Re: [Mesa-dev] [PATCH v2] nv50: fix PIPE_QUERY_TIMESTAMP_DISJOINT, based on nvc0

On 05/23/2015 06:59 PM, Ilia Mirkin wrote: On Sat, May 23, 2015 at 12:59 PM, Samuel Pitoiset samuel.pitoi...@gmail.com wrote: On 05/23/2015 06:53 PM, Ilia Mirkin wrote: Reviewed-by: Ilia Mirkin imir...@alum.mit.edu Seems in line with nvc0, but I assume you tested this? e.g. the occlquery

Re: [Mesa-dev] [PATCH] nv50: fix PIPE_QUERY_TIMESTAMP_DISJOINT, based on nvc0

On 05/23/2015 06:47 PM, Ilia Mirkin wrote: On Sat, May 23, 2015 at 12:46 PM, Samuel Pitoiset samuel.pitoi...@gmail.com wrote: PIPE_QUERY_TIMESTAMP_DISJOINT could not work because q-ready was always set to FALSE. To fix this issue, add more different states for queries according to nvc0

[Mesa-dev] [PATCH v2] nv50: fix PIPE_QUERY_TIMESTAMP_DISJOINT, based on nvc0

PIPE_QUERY_TIMESTAMP_DISJOINT could not work because q-ready was always set to FALSE. To fix this issue, add more different states for queries according to nvc0. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_query.c | 39

[Mesa-dev] [PATCH 1/2] nv50: use a fence for 64 bits queries, based on nvc0

A sequence number is written for 32-bits queries to make sure they are ready, but not for 64-bits queries. Instead, we have to use a fence. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_query.c | 14 +- 1 file changed, 13

[Mesa-dev] [PATCH 2/2] nv50: make pipeline statistics queries use 64-bits, based on nvc0

Tested on NVA8. No regression for ARB_pipeline_statistics piglit tests. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_query.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nouveau/nv50/nv50_query.c

Re: [Mesa-dev] [PATCH 1/2] nv50: fix 64 bits queries, based on nvc0

... As an aside, nv50_query_end for PIPE_QUERY_TIMESTAMP_DISJOINT *is* busted... it sets q-ready = TRUE and then immediately sets it to false :( Yep :/ On Thu, May 21, 2015 at 3:28 PM, Samuel Pitoiset samuel.pitoi...@gmail.com wrote: According to nvc0, 64-bits queries use a fence to make sure the result

Re: [Mesa-dev] [PATCH 2/2] nv50: fix pipeline statistics queries, based on nvc0

. Do we have piglit tests for those queries? I'm not really sure. On Thu, May 21, 2015 at 3:28 PM, Samuel Pitoiset samuel.pitoi...@gmail.com wrote: These queries use 64 bits. Tested on NVA8. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50

[Mesa-dev] [PATCH 1/2] nv50: fix 64 bits queries, based on nvc0

According to nvc0, 64-bits queries use a fence to make sure the result is available. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_query.c | 14 +- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers

[Mesa-dev] [PATCH 2/2] nv50: fix pipeline statistics queries, based on nvc0

These queries use 64 bits. Tested on NVA8. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_query.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nouveau/nv50/nv50_query.c b/src/gallium/drivers/nouveau

[Mesa-dev] [PATCH] vc4: make vc4_begin_query() returns a boolean

I forgot to make the change in 96f164f6f047833091eb98a73aa80c31dc94f962. This fixes a warning with GCC and probably an error with Clang. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/vc4/vc4_query.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff

Re: [Mesa-dev] [PATCH 06/11] gallium/hud: automatically print % if max_value == 100

On 08/03/2015 05:28 PM, Marek Olšák wrote: On Mon, Aug 3, 2015 at 2:58 PM, Samuel Pitoiset samuel.pitoi...@gmail.com wrote: Patches 1-6 are: Reviewed-by: Samuel Pitoiset samuel.pitoi...@gmail.com But please, fix the commit message for patches 1 and 3 (ie. gallium/hud instead of gallium, hud

[Mesa-dev] [PATCH v2 7/7] nv50: enable GL_AMD_performance_monitor

This exposes a group of global performance counters that enables GL_AMD_performance_monitor. All piglit tests are okay. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_query.c | 35 ++ src/gallium/drivers/nouveau/nv50

[Mesa-dev] [PATCH v2 5/7] nv50: add support for compute/graphics global performance counters

later. Changes since v2: - replace \% by percentage - remove one extra call to PUSH_SPACE - use nouveau_fence instead of my hand-made fence mechanism Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_query.c | 1066 +++- src

[Mesa-dev] [PATCH v2 3/7] nv50: allocate and map a notifier buffer object for PM

in order to prevent stalls when reading queries. Changes since v2: - check return value of nouveau_bo_map() - add a libdrm version check around creating the notifier BO Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_screen.c | 32

