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
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
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
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
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
...
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
.
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
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
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
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
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
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
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
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
/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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
:
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
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
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
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
/listinfo/mesa-dev
--
Best regards,
Samuel Pitoiset.
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
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
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
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
?
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>>
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
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 ++
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.
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
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/
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|
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 +-
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
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:
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
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)
+
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
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
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&
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_
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
@@
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
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 --
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
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.
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
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
*/
+ /* 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
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
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
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
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
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
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
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
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.
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
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
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
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
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
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
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
---
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
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
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 ---
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
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
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
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
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
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
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
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
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
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
, 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
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
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
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
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
---
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
201 - 300 of 5019 matches
Mail list logo