Am Freitag, den 14.12.2018, 12:38 +0100 schrieb Samuel Pitoiset:
>
> On 12/13/18 9:27 PM, Gert Wollny wrote:
> > IMHO allowing MRs is a good thing, so
> > Acked-by: Gert Wollny
> >
>
> Allowing MRs isn't a bad thing. The main problem IMHO is that now we
&g
ntrol.
framebuffer_srgb_enabled*
pass (with MESA_GLES_VERSION_OVERRIDE=3.2).
Best,
Gert
Am Donnerstag, den 15.11.2018, 13:45 +0100 schrieb Gert Wollny:
> From: Gert Wollny
>
> Dear all,
>
> after the RFC and Ilias comments I reworked the series another
> time.
&g
Since Meson will eventually be the only build system deprecate autotools
now. It can still be used by invoking configure with the flag
--enable-autotools
Signed-off-by: Gert Wollny
---
IMO autotools should be properly deprecated prior it its removal, so here
is a patch to do just that. I
Am Freitag, den 14.12.2018, 13:11 +0100 schrieb Gert Wollny:
> Am Freitag, den 14.12.2018, 01:19 -0500 schrieb Ilia Mirkin:
> > I have to say that the user experience for autotools is WAY better
> > than for meson. As a concrete example, I had a meson build. Then I
> > upd
Reviewed-By: Gert Wollny
Am Donnerstag, den 06.12.2018, 15:00 +0100 schrieb Nicolai Hähnle:
> From: Nicolai Hähnle
>
> ---
> src/gallium/drivers/r600/sb/sb_ir.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/src/gallium/drivers/r600/sb/sb
Hi Ilia,
Am Sonntag, den 16.12.2018, 12:40 -0500 schrieb Ilia Mirkin:
> On Sun, Dec 16, 2018 at 6:24 AM Gert Wollny
> wrote:
> >
> > Since Meson will eventually be the only build system deprecate
> > autotools
> > now. It can still be used by invoking configur
Am Mittwoch, den 19.12.2018, 12:19 -0500 schrieb Ilia Mirkin:
> On Sun, Dec 16, 2018 at 6:24 AM Gert Wollny
> wrote:
> >
> > Since Meson will eventually be the only build system deprecate
> > autotools
> > now. It can still be used by invoking configure with the
Hello all,
With the last (minor) update I moved the series to an MR:
https://gitlab.freedesktop.org/mesa/mesa/merge_requests/14
There are only one minor change from the series last posted here
v7: use _mesa_has_EXT_framebuffer_sRGB instead of testing the extension
flag directly.
v6: - rename
Hello Stuart,
Am Donnerstag, den 20.12.2018, 13:40 +1100 schrieb Stuart Young:
> Gert/Ilia,
>
> Could this be reduced this from an error to a warning, with the
> command-line option suppressing the warning?
>
> Perhaps as well as producing the warning, the build could sleep for
> say 30 seconds
Looks good to me,
Reviewed-By: Gert Wollny
Am Donnerstag, den 27.12.2018, 16:11 +1000 schrieb Dave Airlie:
> From: Dave Airlie
>
> Older versions of virglrenderer before
> 33da7361aec486290df0aec4ad8dfa8ff6adde2c
> in vtest mode, misrender gears.
>
> Fixes: 9d81cd8e7c
https://gitlab.freedesktop.org/mesa/mesa/merge_requests/58
The extension NV_depth_clamp is written against OpenGL 1.2.1, and since
GLES 2.0 is based on GL 2.0 there is no reason not to enable this
extension also for GLES >= 2.0.
Best,
Gert
___
mesa-de
Tested it, looks fine,
Reviewed-By: Gert Wollny
Am Freitag, den 28.12.2018, 16:54 +1000 schrieb Dave Airlie:
> From: Dave Airlie
>
> This stores the raster state and calls the correct primconvert
> interface
> using the currently bound raster state.