[Mesa-dev] [PATCH v2 1/7] nouveau: implement the nvif hardware performance counters interface

/nouveau_perfmon.c @@ -0,0 +1,290 @@ +/* + * Copyright 2015 Samuel Pitoiset + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the Software), + * to deal in the Software without restriction, including without limitation

[Mesa-dev] [PATCH v2 2/7] nv50: allocate a software object class

This will allow to monitor global performance counters through the command stream of the GPU instead of using ioctls. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_screen.c | 11 +++ src/gallium/drivers/nouveau/nv50/nv50_screen.h | 1

[Mesa-dev] [PATCH v2 4/7] nv50: configure the ring buffer for reading back PM counters

To write data at the right offset, the kernel has to know some parameters of this ring buffer, like the number of domains and the maximum number of queries. Changes since v2: - only configure the ring buffer if the notifier BO is allocated - only use one BEGIN_NV04() Signed-off-by: Samuel

[Mesa-dev] [PATCH v2 6/7] nv50: expose global performance counters to the HUD

Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_query.c | 41 ++ src/gallium/drivers/nouveau/nv50/nv50_screen.c | 1 + src/gallium/drivers/nouveau/nv50/nv50_screen.h | 3 ++ 3 files changed, 45 insertions(+) diff --git

[Mesa-dev] [PATCH v2 0/7] nv50: expose global performance counters

series which exposes global performance counters for Fermi and Kepler will be submitted once I have got enough reviews for this one. Feel free to make a review. Thanks, Samuel. Samuel Pitoiset (7): nouveau: implement the nvif hardware performance counters interface nv50: allocate a software

Re: [Mesa-dev] [PATCH 3/3] st/mesa: implement DrawTransformFeedbackStream

On 07/30/2015 04:26 PM, Marek Olšák wrote: From: Marek Olšák marek.ol...@amd.com --- src/mesa/state_tracker/st_cb_xformfb.c | 58 ++ src/mesa/state_tracker/st_cb_xformfb.h | 2 +- src/mesa/state_tracker/st_draw.c | 2 +- 3 files changed, 33

Re: [Mesa-dev] [PATCH 06/11] gallium/hud: automatically print % if max_value == 100

On 08/03/2015 08:14 PM, Marek Olšák wrote: On Mon, Aug 3, 2015 at 2:58 PM, Samuel Pitoiset samuel.pitoi...@gmail.com wrote: Patches 1-6 are: Reviewed-by: Samuel Pitoiset samuel.pitoi...@gmail.com But please, fix the commit message for patches 1 and 3 (ie. gallium/hud instead of gallium, hud

Re: [Mesa-dev] [PATCH 06/11] gallium/hud: automatically print % if max_value == 100

Patches 1-6 are: Reviewed-by: Samuel Pitoiset samuel.pitoi...@gmail.com But please, fix the commit message for patches 1 and 3 (ie. gallium/hud instead of gallium, hud). Btw, it would be good to display floating point numbers when percentage is used. What do you think ? On 08/03/2015 02

Re: [Mesa-dev] [PATCH] nvc0: bind a fake tess control program when there isn't one available

On 07/26/2015 06:56 AM, Ilia Mirkin wrote: Apparently this is necessary in order for tess factors to work in a tess eval program without a tess control program bound. Probably because it uses the fake program's shader header to work out the number of patch constants. Fixes

[Mesa-dev] [PATCH] nv50: force cache flush for Uniform Buffer Objects

This fixes the following piglit test: ext_transform_feedback-immediate-reuse-uniform-buffer Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_shader_state.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/gallium/drivers/nouveau/nv50

[Mesa-dev] [PATCH] nvc0: force cache flush when binding a new ubo

This fixes the following piglit test: ext_transform_feedback-immediate-reuse-uniform-buffer Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nvc0/nvc0_state_validate.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/gallium/drivers/nouveau/nvc0

Re: [Mesa-dev] [PATCH v2 1/7] nouveau: implement the nvif hardware performance counters interface

On 07/22/2015 10:29 PM, Martin Peres wrote: On 01/07/15 01:01, Samuel Pitoiset wrote: This commit implements the base interface for hardware performance counters that will be shared between nv50 and nvc0 drivers. TODO: Bump libdrm version of mesa when nvif will be merged. Changes since v2

Re: [Mesa-dev] [PATCH v2 2/7] nv50: allocate a software object class

On 07/22/2015 10:35 PM, Martin Peres wrote: On 01/07/15 01:01, Samuel Pitoiset wrote: This will allow to monitor global performance counters through the command stream of the GPU instead of using ioctls. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers

Re: [Mesa-dev] [PATCH v2 4/7] nv50: configure the ring buffer for reading back PM counters

