On 12/9/20 7:55 AM, Brian Paul wrote:
> On 12/08/2020 05:47 PM, Ian Romanick wrote:
>> On 12/1/20 7:43 AM, Brian Paul wrote:
>>> On 11/26/2020 09:22 AM, Brian Paul wrote:
>>>> On Wed, Nov 25, 2020 at 8:19 PM Ian Romanick >>> <mailto:i...@freedesktop
On 12/1/20 7:43 AM, Brian Paul wrote:
> On 11/26/2020 09:22 AM, Brian Paul wrote:
>> On Wed, Nov 25, 2020 at 8:19 PM Ian Romanick > <mailto:i...@freedesktop.org>> wrote:
>>
>> I was looking at the dlist-beginend subtest test_illegal_begin_mode.
>&g
I was looking at the dlist-beginend subtest test_illegal_begin_mode.
The test compiles a display list with GL_COMPILE_AND_EXECUTE mode. The
glBegin in the test uses an invalid mode. This should generate a
GL_INVALID_OPERATION error immediately. The test checks this, and that
seems correct.
The
----------
> *From:* Ian Romanick
> *Sent:* Monday, July 8, 2019 2:27:16 PM
> *To:* Abel Briggs; piglit@lists.freedesktop.org
> *Subject:* Re: [Piglit] shader_runner: ignoring uniforms in [test]
> header when location cannot be found instead of returning PIGLIT_FAIL
>
> On 7/
On 7/8/19 11:57 AM, Abel Briggs wrote:
> Hi,
>
> Recently, I've been working on scripts for a GLSL fuzz testing framework
> that use piglit's shader_runner to render shaders. As part of the test
> generation process, shaders can be generated that declare uniforms
> without using them (which is
Reviewed-by: Ian Romanick
On 6/25/19 7:15 PM, Dave Airlie wrote:
> From: Dave Airlie
>
> These tests preinit the ssbo contents but expect the driver to
> be doing std430 packing by default, just specify std430 packing.
>
> This fixes the tests on softpipe
> ---
> .../
From: Ian Romanick
These tests preinit the ssbo contents but expect the driver to be doing
std430 packing by default, just specify std430 packing.
Edits done by:
grep -lr ' buffer ' tests/spec/intel_shader_atomic_float_minmax/execution |\
while read f; do
sed --in-place -e 's
Reviewed-by: Ian Romanick
On 6/25/19 7:23 PM, Dave Airlie wrote:
> From: Dave Airlie
>
> These tests preinit the ssbo contents but expect the driver to
> be doing std430 packing by default, just specify std430 packing.
> ---
> .../execution/shared-atomicM
On 4/15/19 5:16 PM, Marek Olšák wrote:
> It's also an execution test for "varying out".
I don't know what that has to do with glBindFragDataLocationEXT vs
glBindFragDataLocation. Can you please explain?
> Marek
>
> On Mon, Apr 15, 2019 at 6:28 PM Ian Romanick <ma
);
> - }
> - else {
> - glBindFragDataLocation(program, 0, "outputFloat");
> - glBindFragDataLocation(program, 1, "outputInt");
> + GLuint program;
> +
> + if (ext_gpu_shader4) {
> + pr
From: Ian Romanick
In Mesa, NIR has an optimization that will convert certain patterns into
instructions that extract an individual byte from a larger integer.
When extracting from a 64-bit value, the i965 compiler backend has some
bugs emitting these instructions. These tests try to reproduce
On 2/22/19 8:25 AM, apinheiro wrote:
>
> On 22/2/19 15:51, Andres Gomez wrote:
>> The property is supported by most of the editors, but not all:
>> https://github.com/editorconfig/editorconfig/wiki/EditorConfig-Properties#max_line_length
>>
>>
>> Cc: Eric Engestrom
>> Cc: Eric Anholt
>>
I made a couple small adjustments to the tests and pushed them.
Thanks!
On 2/15/19 5:43 AM, asimiklit.w...@gmail.com wrote:
> From: Andrii Simiklit
>
> The tests for a compiler which leads to mismatch between max_array_access
> and the var type when the compiler tries to optimize unused ssbo
From: Ian Romanick
tests/texturing/s3tc-errors.c: In function ‘test_format’:
tests/texturing/s3tc-errors.c:175:9: warning: unused variable ‘format2’
[-Wunused-variable]
GLenum format2;
^~~
At top level:
tests/texturing/s3tc-errors.c:155:1: warning: ‘check_gl_error2_’ defined
From: Ian Romanick
tests/fbo/fbo-blending-format-quirks.c: In function ‘piglit_display’:
tests/fbo/fbo-blending-format-quirks.c:157:40: warning: passing argument 2 of
‘test_formats’ discards ‘const’ qualifier from pointer target type
[-Wdiscarded-qualifiers]
result = test_formats(cases[i
On 11/28/18 6:55 PM, Marek Olšák wrote:
> From: Marek Olšák
>
> ---
> tests/opengl.py | 6 +
> tests/spec/CMakeLists.txt | 1 +
> .../CMakeLists.gl.txt | 7 ++
> .../CMakeLists.gles2.txt | 4 +
t;
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104297
> Signed-off-by: Danylo Piliaiev
Reviewed-by: Ian Romanick
> ---
> .../tcs-output-unmatched.shader_test | 67 +++
> 1 file changed, 67 insertions(+)
> create mode 100644
> tests/
On 1/2/19 7:39 AM, Danylo Piliaiev wrote:
> Test that TCS per-vertex outputs which are used only in
> TCS stage are not converted to local variables and indeed
> share data within the patch.