>
> This fixes pigl
Am Donnerstag, den 03.01.2019, 14:28 +0100 schrieb Gert Wollny:
> Tested it, looks fine,
> Reviewed-By: Gert Wollny
Sorry, that should be
>
> Am Freitag, den 28.12.2018, 16:54 +1000 schrieb Dave Airlie:
> > From: Dave Airlie
> >
> > This stores the rast
Am Montag, den 19.02.2018, 14:06 +1000 schrieb Dave Airlie:
> On 15 February 2018 at 01:26, Gert Wollny
> wrote:
> > Am Mittwoch, den 14.02.2018, 17:18 +1000 schrieb Dave Airlie:
> > > From: Dave Airlie
> > >
> > > A bunch of CTS tests led me to writ
ALU_EXTENDED needs 4 DWORDS instead of the usual 2, hence if the last ALU
clause within a IF-JUMP or ELSE branch is ALU_EXTENDED the target jump
offset needs to be adjusted accordingly.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104654
Signed-off-by: Gert Wollny
---
* An additional
Limit the length of acceptable cpu names for use in hud_get_num_cpufreq
in order to avoid a buffer overflow later in add_object when this name
is copied into cpufreq_info::name.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105274
Signed-off-by: Gert Wollny
---
Submitter has no git
For me this also fixed
tests/spec/arb_compute_shader/execution/
multiple-texture-reading.shader_test
With that Tested-by: Gert Wollny
Am Mittwoch, den 28.02.2018, 16:47 +1000 schrieb Dave Airlie:
> From: Dave Airlie
>
> The delayed loading code was fail if we had con
's
> make Meson find our wanted version sooner than the one at
> /usr/local/bin
>
> Once this is corrected, we would still need a patch similar to:
> https://lists.freedesktop.org/archives/mesa-dev/2017-December/180217.
> html
>
> Cc: Eric Engestrom
> Cc: Dylan Baker
&
Looks fine to me,
Reviewed-By: Gert Wollny
Am Mittwoch, den 28.02.2018, 23:18 +0200 schrieb Andres Gomez:
> Travis CI has moved to LLVM 5.0, and meson is detecting automatically
> the available version in /usr/local/bin based on the PATH env
> variable
> order preference.
>
Am Dienstag, den 06.03.2018, 01:10 +0100 schrieb Germano Massullo:
>
> Question: is LIBGL_DRIVERS_PATH enough to let Firefox use a different
> Mesa version?
You will also need to set LIBGL_DRIVERS_PATH
best,
Gert
> I am asking this because on the console I am getting
>
> libGL error: unable
Am Dienstag, den 06.03.2018, 13:10 +0100 schrieb Germano Massullo:
> Il 06/03/2018 09:06, Gert Wollny ha scritto:
> > Am Dienstag, den 06.03.2018, 01:10 +0100 schrieb Germano Massullo:
> > > Question: is LIBGL_DRIVERS_PATH enough to let Firefox use a
> > > d
detected that t is only read in the
"if (a)" scope that encloses the "if (b)/else" constructs where t is written
in both branches.
This patch corrects the minimal live range estimation and adds an according
unit test for this case.
Signed-off-by: Gert Wollny
--
64, and I get 2976 of 2995 test
pass, skip: 16, fail: 3 (like Elie pointed out: ssbo and arrays of
arrays are the culprits).
I'm wonderig a bit what is the subset that gives you 1375 piglits ...
In any case: Tested-By: Gert Wollny
>
> This patchset has all the glsl lowering code. (
Am Mittwoch, den 14.03.2018, 21:05 +1100 schrieb Timothy Arceri:
>
>
> Anyway I'm not trying to convince Dave as I'm sure he has a good idea
> of the work involve. Just thought I'd mention this stuff in case
> anyone else has considered working on NIR for r600 and are scared by
> Daves time esti
Am Freitag, den 16.03.2018, 12:42 +0100 schrieb Juan A. Suarez Romero:
>
> Gert Wollny (1):
> c7cadcbda4 r600: Take ALU_EXTENDED into account when evaluating
> jump offsets
> Reason: The commit requires earlier commit a03d456f5a41 which did not
> land in branch.