On 07/22/2015 10:54 PM, Martin Peres wrote: On 01/07/15 01:01, Samuel Pitoiset wrote: To write data at the right offset, the kernel has to know some parameters of this ring buffer, like the number of domains and the maximum number of queries. Changes since v2: - only configure the ring

Re: [Mesa-dev] [PATCH v2 5/7] nv50: add support for compute/graphics global performance counters

On 07/23/2015 12:05 AM, Martin Peres wrote: On 01/07/15 01:01, Samuel Pitoiset wrote: This commit adds support for both compute and graphics global performance counters which have been reverse engineered with CUPTI (Linux) and PerfKit (Windows). Currently, only one query type can

Re: [Mesa-dev] [PATCH v2 7/7] nv50: enable GL_AMD_performance_monitor

On 07/23/2015 12:14 AM, Martin Peres wrote: On 01/07/15 01:01, Samuel Pitoiset wrote: This exposes a group of global performance counters that enables GL_AMD_performance_monitor. All piglit tests are okay. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers

[Mesa-dev] [PATCH] nv50: limit the maximum number of samplers to 16

NV50_3D_BIND_TSC only allows to bind 16 samplers, and since we don't want to do anything with NV50_3D_BIND_TSC2, just limit the maximum number of samplers to 16 like for nvc0. This fixes dmesg fails with the following piglit test: max-samplers But the test still fails. Signed-off-by: Samuel

[Mesa-dev] [PATCH 1/2] nv50: add nesting support for occlusion queries

This is loosely based on nvc0. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_query.c | 27 -- src/gallium/drivers/nouveau/nv50/nv50_screen.h | 2 ++ 2 files changed, 19 insertions(+), 10 deletions(-) diff --git a/src

[Mesa-dev] [PATCH 2/2] nv50: turn samples counts off during blit

Fixes the following piglit test: occlusion_query_meta_no_fragments Signed-off-by: Samuel Pitoiset sameul.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_surface.c | 11 +++ 1 file changed, 11 insertions(+) diff --git a/src/gallium/drivers/nouveau/nv50/nv50_surface.c b

[Mesa-dev] [PATCH 2/2] nv50/nvc0: force cache flush for constbufs

This fixes the following piglit test: ext_transform_feedback-immediate-reuse-uniform-buffer I didn't test on nvc0 but this should work as expected. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_shader_state.c | 2 ++ src/gallium/drivers

[Mesa-dev] [PATCH 1/2] nouveau: always align buffers to 0x100

: ext_transform_feedback-immediate-reuse-uniform-buffer -auto -fbo Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nouveau_buffer.c | 7 +-- 1 file changed, 1 insertion(+), 6 deletions(-) diff --git a/src/gallium/drivers/nouveau/nouveau_buffer.c b/src/gallium

[Mesa-dev] [PATCH] nvc0: add a missing parameter to nvc0_set_shader_images()

This fixes a compilation warning introduced in commit 05a12c5 (gallium: add interface for writable shader images). While we are at it, fix indentation and rename parameters according to the gallium interface. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers

Re: [Mesa-dev] [Nouveau] [PATCH] nv50: adjust min/max lod by base level on G80

Reviewed-by: Samuel Pitoiset samuel.pitoi...@gmail.com On 07/20/2015 09:26 AM, Ilia Mirkin wrote: Make the assumption that there's a 1:1 TIC - TSC connection, and increase min/max lod by the relevant texture's base level. Also if there's no mipfilter, we have to enable it while forcing min/max

[Mesa-dev] [PATCH 1/2] nv50: add nesting support for occlusion queries

This is loosely based on nvc0. Signed-off-by: Samuel Pitoiset samuel.pitoi...@gmail.com --- src/gallium/drivers/nouveau/nv50/nv50_query.c | 29 -- src/gallium/drivers/nouveau/nv50/nv50_screen.h | 2 ++ 2 files changed, 20 insertions(+), 11 deletions(-) diff --git a/src

Re: [Mesa-dev] [PATCH] nvc0: fix geometry program revalidation of clipping params

/listinfo/mesa-dev -- Best regards, Samuel Pitoiset. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH] nv50: avoid using inline vertex data submit when gl_VertexID is used

be happy to have a look at this second approach. On Mon, Aug 24, 2015 at 4:07 PM, Samuel Pitoiset samuel.pitoi...@gmail.com wrote: Reviewed-by: Samuel Pitoiset samuel.pitoi...@gmail.com This fix is simpler than I was expected. What about the edge flag stuff now? :) On 08/24/2015 05:51 PM, Ilia

Re: [Mesa-dev] [PATCH] nv50: avoid using inline vertex data submit when gl_VertexID is used

Reviewed-by: Samuel Pitoiset samuel.pitoi...@gmail.com This fix is simpler than I was expected. What about the edge flag stuff now? :) On 08/24/2015 05:51 PM, Ilia Mirkin wrote: The hardware only generates vertexid when vertices come from a VBO. This fixes: vertexid-drawelements

