jan 2015 21:35:21 +0000 (13:35 -0800)]
vc4: Add miptree/texture state support for ETC1 compressed textures.

The format isn't flagged as enabled at runtime yet, because we need kernel
validation support.

8 years agovc4: Fix use of undefined values since the ralloc zeroing changes.
Eric Anholt [Wed, 2 Nov 2016 20:51:10 +0000 (13:51 -0700)]
vc4: Fix use of undefined values since the ralloc zeroing changes.

reralloc() no longer zeroes the new contents, so switch to using
rzalloc_array() instead.

8 years agonir: Make sure to set the texsrc type in nir drawpixels/bitmap lowering.
Eric Anholt [Thu, 3 Nov 2016 23:28:49 +0000 (16:28 -0700)]
nir: Make sure to set the texsrc type in nir drawpixels/bitmap lowering.

We were leaving an undefined value since the ralloc zeroing changes.
Fixes nir_validate() failures on vc4.

v2: Fix the color-index case of drawpixels as well.

Reviewed-by: Rob Clark <robdclark@gmail.com> (v1)
8 years agodraw: fix undefined input handling some more...
Roland Scheidegger [Fri, 4 Nov 2016 00:48:22 +0000 (01:48 +0100)]
draw: fix undefined input handling some more...

Previous fixes were incomplete - some code still iterated through the number
of elements provided by velem layout instead of the number stored in the key
(which is the same as the number defined by the vs). And also actually
accessed the elements from the layout directly instead of those in the key.
This mismatch could still cause crashes.
(Besides, it is a very good idea to only use data stored in the key anyway.)
v2: move null format check, remove now unnecessary function parameter,
some minor prettify

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agogallium/hud: call fflush() after printing error messages
Brian Paul [Tue, 1 Nov 2016 14:32:04 +0000 (08:32 -0600)]
gallium/hud: call fflush() after printing error messages

For Windows.  Otherwise, we don't see the message until the program exits.

Reviewed-by: Charmaine Lee <charmainel@vmware.com>
8 years agosvga: move svga_mark_surfaces_dirty() prototype to svga_surface.h
Brian Paul [Tue, 1 Nov 2016 14:19:42 +0000 (08:19 -0600)]
svga: move svga_mark_surfaces_dirty() prototype to svga_surface.h

Trivial.

8 years agosvga: whitespace / formatting clean-up in svga_context.c
Brian Paul [Tue, 1 Nov 2016 14:17:05 +0000 (08:17 -0600)]
svga: whitespace / formatting clean-up in svga_context.c

Trivial.

8 years agosvga: collect stats for time spent in svga_context_finish()
Brian Paul [Fri, 28 Oct 2016 20:04:32 +0000 (13:04 -0700)]
svga: collect stats for time spent in svga_context_finish()

This should have appeared with commit "svga: add guest statistic
gathering interface" from August 4, but was somehow lost.

8 years agosvga: invalidate new surface before it is bound to a render target view
Charmaine Lee [Wed, 26 Oct 2016 23:15:23 +0000 (16:15 -0700)]
svga: invalidate new surface before it is bound to a render target view

Invalidate a "new" surface before it is bound to a render target view or
depth stencil view in order to avoid the unnecessary host side copy
of the surface data before it is rendered to.
Note that, recycled surface is already invalidated before it is reused.

Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agoRevert "svga: use untyped surface formats in most cases"
Charmaine Lee [Wed, 26 Oct 2016 22:46:49 +0000 (15:46 -0700)]
Revert "svga: use untyped surface formats in most cases"

Using untyped surface formats causes huge performance degradation on Fusion.
This reverts commit eb0ced74f6decd1bf1e111b162e1389bede89af6 until
the backend has a better solution to address typeless surface formats.

8 years agosvga: allow quad blit for more formats
Charmaine Lee [Fri, 28 Oct 2016 18:48:34 +0000 (11:48 -0700)]
svga: allow quad blit for more formats

Currently blitter will fail if the blit format is different and
view-incompatible to the resource format. Instead of punting
to software blit which will stall the pipeline, we will
create temporary resource to allow blitter to work.

Fixes piglit test arb_copy_image-formats.
Also tested with MTT piglit, glretrace.

Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agosvga: create BGRX render target view for BGRX_UNORM surface
Charmaine Lee [Tue, 25 Oct 2016 20:56:52 +0000 (13:56 -0700)]
svga: create BGRX render target view for BGRX_UNORM surface

Currently we adjust the view format when we are asked to create a
BGRA render target view for BGRX surface. But we only look for
SVGA3D_B8G8R8X8_TYPELESS surface format.
With this patch, we will also check for SVGA3D_B8G8R8X8_UNORM surface format,
and use SVGA3D_B8G8R8X8_UNORM as the view format for that case.

Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agosvga: add a helper function to check for typeless format
Charmaine Lee [Mon, 24 Oct 2016 17:50:29 +0000 (10:50 -0700)]
svga: add a helper function to check for typeless format

This patch adds a helper function svga_format_is_typeless() which
returns TRUE if the specified format is typeless.

Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agosvga: add SVGA_NEW_FRAME_BUFFER to svga_hw_tss_binding state atom
Brian Paul [Tue, 27 Sep 2016 16:06:46 +0000 (10:06 -0600)]
svga: add SVGA_NEW_FRAME_BUFFER to svga_hw_tss_binding state atom

We may need to re-emit texture bindings when the framebuffer state
changes.  In particular, emitting the texture binding can also involve
updating a texture from its backing copy during sampler view validation.
The backing copy is made during framebuffer validation.

This helps to fix an issue with Photoshop on VGPU9 (VMware bug 1723971).

Reviewed-by: Charmaine Lee <charmainel@vmware.com>
8 years agosvga: allow copy_region if sample counts match
Charmaine Lee [Fri, 28 Oct 2016 18:39:47 +0000 (11:39 -0700)]
svga: allow copy_region if sample counts match

With this patch, we will allow blit with copy_region if the
source and destination textures have the same sample counts.

Fixes failures with piglit tests
 spec@arb_texture_float@multisample-formats 2 gl_arb_texture_float
 spec@arb_texture_rg@multisample-formats 2 gl_arb_texture_rg-float

Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agosvga: set rendered-to flag after updating the texture using PredCopyRegion
Charmaine Lee [Mon, 3 Oct 2016 20:29:58 +0000 (13:29 -0700)]
svga: set rendered-to flag after updating the texture using PredCopyRegion

This patch sets the rendered-to flag for the subresource after it is
updated using the PredCopyRegion command. This is to ensure that the GB surface
will be sync up properly before it will be directly mapped to.

Tested with MTT piglit, glretrace.

Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agosvga: add can_use_upload flag
Charmaine Lee [Fri, 30 Sep 2016 23:41:12 +0000 (16:41 -0700)]
svga: add can_use_upload flag

This patch adds a flag "can_use_upload" to svga_texture structure
to avoid some checking of the upload availability at each transfer map time.

Tested with Lightsmark2008, Tropics, MTT glretrace, piglit.

Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agosvga: fix texture upload path condition
Charmaine Lee [Fri, 30 Sep 2016 22:52:14 +0000 (15:52 -0700)]
svga: fix texture upload path condition

As Thomas suggested, we'll first try to map directly to a GB surface.
If it is blocked, then we'll use texture upload buffer.
Also if a texture is already "rendered to", that is, the GB surface
is already out of sync, then we'll use the texture upload buffer
to avoid syncing the GB surface.

Tested with Lightsmark2008, Tropics, MTT piglit, glretrace.

Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agosvga: set rendered_to flag with texture uploaded using TransferFromBuffer command
Charmaine Lee [Thu, 29 Sep 2016 23:41:21 +0000 (16:41 -0700)]
svga: set rendered_to flag with texture uploaded using TransferFromBuffer command

This patch sets the rendered_to flag for the texture subresource that
is uploaded using the TransferFromBuffer command. This is to ensure that
the subresource will be read back or invalidated before it will be
directly mapped to. This makes sure that the content of the GB surface
will not be accidentally overwritten by the device at suspend/resume time.

Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agosvga: Add render_condition boolean flag in struct svga_context
Neha Bhende [Wed, 28 Sep 2016 17:48:45 +0000 (10:48 -0700)]
svga: Add render_condition boolean flag in struct svga_context

set render_condition flag when driver performs conditional rendering.
Blit using DXPredCopyRegion command gets affected by conditional rendering so
We should check this flag while performing blit operation

Tested with piglit tests.

v2: As per Charmaine's comment, setting render_condition flag if svga_query is valid.
Tested with pigit tests.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
8 years agosvga: Allow DXPredCopyRegion for depth_and_stencil formats.
Neha Bhende [Wed, 21 Sep 2016 19:27:14 +0000 (12:27 -0700)]
svga: Allow DXPredCopyRegion for depth_and_stencil formats.

DXPredCopyRegion supports copy between src and dst for depth_and_stencil
formats if src and dst have same formats.

tested ith piglit

v2: As per Brian's comment, allow DXPredCopyRegion for depth+stencil buffers
if the blit mask is PIPE_MASK_ZS.

Tested with piglit tests and added new piglit test
arb_framebuffer_object-depth-stencil-blit to test this particular testcase.

Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agosvga: fix memory leak in svga_clear_texture()
Neha Bhende [Fri, 28 Oct 2016 18:29:11 +0000 (11:29 -0700)]
svga: fix memory leak in svga_clear_texture()

Piglit tests which uses arb_clear_texture extension, have memory leak issue.
pipe_surface created in svga_clear_texture() was not deleted which happens to be
the cause for memory leak.

tested all arb_clear_texture-* piglit tests with valgrid.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
8 years agosvga: Implement the pipe clear_render_target functionality v2
Thomas Hellstrom [Fri, 28 Oct 2016 18:33:53 +0000 (11:33 -0700)]
svga: Implement the pipe clear_render_target functionality v2

v2: Accounted for the fact that svga_try_clear_render_target also
honors conditional rendering.

Testing done: Excercised all functions in a separate feature branch. Forced
emission of conditional rendering commands when necessary.

Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agosvga: add SVGA_3D_CMD_INVALIDATE_GB_SURFACE support
Charmaine Lee [Wed, 2 Mar 2016 17:31:58 +0000 (09:31 -0800)]
svga: add SVGA_3D_CMD_INVALIDATE_GB_SURFACE support

This command will be used in a subsequent patch to invalidate a surface.

Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agonir: Flip gl_SamplePosition in nir_lower_wpos_ytransform().
Francisco Jerez [Tue, 1 Nov 2016 18:56:13 +0000 (11:56 -0700)]
nir: Flip gl_SamplePosition in nir_lower_wpos_ytransform().

Assuming the hardware is set up to use a screen coordinate system
flipped vertically with respect to the GL's window coordinate system,
the SYSTEM_VALUE_SAMPLE_POS vector will also be flipped vertically
with respect to the value expected by the GL, so we need to give it
the same treatment as gl_FragCoord.  Fixes the following CTS tests on
i965:

 ES31-CTS.functional.shaders.multisample_interpolation.interpolate_at_offset.at_sample_position.default_framebuffer
 ES31-CTS.functional.shaders.sample_variables.sample_pos.correctness.default_framebuffer

when run with any multisample configuration, e.g. rgba8888d24s8ms4.

Cc: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
8 years agoisl: Only allow Y-tiling for ASTC textures
Nanley Chery [Thu, 22 Sep 2016 21:51:37 +0000 (14:51 -0700)]
isl: Only allow Y-tiling for ASTC textures

Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
8 years agoanv/blorp: Don't create linear ASTC surfaces for buffers
Nanley Chery [Wed, 2 Nov 2016 23:18:44 +0000 (16:18 -0700)]
anv/blorp: Don't create linear ASTC surfaces for buffers

Such a surface is not possible on our hardware. Without this change, ISL
surface creation would fail with the next patch.

Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
8 years agoanv/formats: Disallow linear ASTC textures
Nanley Chery [Thu, 22 Sep 2016 22:43:07 +0000 (15:43 -0700)]
anv/formats: Disallow linear ASTC textures

Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
8 years agoanv/formats: Disallow 1D compressed textures
Nanley Chery [Thu, 22 Sep 2016 23:44:19 +0000 (16:44 -0700)]
anv/formats: Disallow 1D compressed textures

Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
8 years agoi965: Use rzalloc for cfg_t
Chris Wilson [Wed, 2 Nov 2016 07:01:35 +0000 (07:01 +0000)]
i965: Use rzalloc for cfg_t

Valgrind reports that we use cfg.cycle_count uninitialised, so zero the
cfg_t on construction.