Th
Don't allocate a zero-sized array, when no texture offsets are given.
Reviewed-by: Nicolai Hähnle
Signed-off-by: Gert Wollny
---
.../tests/test_glsl_to_tgsi_lifetime.cpp | 23 +++---
1 file changed, 16 insertions(+), 7 deletions(-)
diff --git a/src/mesa/state_tr
---
src/mesa/program/prog_print.c | 16 ++--
1 file changed, 14 insertions(+), 2 deletions(-)
diff --git a/src/mesa/program/prog_print.c b/src/mesa/program/prog_print.c
index 9462510f3a..b273bbf4c8 100644
--- a/src/mesa/program/prog_print.c
+++ b/src/mesa/program/prog_print.c
@@ -50,1
ss to mesa-git.
Gert Wollny (10):
mesa/st/tests: Fix zero-byte allocation leaks
mesa/st/tests: unify MockCodeLine* classes
mesa/st/tests: base check of number of registers on opcode info
mesa/st/tests: cleanup whitespace usage and correct some comments
mesa/st/glsl_to_tgsi: Correct
Add the equal operator and the "<<" stream write operator for the
st_*_reg classes and the "<<" operator to the instruction class, and
make use of these operators in the debugging output.
---
src/mesa/state_tracker/st_glsl_to_tgsi_private.cpp | 168 +
src/mesa/state_tracker/st_
Improve the life-time evaluation of temporary registers by also tracking
writes in both if and else branches and in up to 32 nested scopes.
As a result the estimated required register life-times can be further
reduced enabling more registers to be merged.
Signed-off-by: Gert Wollny
Don't allocate a zero-sized array, when no texture offsets are given.
v5: correct spaces and empty lines
Reviewed-by: Brian Paul (v4)
Reviewed-by: Nicolai Hähnle (v1)
Signed-off-by: Gert Wollny
---
.../tests/test_glsl_to_tgsi_lifetime.cpp | 22 +++---
1 file ch
Reviewed-by: Brian Paul
Signed-off-by: Gert Wollny
---
src/mesa/program/prog_print.c | 16 ++--
1 file changed, 14 insertions(+), 2 deletions(-)
diff --git a/src/mesa/program/prog_print.c b/src/mesa/program/prog_print.c
index 9462510f3a..b273bbf4c8 100644
--- a/src/mesa/program
Add the equal operator and the "<<" stream write operator for the
st_*_reg classes and the "<<" operator to the instruction class, and
make use of these operators in the debugging output.
v5: Fix empty lines
Reviewed-by: Brian Paul (v4)
Signed-off-by: Gert Wol
* Base assertion of correct number of src and dst registers in tests
on what the operatand actually expects
* Fix number of destinations in one test
Signed-off-by: Gert Wollny
---
src/mesa/state_tracker/tests/Makefile.am | 7 +
src/mesa/state_tracker/tests/st_tests_common.cpp | 397
r the use of address
registers
in register_merge (I was considering to add this to 8, but since that one was
already
reviewed ...)
many thanks for any comments,
Gert
PS: I have no write access to mesa-git.
Gert Wollny (10):
mesa/program: Add missing file types to printout
mesa/st/gls
: Gert Wollny
---
.../state_tracker/st_glsl_to_tgsi_temprename.cpp | 312 +++--
1 file changed, 292 insertions(+), 20 deletions(-)
diff --git a/src/mesa/state_tracker/st_glsl_to_tgsi_temprename.cpp
b/src/mesa/state_tracker/st_glsl_to_tgsi_temprename.cpp
index 509342e53f
Signed-off-by: Gert Wollny
---
.../tests/test_glsl_to_tgsi_lifetime.cpp | 127 ++---
1 file changed, 63 insertions(+), 64 deletions(-)
diff --git a/src/mesa/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
b/src/mesa/state_tracker/tests
Add a code line type that accepts one layer of indirect addressing and
add tests to check that temporary register access used for indirect
addressing is accounted for in the lifetime estimation.