Re: [Mesa-dev] [PATCH] nouveau: fix double free when screen_create fails

On 10/22/2015 01:16 AM, Julien Isorce wrote: The real fix is in nouveau_drm_winsys.c by setting dev to 0. Which means dev's ownership has been passed to previous call. Other changes are there to be consistent with what the screen_create functions already do on errors. This actually happens

Re: [Mesa-dev] [PATCH v2 1/7] nvc0: fix crash when nv50_miptree_from_handle fails

? Anyway, this patch is: Reviewed-by: Samuel Pitoiset <samuel.pitoi...@gmail.com <mailto:samuel.pitoi...@gmail.com>> On 20/10/2015 18:34, Julien Isorce wrote: Signed-off-by: Julien Isorce <j.iso...@samsung.com <mailto:j.iso...@samsung.com>>

Re: [Mesa-dev] [PATCH] nouveau: fix double free when screen_create fails

On 10/27/2015 02:01 PM, samuel.pitoiset wrote: On 27/10/2015 12:52, Emil Velikov wrote: On 27 October 2015 at 10:50, samuel.pitoiset wrote: On 27/10/2015 11:37, Emil Velikov wrote: On 22 October 2015 at 00:16, Julien Isorce wrote: The

[Mesa-dev] [PATCH] nvc0: expose a group of performance metrics on Fermi

This allows to monitor those performance metrics through GL_AMD_performance_monitor. Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> --- src/gallium/drivers/nouveau/nvc0/nvc0_query.c | 14 +- src/gallium/drivers/nouveau/nvc0/nvc0_query.h | 3 ++

Re: [Mesa-dev] [PATCH v2 1/7] nvc0: fix crash when nv50_miptree_from_handle fails

On 10/26/2015 01:44 PM, Julien Isorce wrote: On 25 October 2015 at 21:38, Samuel Pitoiset <samuel.pitoi...@gmail.com <mailto:samuel.pitoi...@gmail.com>> wrote: Do you need someone to push this patch? Yes please Pushed.

[Mesa-dev] [PATCH v2 3/4] nv50: move HW queries to nv50_query_hw.c/h files

Changes since v2: - remove unused 'nv50_hw_query_funcs' struct Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> --- src/gallium/drivers/nouveau/Makefile.sources | 2 + src/gallium/drivers/nouveau/nv50/nv50_query.c | 354 ++- src/gallium/drivers/nouvea

[Mesa-dev] [PATCH v2 4/4] nv50: do not create an invalid HW query type

While we are at it, store the rotate offset for occlusion queries to nv50_hw_query like on nvc0. Changes since v2: - remove useless 'space' variable Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> Reviewed-by: Pierre Moreau <pierre.mor...@free.fr> --- src/gallium/drivers/

Re: [Mesa-dev] [PATCH 3/3] nvc0: add ARB_clear_texture support

On 11/09/2015 09:03 PM, Ilia Mirkin wrote: On Mon, Nov 9, 2015 at 2:58 PM, Samuel Pitoiset <samuel.pitoi...@gmail.com> wrote: On 11/09/2015 07:40 PM, Ilia Mirkin wrote: Signed-off-by: Ilia Mirkin <imir...@alum.mit.edu> --- docs/GL3.txt|

Re: [Mesa-dev] [PATCH 3/3] nvc0: add ARB_clear_texture support

On 11/09/2015 07:40 PM, Ilia Mirkin wrote: Signed-off-by: Ilia Mirkin --- docs/GL3.txt| 2 +- docs/relnotes/11.1.0.html | 1 + src/gallium/drivers/nouveau/nvc0/nvc0_screen.c | 2 +-

Re: [Mesa-dev] [PATCH 3/3] nvc0: add ARB_clear_texture support

On 11/09/2015 09:14 PM, Ilia Mirkin wrote: On Mon, Nov 9, 2015 at 3:13 PM, Samuel Pitoiset <samuel.pitoi...@gmail.com> wrote: On 11/09/2015 09:03 PM, Ilia Mirkin wrote: On Mon, Nov 9, 2015 at 2:58 PM, Samuel Pitoiset <samuel.pitoi...@gmail.com> wrote: On 11/09/2015 07

Re: [Mesa-dev] [PATCH] nouveau: fix double free when screen_create fails

Hi Emil, On 11/10/2015 04:35 PM, Emil Velikov wrote: Hi Samuel, Sorry about this I thought I already replied :-\ On 29 October 2015 at 22:22, Samuel Pitoiset <samuel.pitoi...@gmail.com> wrote: On 10/27/2015 02:01 PM, samuel.pitoiset wrote: On 27/10/2015 12:52, Emil Velikov wrote:

Re: [Mesa-dev] [PATCH] nouveau: fix double free when screen_create fails

On 11/12/2015 08:51 PM, Samuel Pitoiset wrote: Hi Emil, On 11/10/2015 04:35 PM, Emil Velikov wrote: Hi Samuel, Sorry about this I thought I already replied :-\ On 29 October 2015 at 22:22, Samuel Pitoiset <samuel.pitoi...@gmail.com> wrote: On 10/27/2015 02:01 PM, samuel.pitoiset

Re: [Mesa-dev] [PATCH 1/5] nv50: implement a basic compute support

On 11/13/2015 01:15 AM, Ilia Mirkin wrote: On Thu, Nov 12, 2015 at 7:04 PM, Samuel Pitoiset <samuel.pitoi...@gmail.com> wrote: +static bool +nv50_compute_validate_program(struct nv50_context *nv50) +{ + struct nv50_program *prog = nv50->compprog; + + if (prog->mem) +

[Mesa-dev] [PATCH 1/5] nv50: implement a basic compute support

versa. Note that, textures, samplers and surfaces still need to be implemented. Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> Tested-by: Pierre Moreau <pierre.mor...@free.fr> --- src/gallium/drivers/nouveau/Makefile.sources | 1 + .../drivers/nouveau/codegen/nv50

[Mesa-dev] [PATCH 4/5] nv50: expose two groups of compute-related MP perf counters

This turns on GL_AMD_performance_monitor which needs at least one group of GPU counters to be enabled. Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> --- src/gallium/drivers/nouveau/nv50/nv50_query.c | 55 ++ src/gallium/drivers/nouveau/nv50/nv50_q

[Mesa-dev] [PATCH 2/5] nv50: add compute-related MP perf counters on G84+

Fermi has 8. Only G84+ is supported because G80 is an old and weird card. Tested on G84, G96, G200, MCP79 and GT218 with glxgears, glxspheres64, xonotic-glx, heaven and valley. Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> Tested-by: Pierre Moreau <pierre.mor...@free.fr&

[Mesa-dev] [PATCH 3/5] nv50: add support for performance metrics on G84+

Currently only one metric is exposed but more will be added later. Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> Tested-by: Pierre Moreau <pierre.mor...@free.fr> --- src/gallium/drivers/nouveau/Makefile.sources | 2 + src/gallium/drivers/nouveau/nv50/nv50_query_

[Mesa-dev] [PATCH 5/5] docs: mark GL_AMD_performance_monitor for nv50

Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> --- docs/relnotes/11.1.0.html | 1 + 1 file changed, 1 insertion(+) diff --git a/docs/relnotes/11.1.0.html b/docs/relnotes/11.1.0.html index 82ee3c4..14dc5e4 100644 --- a/docs/relnotes/11.1.0.html +++ b/docs/relnotes/11.1.0.html @@

[Mesa-dev] [PATCH 0/5] nv50: compute support & some perf counters

tested by Pierre Moreau and myself on different chipsets. I didn't run a full piglit but I'll do it in the next few days before merging the series. :-) Feel free to review. Thanks! Samuel Pitoiset (5): nv50: implement a basic compute support nv50: add compute-related MP perf counters on G84

Re: [Mesa-dev] [PATCH 6/9] st/mesa: maintain active perfmon counters in an array

On 11/13/2015 04:57 PM, Nicolai Hähnle wrote: It is easy enough to pre-determine the required size, and arrays are generally better behaved especially when they get large. --- src/mesa/state_tracker/st_cb_perfmon.c | 78 --

Re: [Mesa-dev] [PATCH 9/9] st/mesa: add support for batch driver queries to perfmon

On 11/13/2015 04:57 PM, Nicolai Hähnle wrote: --- src/mesa/state_tracker/st_cb_perfmon.c | 75 ++ src/mesa/state_tracker/st_cb_perfmon.h | 6 +++ 2 files changed, 74 insertions(+), 7 deletions(-) diff --git a/src/mesa/state_tracker/st_cb_perfmon.c

Re: [Mesa-dev] [PATCH 4/9] st/mesa: store mapping from perfmon counter to query type

this code one year ago for the nvc0 driver, this wasn't a real issue because it didn't expose lots of queries. Anyway, this looks like a good improvement. Thanks! Reviewed-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> --- src/mesa/state_tracker/st_cb_perfmon.