Fixes: 52d2b28f7f10 ("ralloc: use rzalloc where it's necessary")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agopipe-loader: add libamd_common for radeonsi
Nicolai Hähnle [Thu, 3 Nov 2016 14:25:45 +0000 (15:25 +0100)]
pipe-loader: add libamd_common for radeonsi

This fixes a build regression of commit 7115e56c21ace07cf04f5073ba73a533e2182099.
Sorry for the breakage, this second location for link dependencies escaped
my build tests.

Bugzilla: https://patchwork.freedesktop.org/patch/119816/
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
8 years agoglx/windows: Add wgl.h to the sources list
Andreas Boll [Wed, 2 Nov 2016 11:11:54 +0000 (12:11 +0100)]
glx/windows: Add wgl.h to the sources list

Otherwise it won't be picked in the tarball and the build will fail.

Fixes: 533b3530c12 ("direct-to-native-GL for GLX clients on Cygwin
("Windows-DRI")")
Cc: "13.0" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Andreas Boll <andreas.boll.dev@gmail.com>
Reviewed-by: Jon Turney <jon.turney@dronecode.org.uk>
8 years agoi965: use rzalloc instead of calloc in brwNewProgram
Tapani Pälli [Thu, 3 Nov 2016 09:36:47 +0000 (11:36 +0200)]
i965: use rzalloc instead of calloc in brwNewProgram

commit cc6aa1d161280f10ded7834d1ec2413bc97589fe changed to using rzalloc
for gl_program creation but one instance for program creation was still
using calloc.

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
8 years agoradeonsi: generate GS prolog to (partially) fix triangle strip adjacency rotation
Nicolai Hähnle [Mon, 31 Oct 2016 11:50:09 +0000 (12:50 +0100)]
radeonsi: generate GS prolog to (partially) fix triangle strip adjacency rotation

Fixes GL45-CTS.geometry_shader.adjacency.adjacency_indiced_triangle_strip and
others.

This leaves the case of triangle strips with adjacency and primitive restarts
open. It seems that the only thing that cares about that is a piglit test.
Fixing this efficiently would be really involved, and I don't want to use the
hammer of degrading to software handling of indices because there may well
be software that uses this draw mode (without caring about the precise
rotation of triangles).

v2:
- skip the GS prolog entirely if workaround is not needed
- only check for TES (TES is always non-null when tessellation is used)

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: remove si_shader_context::is_gs_copy_shader
Nicolai Hähnle [Mon, 31 Oct 2016 20:13:12 +0000 (21:13 +0100)]
radeonsi: remove si_shader_context::is_gs_copy_shader

It has become redundant.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: make the GS copy shader owned by the GS selector
Nicolai Hähnle [Mon, 31 Oct 2016 20:10:37 +0000 (21:10 +0100)]
radeonsi: make the GS copy shader owned by the GS selector

The copy shader only depends on the selector. This change avoids creating
separate code paths for monolithic vs. non-monolithic geometry shaders.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: si_shader_vs only depends on the GS selector
Nicolai Hähnle [Mon, 31 Oct 2016 20:09:20 +0000 (21:09 +0100)]
radeonsi: si_shader_vs only depends on the GS selector

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: si_vgt_gs_mode only depends on the selector
Nicolai Hähnle [Mon, 31 Oct 2016 20:07:53 +0000 (21:07 +0100)]
radeonsi: si_vgt_gs_mode only depends on the selector

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: make si_generate_gs_copy_shader usable as a standalone function
Nicolai Hähnle [Mon, 31 Oct 2016 19:53:37 +0000 (20:53 +0100)]
radeonsi: make si_generate_gs_copy_shader usable as a standalone function

It really only depends on the shader selector.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: unify the si_compile_* functions for prologs and epilogs
Nicolai Hähnle [Fri, 28 Oct 2016 20:26:33 +0000 (22:26 +0200)]
radeonsi: unify the si_compile_* functions for prologs and epilogs

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: get rid of no_{prolog,epilog}
Nicolai Hähnle [Mon, 31 Oct 2016 15:30:17 +0000 (16:30 +0100)]
radeonsi: get rid of no_{prolog,epilog}