Signed-off-by: Gert Wollny
---
src/mesa/state_tracker/tests/st_tests_common.cpp | 95
Additional teste are added that check the tracking of access to temporaries
in if-else branches.
Signed-off-by: Gert Wollny
---
.../tests/test_glsl_to_tgsi_lifetime.cpp | 493 -
1 file changed, 486 insertions(+), 7 deletions(-)
diff --git a/src/mesa/state_tracker
lifetime in the cases when the load to the address register was emitted.
v2: no changes
v3: Use debug_log variable instead of directly writing to std::err in debugging
output.
Reviewed-by: Nicolai Hähnle (v1)
Signed-off-by: Gert Wollny
---
.../state_tracker/st_glsl_to_tgsi_temprename.cpp
With the implementation of the tracking of the registers used in reladdr
asserting that a driver calling merge_register() uses the address register
is no longer needed.
Signed-off-by: Gert Wollny
---
src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 1 -
1 file changed, 1 deletion(-)
diff --git a
On 26.01.2018 01:56, Timothy Arceri wrote:
> ---
> src/mesa/main/program_binary.c | 14 ++
> 1 file changed, 14 insertions(+)
>
> diff --git a/src/mesa/main/program_binary.c b/src/mesa/main/program_binary.c
> index 2786487362..68a15ec258 100644
> --- a/src/mesa/main/program_binary.c
sb optimized away the ELSE in a construct like
while (foo) {
if (bar ) {
do something;
} else {
break;
}
}
resulting in
while (foo) {
if (bar ) {
do something;
break;
}
}
which is obviously wrong.
With this patch an
: 807e2539e512ca6c96f059da855473eb7be99ba1
mesa/st/glsl_to_tgsi: Add tracking of ifelse writes in register merging
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104803
Signed-off-by: Gert Wollny
---
.../state_tracker/st_glsl_to_tgsi_temprename.cpp | 24 +-
.../tests/test_glsl_to_tgsi_lifetime.cpp
Am Montag, den 29.01.2018, 20:32 +0100 schrieb Roland Scheidegger:
>
> Am I correct assuming that for something like
>while (foo) {
> if (bar) {
> do something;
> } else {
> /* nothing */
> }
>}
> The else clause wouldn't get optimized away neither?
> (T
with sb and pass without it are
differnt issues I'll open separate bugs for these, so that this one can
savely be closes when the patch lands.
Tested-By: Gert Wollny
Best,
Gert
Am Dienstag, den 30.01.2018, 16:40 +1000 schrieb Dave Airlie:
> From: Dave Airlie
>
> If there is
@arb_gpu_shader5@execution@samplemaskin-indirect for
debug builds (it already passed in release).
Signed-off-by: Gert Wollny
---
src/gallium/drivers/r600/sb/sb_bc_parser.cpp | 8 +++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/src/gallium/drivers/r600/sb/sb_bc_parser.cpp
b
Am Montag, den 29.01.2018, 13:42 -0700 schrieb Brian Paul:
> ---
> src/mesa/main/mtypes.h | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
> index b0eeeaf..95f6319 100644
> --- a/src/mesa/main/mtypes.h
> +++ b/src/mesa/main
ding *binding = &vao-
> >BufferBinding[index];
This conflicts with d5f42f96e16f4b12, you added an asset there.