Re: [Mesa-dev] [PATCH 7/9] gallium: add the concept of batch queries

On 11/13/2015 04:57 PM, Nicolai Hähnle wrote: Some drivers (in particular radeon[si], but also freedreno judging from a quick grep) may want to expose performance counters that cannot be individually enabled or disabled. Allow such drivers to mark driver-specific queries as requiring a new

Re: [Mesa-dev] [PATCH 2/9] gallium/hud: remove unused field in query_info

Reviewed-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> On 11/13/2015 04:57 PM, Nicolai Hähnle wrote: --- src/gallium/auxiliary/hud/hud_driver_query.c | 1 - 1 file changed, 1 deletion(-) diff --git a/src/gallium/auxiliary/hud/hud_driver_query.c b/src/gallium/auxilia

Re: [Mesa-dev] [PATCH 3/9] st/mesa: map semantic driver query types to underlying type

*/ + /* PIPE_DRIVER_QUERY_TYPE_MICROSECONDS */ When you are at it, please also add /* PIPE_DRIVER_QUERY_TYPE_MICROSECONDS */ to pipe_query_result. With this minor change, this patch is: Reviewed-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> /* PIPE_DRIVER_QUERY_T

Re: [Mesa-dev] [PATCH 8/9] gallium/hud: add support for batch queries

Some comments below. On 11/13/2015 04:57 PM, Nicolai Hähnle wrote: --- src/gallium/auxiliary/hud/hud_context.c | 24 ++- src/gallium/auxiliary/hud/hud_driver_query.c | 248 +++ src/gallium/auxiliary/hud/hud_private.h | 13 +- 3 files changed, 240

Re: [Mesa-dev] [PATCH 0/9] gallium: batch query objects and related cleanups

Hi Nicolai, Did you run amd_performance_monitor piglit tests to make sure all of your changes didn't break anything? Did you test on nvc0 driver which is the only driver that currently exposes GL_AMD_performance_monitor? In case you didn't, I'll test it myself in the next few days. You

Re: [Mesa-dev] [PATCH 1/9] gallium: remove pipe_driver_query_group_info field type

On 11/13/2015 04:57 PM, Nicolai Hähnle wrote: This was only used to implement an unnecessarily restrictive interpretation of the spec of AMD_performance_monitor. The spec says A performance monitor consists of a number of hardware and software counters that can be sampled by the GPU and

Re: [Mesa-dev] [PATCH 5/9] st/mesa: use BITSET_FOREACH_SET to loop through active perfmon counters

Reviewed-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> On 11/13/2015 04:57 PM, Nicolai Hähnle wrote: --- src/mesa/state_tracker/st_cb_perfmon.c | 7 ++- 1 file changed, 2 insertions(+), 5 deletions(-) diff --git a/src/mesa/state_tracker/st_cb_perfmon.c b/src/mesa/state_t

Re: [Mesa-dev] [PATCH 3/9] st/mesa: map semantic driver query types to underlying type

On 11/13/2015 07:22 PM, Nicolai Hähnle wrote: On 13.11.2015 18:34, Samuel Pitoiset wrote: On 11/13/2015 04:57 PM, Nicolai Hähnle wrote: --- src/gallium/include/pipe/p_defines.h | 2 ++ src/mesa/state_tracker/st_cb_perfmon.c | 3 +++ 2 files changed, 5 insertions(+) diff --git a/src

Re: [Mesa-dev] [PATCH 0/9] gallium: batch query objects and related cleanups

On 11/13/2015 07:29 PM, Nicolai Hähnle wrote: Hi Samuel, thanks for taking a look! On 13.11.2015 18:35, Samuel Pitoiset wrote: Did you run amd_performance_monitor piglit tests to make sure all of your changes didn't break anything? Yes, everything passes here. Looks good. Thanks

Re: [Mesa-dev] [PATCH 1/9] gallium: remove pipe_driver_query_group_info field type

On 11/13/2015 07:23 PM, Nicolai Hähnle wrote: On 13.11.2015 18:35, Samuel Pitoiset wrote: On 11/13/2015 04:57 PM, Nicolai Hähnle wrote: This was only used to implement an unnecessarily restrictive interpretation of the spec of AMD_performance_monitor. The spec says A performance monitor

Re: [Mesa-dev] [PATCH 1/9] gallium: remove pipe_driver_query_group_info field type

On 11/13/2015 07:27 PM, Ilia Mirkin wrote: On Fri, Nov 13, 2015 at 1:23 PM, Nicolai Hähnle wrote: So really, this is a question for everybody who cares about nouveau, because nouveau is the only driver that (if a #define is enabled) advertises a CPU driver_query_group.