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: get rid of si_llvm_emit_fs_epilogue
Nicolai Hähnle [Fri, 28 Oct 2016 20:10:00 +0000 (22:10 +0200)]
radeonsi: get rid of si_llvm_emit_fs_epilogue

It is no longer used.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: get rid of get_interp_param
Nicolai Hähnle [Fri, 28 Oct 2016 20:09:26 +0000 (22:09 +0200)]
radeonsi: get rid of get_interp_param

Replace by a simple LLVMGetParam, since ctx->no_prolog is always false.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: get rid of select_interp_param
Nicolai Hähnle [Fri, 28 Oct 2016 20:08:10 +0000 (22:08 +0200)]
radeonsi: get rid of select_interp_param

The condition !ctx->no_prolog is now always true.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: use TCS epilog for monolithic shaders
Nicolai Hähnle [Fri, 28 Oct 2016 19:49:32 +0000 (21:49 +0200)]
radeonsi: use TCS epilog for monolithic shaders

For fixed function TCS, we keep the copying of VS outputs to TES inputs inside
the main function; the call to si_copy_tcs_inputs is moved accordingly.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: extract si_build_tcs_epilog_function
Nicolai Hähnle [Fri, 28 Oct 2016 19:47:02 +0000 (21:47 +0200)]
radeonsi: extract si_build_tcs_epilog_function

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: use VS epilog for monolithic TES
Nicolai Hähnle [Fri, 28 Oct 2016 19:42:47 +0000 (21:42 +0200)]
radeonsi: use VS epilog for monolithic TES

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: use VS prolog and epilog for monolithic shaders
Nicolai Hähnle [Fri, 28 Oct 2016 19:31:27 +0000 (21:31 +0200)]
radeonsi: use VS prolog and epilog for monolithic shaders

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: extract si_build_vs_{prolog,epilog}_function
Nicolai Hähnle [Fri, 28 Oct 2016 19:24:35 +0000 (21:24 +0200)]
radeonsi: extract si_build_vs_{prolog,epilog}_function

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: use PS prolog for monolithic shaders
Nicolai Hähnle [Fri, 28 Oct 2016 13:38:47 +0000 (15:38 +0200)]
radeonsi: use PS prolog for monolithic shaders

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: set num_input_vgprs for fragment shaders in create_function
Nicolai Hähnle [Fri, 28 Oct 2016 13:36:43 +0000 (15:36 +0200)]
radeonsi: set num_input_vgprs for fragment shaders in create_function

So that the prolog generated for monolithic fragment shaders will have the
right signature.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: extract si_build_ps_prolog_function
Nicolai Hähnle [Fri, 28 Oct 2016 12:57:27 +0000 (14:57 +0200)]
radeonsi: extract si_build_ps_prolog_function

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: use PS epilog for monolithic shaders
Nicolai Hähnle [Fri, 28 Oct 2016 12:45:59 +0000 (14:45 +0200)]
radeonsi: use PS epilog for monolithic shaders

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: extract si_build_ps_epilog_function
Nicolai Hähnle [Thu, 27 Oct 2016 16:48:16 +0000 (18:48 +0200)]
radeonsi: extract si_build_ps_epilog_function

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: pass the function name to si_llvm_create_func
Nicolai Hähnle [Thu, 27 Oct 2016 16:57:34 +0000 (18:57 +0200)]
radeonsi: pass the function name to si_llvm_create_func

We will use multiple functions in one module, so they should have
different names.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: split is_monolithic into no_prolog and no_epilog
Nicolai Hähnle [Thu, 27 Oct 2016 15:33:22 +0000 (17:33 +0200)]
radeonsi: split is_monolithic into no_prolog and no_epilog

This helps to achieve a gradual transition towards building monolithic shaders
via inlining.

no_prolog and no_epilog will be removed by the end of the series,
separate_prolog remains in use to control the PS input mapping.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: free data structures when shader compiles fail
Nicolai Hähnle [Thu, 27 Oct 2016 15:22:10 +0000 (17:22 +0200)]
radeonsi: free data structures when shader compiles fail

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: move main TGSI translation into its own function
Nicolai Hähnle [Thu, 27 Oct 2016 15:21:25 +0000 (17:21 +0200)]
radeonsi: move main TGSI translation into its own function