With that patches 1-3 and 5-7 are
Reviewed-by: Gert Wollny
> diff --git a/src/mesa/main/varray.c b/src/mesa/main/varray.c
> index 6c022b4..b2783e2 100644
> --- a/s
Am Dienstag, den 30.01.2018, 08:26 -0700 schrieb Brian Paul:
> On 01/30/2018 01:38 AM, Gert Wollny wrote:
> > Am Montag, den 29.01.2018, 13:42 -0700 schrieb Brian Paul:
> > > Slightly better readbility.
> > > ---
> > > src/mesa/main/arrayobj.c | 2 +-
&
Am Dienstag, den 30.01.2018, 10:14 -0700 schrieb Brian Paul:
> On 01/30/2018 09:59 AM, Mark Janes wrote:
> > This series that was just pushed to master generates ~200
> > assertions on
> > all platforms, eg:
> >
> > piglit.spec.ext_framebuffer_multisample.formats 2
> >
> > ext_framebuffer_multisa
Am Freitag, den 02.02.2018, 02:13 +0100 schrieb Roland Scheidegger:
> Am 30.01.2018 um 09:21 schrieb Gert Wollny:
> > Although gl_SampleMaskIn is declared as an array in GLSL, it is
> > effectively a 32 bit mask on all hardware supported by mesa, so the
> > array indexing is
Am Freitag, den 02.02.2018, 09:04 +0100 schrieb Roland Scheidegger:
>
>
> Yes, the _GL spec_ says it is an array.
> But in gallium it can't be. Therefore I think it's incorrect if we
> end up with array accesses there (albeit I was too lazy to actually
> look at the tgsi, but I'm pretty sure it i
Am Freitag, den 02.02.2018, 06:56 -0500 schrieb Ilia Mirkin:
> On Fri, Feb 2, 2018 at 4:07 AM, Gert Wollny
> wrote:
> > Am Freitag, den 02.02.2018, 09:04 +0100 schrieb Roland Scheidegger:
> > >
> > >
> > > Yes, the _GL spec_ says it is an array.
> &
Thanks, my automake knowledge is a bit rusty so that I didn't know
about this option to limit the build to "make check".
Reviewed-By: Gert Wollny
Am Freitag, den 02.02.2018, 16:59 +0100 schrieb Bas Nieuwenhuizen:
> We don't need the library if we don't build tes
I like that,
Reviewed-By: Gert Wollny
Am Mittwoch, den 07.02.2018, 06:25 +1000 schrieb Dave Airlie:
> From: Glenn Kennard
>
> This is taken from Glenn Kennards scratch series, but separated
> out as a cleanup by me.
>
> Signed-off-by: Dave Airlie
> ---
>
also forbitten.
Test for these cases and reject the optimization if a forbidden combination
of relative access would be created.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=103142
Signed-off-by: Gert Wollny
---
I folded both cases into one patch because they both are required to fix
/show_bug.cgi?id=103142
Signed-off-by: Gert Wollny
---
src/gallium/drivers/r600/sb/sb_expr.cpp | 17 ++
src/gallium/drivers/r600/sb/sb_ir.h | 6 +
src/gallium/drivers/r600/sb/sb_valtable.cpp | 36 +
3 files changed, 55 insertions(+), 4 deletions
corresponds to the value 0 used
before.
Fixes 9f9ce1625fb32ff0b9d5b1858aa5bb7b531ded0f
st/mesa: use TGSI enum types in st_glsl_to_tgsi.cpp
Signed-off-by: Gert Wollny
---
src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/mes
and make the variable any type names
self-explaining, but since I've been staring at this code for a long time I
don't think I am capable of seeing any problems any more, so comments are very
welcome.
Best,
Gert
PS: I have no git write access.
Gert Wollny (11):
mesa/st: gls
This class implements the live range merge and component interleave
remapping logic.
Signed-off-by: Gert Wollny
---
src/mesa/Makefile.sources | 2 +
src/mesa/meson.build | 2 +
.../state_tracker/st_glsl_to_tgsi_array_merge.cpp | 283
On one hand "live range" is the term used in the literature, and on the
other hand a distinction is needed from the array live ranges that will
be introduced later.