Re: [Mesa-dev] llvm TGSI backend (WIP) questions

On 11/13/2015 02:46 PM, Hans de Goede wrote: Hi All, Hey Hans, So as discussed I've started working on a TGSI backend for llvm to use as a way to get compute going on nouveau (and other gpu-s). I'm still learning all the ins and outs of llvm so I do not have much to show yet. I've

Re: [Mesa-dev] [PATCH v2 1/9] gallium: remove pipe_driver_query_group_info field type

There is still one assignment to 'type' in nvc0_query.c at line 246 which needs to be removed. On 11/13/2015 08:17 PM, Nicolai Hähnle wrote: This was only used to implement an unnecessarily restrictive interpretation of the spec of AMD_performance_monitor. The spec says A performance

Re: [Mesa-dev] [PATCH v2 7/9] gallium: add the concept of batch queries

You'll need to update this patch and set 'info->flags = 0' in nv50_query.c as well. I pushed some performance counters for nv50 yesterday. Thanks for doing this change. On 11/13/2015 08:17 PM, Nicolai Hähnle wrote: Some drivers (in particular radeon[si], but also freedreno judging from a

Re: [Mesa-dev] [PATCH v2 8/9] gallium/hud: add support for batch queries

One minor comment below. On 11/13/2015 08:17 PM, Nicolai Hähnle wrote: v2: be more defensive about allocations --- src/gallium/auxiliary/hud/hud_context.c | 24 ++- src/gallium/auxiliary/hud/hud_driver_query.c | 265 +++ src/gallium/auxiliary/hud/hud_private.h

Re: [Mesa-dev] [PATCH v2 9/9] st/mesa: add support for batch driver queries to perfmon

See my comment below. On 11/13/2015 08:17 PM, Nicolai Hähnle wrote: v2: forgot a null-pointer check (spotted by Samuel Pitoiset) --- src/mesa/state_tracker/st_cb_perfmon.c | 78 +++--- src/mesa/state_tracker/st_cb_perfmon.h | 6 +++ 2 files changed, 77

Re: [Mesa-dev] [PATCH v2 6/9] st/mesa: maintain active perfmon counters in an array

Reviewed-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> On 11/13/2015 08:17 PM, Nicolai Hähnle wrote: It is easy enough to pre-determine the required size, and arrays are generally better behaved especially when they get large. v2: make sure init_perf_monitor returns true when no co

Re: [Mesa-dev] [PATCH v2 0/9] gallium: batch query objects and related cleanups

On 11/13/2015 08:17 PM, Nicolai Hähnle wrote: Hi, I have updated patches 6 - 9. Samuel, thank you for your input and I hope you find your points to be resolved satisfactorily ;) Yeah, I have no strong objections to remove those groups of GPU counters. :-) Cheers, Nicolai ---

Re: [Mesa-dev] [PATCH] nvc0: fix wrong value for NVC8_COMPUTE_CLASS

On 11/16/2015 11:55 AM, Emil Velikov wrote: On 9 October 2015 at 14:10, Samuel Pitoiset <samuel.pitoi...@gmail.com> wrote: Compute class value for GF110+ is 0x91c0 and not 0x92c0. This fixes compute support and MP performance counters on GF110. Signed-off-by: Samuel Pitoiset <sam

Re: [Mesa-dev] [PATCH 1/5] nv50: implement a basic compute support

On 11/16/2015 11:47 AM, Emil Velikov wrote: On 16 November 2015 at 10:39, Emil Velikov <emil.l.veli...@gmail.com> wrote: Hi Samuel, On 13 November 2015 at 00:04, Samuel Pitoiset <samuel.pitoi...@gmail.com> wrote: This adds the ability to launch simple compute kernels like the on

[Mesa-dev] [PATCH 1/2] nvc0: handle NULL pointer in nvc0_get_compute_param()

To get the size (in bytes) of a compute parameter, clover first calls get_compute_param() with a NULL data pointer. The RET() macro is based on nv50. Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> --- src/gallium/drivers/nouveau/nvc0/nvc0_screen.c | 45 ---

[Mesa-dev] [PATCH 2/2] nvc0: add missing compute parameters required by clover

This fixes crashes with some piglit OpenCL tests. Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> --- src/gallium/drivers/nouveau/nvc0/nvc0_screen.c | 11 ++- 1 file changed, 10 insertions(+), 1 deletion(-) diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c

Re: [Mesa-dev] [PATCH 1/2] nvc0: handle NULL pointer in nvc0_get_compute_param()

On 11/03/2015 07:26 PM, Ilia Mirkin wrote: On Tue, Nov 3, 2015 at 1:35 PM, Samuel Pitoiset <samuel.pitoi...@gmail.com> wrote: To get the size (in bytes) of a compute parameter, clover first calls get_compute_param() with a NULL data pointer. The RET() macro is based on nv50. Sign

[Mesa-dev] [PATCH v2 1/2] nvc0: handle NULL pointer in nvc0_get_compute_param()

To get the size (in bytes) of a compute parameter, clover first calls get_compute_param() with a NULL data pointer. The RET() macro is based on nv50. Changes since v2: - get rid of ul suffixes when they are unnecessary Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> --- src/g

[Mesa-dev] [PATCH v2 2/2] nvc0: add missing compute parameters required by clover

This fixes crashes with some piglit OpenCL tests. Changes since v2: - get rid of ul suffixes when they are unnecessary Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> --- src/gallium/drivers/nouveau/nvc0/nvc0_screen.c | 11 ++- 1 file changed, 10 insertions(+), 1 de

Re: [Mesa-dev] [PATCH] gallium/hud: document GALLIUM_HUD_PERIOD in envvars.html.