The idea is that adding prolog and epilog code will be pulled out into the
caller.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: add always-inline pass to si_llvm_finalize_module
Nicolai Hähnle [Fri, 28 Oct 2016 12:42:46 +0000 (14:42 +0200)]
radeonsi: add always-inline pass to si_llvm_finalize_module

Change the pass manager as well, since this is a module-level pass. No
noticeable run-time difference on shader-db.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: fix signature of export intrinsic in VS epilog
Nicolai Hähnle [Mon, 31 Oct 2016 10:36:35 +0000 (11:36 +0100)]
radeonsi: fix signature of export intrinsic in VS epilog

The incompatible signature becomes an issue when the VS epilog gets merged
with the main vertex shader at the IR level.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: link against amd_common
Nicolai Hähnle [Fri, 28 Oct 2016 12:41:10 +0000 (14:41 +0200)]
radeonsi: link against amd_common

Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoamd/common: add ac_is_sgpr_param helper
Nicolai Hähnle [Fri, 28 Oct 2016 12:40:24 +0000 (14:40 +0200)]
amd/common: add ac_is_sgpr_param helper

Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoamd/common: build also for gallium drivers
Nicolai Hähnle [Mon, 31 Oct 2016 10:56:09 +0000 (11:56 +0100)]
amd/common: build also for gallium drivers

At least when LLVM is used, which is basically always (unless you're only
building r600 without OpenCL).

Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoamd/common: move llvm helper prototype to ac_llvm_util.h
Nicolai Hähnle [Thu, 27 Oct 2016 14:48:42 +0000 (16:48 +0200)]
amd/common: move llvm helper prototype to ac_llvm_util.h

Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoglsl: fix lowering of UBO references of named blocks
Nicolai Hähnle [Mon, 31 Oct 2016 13:03:10 +0000 (14:03 +0100)]
glsl: fix lowering of UBO references of named blocks

When a UBO reference has the form block_name.foo where block_name refers
to a block where the first member has a non-zero offset, the base offset
was incorrectly added to the reference.

Fixes an assertion triggered in debug builds by
GL45-CTS.enhanced_layouts.uniform_block_layout_qualifier_conflict. That test
doesn't properly check for correct execution in this case, so I am also
going to send out a piglit test.

Cc: 13.0 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
8 years agoglsl: Update deref types when resizing implicitly sized arrays.
Kenneth Graunke [Wed, 2 Nov 2016 20:35:30 +0000 (13:35 -0700)]
glsl: Update deref types when resizing implicitly sized arrays.

At link time, we resolve the size of implicitly sized arrays.
When doing so, we update the type of the ir_variables.  However,
we neglected to update the type of ir_dereference nodes which
reference those variables.

It turns out array_resize_visitor (for GS/TCS/TES interface array
handling) already did 2/3 of the cases for this, so we can simply
refactor the code and reuse it.

This fixes:
GL45-CTS.shader_storage_buffer_object.basic-syntax
GL45-CTS.shader_storage_buffer_object.basic-syntaxSSO

which have an SSBO containing an implicitly sized array, followed
by some other members.  setup_buffer_access uses the dereference
types to compute offsets to fields, and it had a stale type where
the implicitly sized array's length was still 0 instead of the
actual length.

While we're here, we can also fix update_array_sizes to properly
update deref types as well, fixing a FINISHME from 2010.

Cc: mesa-stable@lists.freedesktop.org
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
8 years agomesa/glsl: delete previously linked shaders earlier when linking
Timothy Arceri [Tue, 1 Nov 2016 22:49:58 +0000 (09:49 +1100)]
mesa/glsl: delete previously linked shaders earlier when linking

This moves the delete linked shaders call to
_mesa_clear_shader_program_data() which makes sure we delete them
before returning due to any validation problems.

It also reduces some code duplication.

From the OpenGL 4.5 Core spec:

   "If LinkProgram failed, any information about a previous link of
   that program object is lost. Thus, a failed link does not restore
   the old state of program.

   ...

   If one of these commands is called with a program for which
   LinkProgram failed, no error is generated unless otherwise noted.
   Implementations may return information on variables and interface
   blocks that would have been active had the program been linked
   successfully. In cases where the link failed because the program
   required too many resources, these commands may help applications
   determine why limits were exceeded."