Signed-off-by: Gert Wollny
---
src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 16 ++---
.../sta
Implememt a class that holds the information required by the array merging
and interleave algorithm, namely array ID, live range, access mask,
accessed components, and the number of accessed components.
Signed-off-by: Gert Wollny
---
.../state_tracker/st_glsl_to_tgsi_array_merge.cpp | 69
Signed-off-by: Gert Wollny
---
src/mesa/state_tracker/tests/Makefile.am | 20 +-
src/mesa/state_tracker/tests/st_tests_common.h | 7 +-
.../tests/test_glsl_to_tgsi_array_merge.cpp| 296 +
3 files changed, 317 insertions(+), 6 deletions(-)
create mode
Add the functions that evaluate array live range merging and array
interleaving based on the array information.
Signed-off-by: Gert Wollny
---
.../state_tracker/st_glsl_to_tgsi_array_merge.cpp | 385 -
.../state_tracker/st_glsl_to_tgsi_array_merge.h| 26 +-
2 files
to split_arrays
* unify the tracking and remapping methods for src and st registers
* also track access to arrays via reladdr*
Signed-off-by: Gert Wollny
---
src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 107 +
1 file changed, 107 insertions(+)
diff --git a/src/mes
available,
so some are added as a placeholder in get_temp_registers_required_live_ranges.
Signed-off-by: Gert Wollny
---
.../state_tracker/st_glsl_to_tgsi_temprename.cpp | 61 +-
1 file changed, 49 insertions(+), 12 deletions(-)
diff --git a/src/mesa/state_tracker
Add a class implementing the (rather simplified) live range evaluation for
arrays.
Remark: Since the class is local and not yet used, the compiler will emit
some warnings about unused functions.
Signed-off-by: Gert Wollny
---
.../state_tracker/st_glsl_to_tgsi_temprename.cpp | 102
* rename access_record to register_merge_record because the name is more
appropriate and we need to distinguish from the array access introduced
later.
* rename member variable acc to temp_acc
* make comparison operator in register_merge_record const
Signed-off-by: Gert Wollny
In preparation of the array live range tracking the evaluation of the read
mask is moved out the register live range tracking to the enclosing call
of the generalized read access tracking.
Signed-off-by: Gert Wollny
---
src/mesa/state_tracker/st_glsl_to_tgsi_temprename.cpp | 15
-multiple-large-local-arrays
on r600/barts unneccessary.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=100200
Signed-off-by: Gert Wollny
---
src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 25 +-
.../state_tracker/st_glsl_to_tgsi_temprename.cpp | 13
Am Mittwoch, den 14.02.2018, 17:18 +1000 schrieb Dave Airlie:
> From: Dave Airlie
>
> A bunch of CTS tests led me to write
> tests/shaders/ssa/fs-while-loop-rotate-value.shader_test
> which r600/sb always fell over on.
>
> This patch fixes it, but I'll probably never be 100% sure why.
Unfortuna
Am Freitag, den 16.02.2018, 14:44 +1000 schrieb Dave Airlie:
> From: Dave Airlie
>
> A bunch of CTS tests led me to write
> tests/shaders/ssa/fs-while-loop-rotate-value.shader_test
> which r600/sb always fell over on.
>
> GCM seems to move some of the copys into other basic blocks,
> if we don't
Am Freitag, den 18.08.2017, 09:43 +1000 schrieb Timothy Arceri:
> This fixes an assert during IR validation in LLVMpipe.
>
> Fixes: e2e2c5abd279 (glsl: calculate number of operands in an
> expression once)
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=102274
> Cc: Brian Paul
> ---
> s
Am Freitag, den 18.08.2017, 18:25 +0200 schrieb Marek Olšák:
>
> Sure. The changes are here:
> https://cgit.freedesktop.org/~mareko/mesa/log/?h=blitter-optz
>
> If somebody wants to test it on r600g, that would be appreciated too.