Reviewed-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> On 11/04/2015 06:24 AM, Jimmy Berry wrote: --- docs/envvars.html | 2 ++ 1 file changed, 2 insertions(+) diff --git a/docs/envvars.html b/docs/envvars.html index bdfe999..173c941 100644 --- a/docs/envvars.html +++ b/docs/envvar

Re: [Mesa-dev] [PATCH v2] gallium/hud: control visibility at startup and runtime.

Hi Jimmy, Some comments below. On 11/04/2015 06:17 AM, Jimmy Berry wrote: - env GALLIUM_HUD_VISIBLE: control default visibility - env GALLIUM_HUD_SIGNAL_TOGGLE: toggle visibility via signal --- Thanks for the feedback. I believe all the suggested changes have been implemented. One note, all

Re: [Mesa-dev] [PATCH] nvc0: enable compute support on Fermi

On 11/06/2015 12:43 AM, Ilia Mirkin wrote: On Thu, Nov 5, 2015 at 6:41 PM, Samuel Pitoiset <samuel.pitoi...@gmail.com> wrote: Altough the compute support is still not complete because textures and surfaces need to be implemented, it allows to launch very simple compute kernel like one

[Mesa-dev] [PATCH] nvc0: enable compute support on Fermi

Altough the compute support is still not complete because textures and surfaces need to be implemented, it allows to launch very simple compute kernel like one which reads reading MP performance counters. This turns on PIPE_CAP_COMPUTE and PIPE_SHADER_COMPUTE. Signed-off-by: Samuel Pitoiset

Re: [Mesa-dev] [PATCH] nvc0: enable compute support on Fermi

On 11/06/2015 11:23 AM, Hans de Goede wrote: Hi, On 06-11-15 00:51, Samuel Pitoiset wrote: On 11/06/2015 12:43 AM, Ilia Mirkin wrote: On Thu, Nov 5, 2015 at 6:41 PM, Samuel Pitoiset <samuel.pitoi...@gmail.com> wrote: Altough the compute support is still not complete because te

Re: [Mesa-dev] [PATCH 07/16] nvc0: fix queries which use multiple MP counters on Fermi

he second one. Now, I check if the slot is free before pushing the configuration through the pushbuf. On Fri, Oct 16, 2015 at 1:22 PM, Samuel Pitoiset <samuel.pitoi...@gmail.com> wrote: Queries which use more than one MP counters was misconfigured and computing the final result was als

Re: [Mesa-dev] [PATCH 16/16] nvc0: add a note about MP counters on GF100/GF110

, 2015 at 1:22 PM, Samuel Pitoiset <samuel.pitoi...@gmail.com> wrote: MP counters on GF100/GF110 (compute capability 2.0) are buggy because there is a context-switch problem that we need to fix. Results might be wrong sometimes, be careful! Signed-off-by: Samuel Pitoiset <samuel.pit

[Mesa-dev] [PATCH 00/16] nvc0: improve MP counters support on Fermi

with piglit. I'll submit an other series in the next few days which adds some performance monitoring metrics on Fermi and I'll double check MP counters on Kepler. Thanks. Samuel Pitoiset (16): nvc0: remove useless call to query_get_cfg() in nvc0_hw_sm_query_end() nvc0: split out begin_query() hook

[Mesa-dev] [PATCH 12/16] nvc0: allow only one active query for the MP counters group

by AMD_performance_monitor. Like for Kepler, this limits the maximum number of active queries to 1 on Fermi. Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> --- src/gallium/drivers/nouveau/nvc0/nvc0_query.c | 20 +--- 1 file changed, 9 insertions(+), 11 deletions(-) diff --git

[Mesa-dev] [PATCH 16/16] nvc0: add a note about MP counters on GF100/GF110

MP counters on GF100/GF110 (compute capability 2.0) are buggy because there is a context-switch problem that we need to fix. Results might be wrong sometimes, be careful! Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> --- src/gallium/drivers/nouveau/nvc0/nvc0_query_hw_sm

[Mesa-dev] [PATCH 15/16] nvc0: add MP counters variants for GF100/GF110

GF100 and GF110 chipsets are compute capability 2.0, while the other Fermi chipsets are compute capability 2.1. That's why, some MP counters are different between these chipsets and we need to handle variants. Signed-off-by: Samuel Pitoiet ---

[Mesa-dev] [PATCH 11/16] nvc0: read MP counters of all GPCs on Fermi

performance counters of all TPCs. Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> --- src/gallium/drivers/nouveau/nvc0/nvc0_query_hw_sm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_query_hw_sm.c b/src/gallium/drivers/nouvea

<    1   2   3   4   5   6   7   8   9   10   >