Therefore it's expected that we shouldn't be able to query the
program that failed to link and retrieve information about a
previously successful link.

Before this change the linker was doing validation before freeing
the previously linked shaders and therefore could exit on failure
before they were freed.

This change also fixes an issue in compat profile where a program
with no shaders attached is expect to fall back to fixed function
but was instead trying to relink IR from a previous link.

Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=97715
Cc: "13.0" <mesa-stable@lists.freedesktop.org>
8 years agonir: fix nir_shader_clone() and nir_sweep()
Timothy Arceri [Wed, 2 Nov 2016 22:18:19 +0000 (09:18 +1100)]
nir: fix nir_shader_clone() and nir_sweep()

These were broken in e1af20f18a8 when the info field in nir_shader was
turned into a pointer.

Clone was copying the pointer rather than the data and nir_sweep was
cleaning up shader_info rather than claiming it.

Reviewed-by: Eric Anholt <eric@anholt.net>
8 years agomesa: move shader_info to the start of gl_program
Timothy Arceri [Wed, 2 Nov 2016 22:18:18 +0000 (09:18 +1100)]
mesa: move shader_info to the start of gl_program

This will allow use to use ralloc_parent() on the info field and fix
a regression in nir_sweep() caused by e1af20f18a8.

This is intended to be a temporary requirement that will be removed
when we finish separating shader_info from nir_shader.

Reviewed-by: Eric Anholt <eric@anholt.net>
8 years agost/mesa/r200/i915/i965: use rzalloc() to create gl_program
Timothy Arceri [Wed, 2 Nov 2016 22:18:17 +0000 (09:18 +1100)]
st/mesa/r200/i915/i965: use rzalloc() to create gl_program

This allows us to use ralloc_parent() to see which data structure owns
shader_info which allows us to fix a regression in nir_sweep().

This will also allow us to move some fields from gl_linked_shader to
gl_program, which will allow us to do some clean-ups like storing
gl_program directly in the CurrentProgram array in gl_pipeline_object
enabling some small validation optimisations at draw time.

Also it is error prone to depend on the gl_linked_shader for
programs in current use because a failed linking attempt will free
infomation about the current program. In i965 we could be trying
to recompile a shader variant but may have lost some required fields.

Reviewed-by: Eric Anholt <eric@anholt.net>
8 years agonv50,nvc0: stop limiting the number of active queries to 1
Samuel Pitoiset [Wed, 26 Oct 2016 22:08:30 +0000 (00:08 +0200)]
nv50,nvc0: stop limiting the number of active queries to 1

This limitation was initially here because AMD_performance_monitor
doesn't allow to expose the real number of hardware counters. But
this actually really annoying when profiling with qapitrace.

Anyways, performance counters are mostly for developers and
failures are expected if you try to monitor more queries than
supported.

This breaks amd_performance_monitor_measure but it's expected.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
8 years agonvc0: add new warp_nonpred_execution_efficiency metric on SM35
Samuel Pitoiset [Wed, 2 Nov 2016 22:50:03 +0000 (23:50 +0100)]
nvc0: add new warp_nonpred_execution_efficiency metric on SM35

Event not_predicated_off_thread_inst_executed is SM35+.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: add missing metric-issue_slot on SM35
Samuel Pitoiset [Tue, 1 Nov 2016 18:34:44 +0000 (19:34 +0100)]
nvc0: add missing metric-issue_slot on SM35

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: do not expose metric-inst_issued twice on SM35
Samuel Pitoiset [Tue, 1 Nov 2016 18:19:15 +0000 (19:19 +0100)]
nvc0: do not expose metric-inst_issued twice on SM35

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: add new warp_execution_efficiency metric on SM30+
Samuel Pitoiset [Wed, 2 Nov 2016 22:57:25 +0000 (23:57 +0100)]
nvc0: add new warp_execution_efficiency metric on SM30+

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: respect 80-chars for perf metrics descriptions
Samuel Pitoiset [Tue, 1 Nov 2016 18:23:21 +0000 (19:23 +0100)]
nvc0: respect 80-chars for perf metrics descriptions

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: sort performance metrics alphabetically
Samuel Pitoiset [Tue, 1 Nov 2016 18:22:37 +0000 (19:22 +0100)]
nvc0: sort performance metrics alphabetically

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>