>
>
I did this and unfortunately it crashes with 744c3e2254 appl
Am Samstag, den 19.08.2017, 12:13 +0200 schrieb Gert Wollny:
>
> I did this and unfortunately it crashes with 744c3e2254 applied.
> Specifically, in
>
> src/gallium/drivers/r600/evergreen_state.c2302:
> evergreen_emit_vertex_fetch_shader
>
> de-referenc
Am Samstag, den 19.08.2017, 18:35 +0200 schrieb Marek Olšák:
> Can you try the attached patch?
While if fixes the crash and lets me run glxgears and some other rathe
simple OpenGL programs it completely messes up blender (menu unreadable
etc) and UE4Editor Window is also flickering.
Best,
Gert
Am Sonntag, den 20.08.2017, 15:48 +0200 schrieb Marek Olšák:
> The patch seems OK to me.
Hmm,
you're right in that I was wrong about "program", I still need learn to
think in patches.
However, when I look at si_delete_compute_state where
si_compute_reference is called:
>+ si_compute_refe
Am Montag, den 21.08.2017, 10:22 +1000 schrieb Dave Airlie:
> Hi Gert,
>
> Can you test this along with the fetch shader patch Marek sent?
This patch fixes the flickering in Blender and UE4Editor.
Not applying Marks fetch patch results in a segfault in glxgears.
I've also tested with the han
Hello Elie,
thanks for you work on this. I'd thought I'd share my test results on
BARTS:
- your tree was build with debug enabled, so some of the crashes
relate to assertion failures (at least one in r600/sb which also
fails with R600_DEBUG=nosb).
- piglit results
* arb_gpu_shader_f
From: Dave Airlie
Instead of having to search the whole array, just use the whole
thing and store a valid bit in there with the rename.
Removes this from the profile on some of the fp64 tests
Reviewed-by: Timothy Arceri
Signed-off-by: Dave Airlie
---
src/mesa/state_tracker/st_glsl_to_tgsi.cp
ed that the register lifetime estimation
will bailout and signal that no renaming can take place (because
subroutines
are not tracked),
* add array-id to debug output for PROGRAM_ARRAY,
* rebase to 39f3e2507c6
Dave Airlie (1):
st_glsl_to_tgsi: rewrite rename registers to use a
gt;info->num_src - 1 : op->info->num_src;
-}
glsl_to_tgsi_instruction *
glsl_to_tgsi_visitor::emit_asm(ir_instruction *ir, unsigned op,
diff --git a/src/mesa/state_tracker/st_glsl_to_tgsi_private.cpp
b/src/mesa/state_tracker/st_glsl_to_tgsi_private.cpp
new file mode 100644
index 000
/mesa/state_tracker/st_glsl_to_tgsi_temprename.cpp
new file mode 100644
index 00..9690e47fd6
--- /dev/null
+++ b/src/mesa/state_tracker/st_glsl_to_tgsi_temprename.cpp
@@ -0,0 +1,886 @@
+/*
+ * Copyright © 2017 Gert Wollny
+ *
+ * Permission is hereby granted, free of charge, to any person
The patch adds tests for the register rename mapping evaluation and
combined life time estimation and renaming.
---
.../tests/test_glsl_to_tgsi_lifetime.cpp | 192 +
1 file changed, 192 insertions(+)
diff --git a/src/mesa/state_tracker/tests/test_glsl_to_tgsi_lifetim
This patch replaces the old register lifetime estiamtion and
rename mapping evaluation with the new one.
Performance to compare between the current and the new implementation
were measured by running the shader-db in one thread.
---
The remapping evaluator first sorts the temporary registers ascending
based on their first life time instruction, and then uses a binary search
to find merge canidates.
For the initial sorting it uses std::sort because qsort is quite slow in
comparison. By removing the define USE_STL_SORT in
src/
si_lifetime.cpp
@@ -0,0 +1,1405 @@
+/*
+ * Copyright © 2017 Gert Wollny
+ *
+ * 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 witho
Am Freitag, den 25.08.2017, 10:20 +0900 schrieb Michel Dänzer:
> On 25/08/17 02:38 AM, Gert Wollny wrote:
> >
> > The patch doesn't introduce piglit regression (I tested the shader
> > subset).
>
> I'd recommend testing at least the gpu profile, ideally runn
I've tested this on r600g (piglit/shader and a bunch of games) and
Unigine benchmarks, no problems found.
For the series Tested-by: Gert Wollny
Am Montag, den 28.08.2017, 10:58 +0200 schrieb Nicolai Hähnle:
> Hi all,
>
> This series changes all CONST accesses to be two-di
ice some minor things, I'm about to send out patches to
> clean things up afterwards.
I will test these patches ASAP.
Best,
Gert
>
> Cheers,
> Nicolai
>
>
> On 24.08.2017 19:38, Gert Wollny wrote:
> > Dear all,
> >
> > I thought I might sen
If is included then max is a macro that clashes
with std::numeric_limits::max, hence undefine it.
For some reason the struct access_record is not recognizes
outside the anonymouse namespace, make it a class.
The patch successfully was tested on AppVeyor.
---
src/mesa/state_tracker/st_glsl_to_tgs
For the series
Reviewed-By: Gert Wollny
Am Mittwoch, den 06.09.2017, 11:54 +0200 schrieb Nicolai Hähnle:
> From: Nicolai Hähnle
>
> enclosing_scope already contains enclosing_scope_first_read.
> What we really want to check here -- not for correctness, but
> for spee
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=102565
Add the according flags to link with libunwind.
---
src/mesa/state_tracker/tests/Makefile.am | 1 +
1 file changed, 1 insertion(+)
diff --git a/src/mesa/state_tracker/tests/Makefile.am
b/src/mesa/state_tracker/tests/Makefile.am
index fb
Fixes the build in classic only mode, i.e. the new state tracker tests are
only build when Gallium is enabled.
---
src/mesa/Makefile.am | 7 ++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/src/mesa/Makefile.am b/src/mesa/Makefile.am
index 865735be27..f2097eb209 100644
--- a/sr
Am Freitag, den 08.09.2017, 11:19 +0200 schrieb Nicolai Hähnle:
>
>
> Sorry about the slip-up, I thought you were building with both
> patches. In any case, I made sure to test with the Travis build, so
> it should be all good (not a guarantee of course, but it has pretty
> good coverage).
Actua
Fixes: build failure in travis configuration "make Gallium ST Other"
with libunwind enabled (https://travis-ci.org/gerddie/mesa/jobs/273673597)
---
src/gallium/targets/dri/Makefile.am | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/src/gallium/targets/dri/Makefile.am
b/src/g
configure picks up libunwind automatically if it is available and
links libmesagallium against it. Because static linking is used
all targets that use libmesagallium also need to link against libunwind
in this case, and this should be checked on travis-ci.
---
.travis.yml | 4
1 file changed,
he gallium/make
configuratuons thereby enabling building against it.
Best,
Gert
Gert Wollny (2):
gallium/targets/dri: Add libunwind to linker flags
travis-ci: Add libunwind-dev to gallium/make builds
.travis.yml | 4
src/gallium/targets/dri/Makefile.am | 3 ++-
2 f
Hi Michel,
I can confirm that the problem exists on
8d6b97a815874c7b695471a259fe4b6689890206
i.e. for me ir_expression_operation.h is created in the build
directory - and the patch fixes compiling the test in an out of source
build.
Hence -
Tested-by: Gert Wollny
Am Sonntag, den
501 - 600 of 827 matches
Mail list logo