>
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104297
> Signed-off-by: Danylo Piliaiev
> ---
From: Ian Romanick
This reproduces bug #109075. In that bug, '-int(XYZ) & anything' was
optimized as though it were '-int(XYZ) & floatBitsToInt(1.0)'.
Signed-off-by: Ian Romanick
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=109075
---
...glsl-fs-bitwise-and-Boolean
From: Ian Romanick
Currently on Mesa this test case is pessimal for memory usage.
According to Valgrind massif, it require > 5GiB of memory to compile.
Defining the macro FAIL_REGISTER_ALLOCATION causes the test case to
fall off a cliff on the i965 driver. The scheduler causes > 22,00
Ah, I guess I didn't realize this took a format string.
Reviewed-by: Ian Romanick
On 11/02/2018 10:50 AM, Vinson Lee wrote:
> idtech-extension-strings.c: In function ‘check_extension_list’:
> idtech-extension-strings.c:131:10: warning: format not a string literal and
> no format
Reviewed-by: Ian Romanick
On 10/11/2018 05:59 AM, Sergii Romantsov wrote:
> Trivial fix of memory leaking for test 'initialized-vbo'.
>
> Fixes: 6a467e00cfae (security/initialized-vbo: test contents of uninitialized
> VBOs)
> CC: Brian Paul
> Signed-off-by: Sergii Romants
gt; Tested-by: Józef Kucia <mailto:joseph.ku...@gmail.com>>
> Reviewed-by: Matt Turner <mailto:matts...@gmail.com>>
I'll take that as, "Yes." ;) If you add a reference to the bug in the
commit message (so that nobody comes along later to "opti
These changes look good, but I'm wondering why. :) Did you find some
bug, or ... ?
On 10/04/2018 10:42 AM, Jason Ekstrand wrote:
> ---
> .../fs_pack.shader_test.mako | 14 +++---
> .../fs_unpack.shader_test.mako | 6 +++---
>
From: Ian Romanick
glslangValidator rejects all of the shaders that try to change the
qualifier with errors like:
ERROR: 0:8: 'redeclaration' : cannot change qualification of gl_TexCoord
Mesa currently allows all of them to compile. Closed-source drivers
have not been tested
From: Ian Romanick
glslangValidator rejects all of the shaders that try to change the
qualifier with errors like:
ERROR: 0:8: 'redeclaration' : cannot change qualification of gl_ClipDistance
Mesa currently allows all of them to compile. Closed-source drivers
have not been tested
From: Ian Romanick
glslangValidator rejects all of the shaders that try to change the
qualifier with errors like:
ERROR: 0:8: 'redeclaration' : cannot change qualification of gl_ClipDistance
Mesa currently allows all of them to compile. Closed-source drivers
have not been tested
From: Ian Romanick
See https://lists.freedesktop.org/archives/piglit/2013-October/007935.html
for the discussion.
Signed-off-by: Ian Romanick
Cc: Andrii Simiklit
---
.../glsl-1.50/execution/geometry/clip-distance-out-values.shader_test | 2 +-
1 file changed, 1 insertion(+), 1 deletion
From: Ian Romanick
As people dig up other games, we can (and should) easily add them.
Signed-off-by: Ian Romanick
Cc: Federico Dossena
Cc: Roland Scheidegger
---
tests/general/CMakeLists.gl.txt | 1 +
tests/general/idtech-extension-strings.c | 149
On 09/15/2018 09:22 AM, Alejandro Piñeiro wrote:
> Right now UBO data filling is based on using the name of the ubo and
> their components, in order to get the block and uniform index. It also
> uses it to set the binding (so it forces block_index and block_binding
> to be equal).
>
> Since
On 09/14/2018 03:16 PM, Dylan Baker wrote:
> Quoting Ian Romanick (2018-09-12 16:29:55)
>> From: Ian Romanick
>>
>> The optimizer recently added the ability to replace a compare with a
>> subtraction under certain circumstances. This can fail for integers.
>>
On 09/12/2018 09:17 PM, Ilia Mirkin wrote:
> On Wed, Sep 12, 2018 at 7:59 PM, Ian Romanick wrote:
>> On 09/12/2018 04:33 PM, Ilia Mirkin wrote:
>>> On Wed, Sep 12, 2018 at 7:29 PM, Ian Romanick wrote:
>>>> From: Ian Romanick
>>>>
>>>&g
On 09/12/2018 04:33 PM, Ilia Mirkin wrote:
> On Wed, Sep 12, 2018 at 7:29 PM, Ian Romanick wrote:
>> From: Ian Romanick
>>
>> The optimizer recently added the ability to replace a compare with a
>> subtraction under certain circumstances. This can fail for integers.
&
From: Ian Romanick
The optimizer recently added the ability to replace a compare with a
subtraction under certain circumstances. This can fail for integers.
For inputs a = 0x8000, b = 4, int(0x8000) < 4, but
int(0x8000) - 4 overflows and results in 0x7ffc. That's not
less t
From: Ian Romanick
These tests caught a bug in an optimized version of nir_op_fsign that I
was working on for i965. I thought of a couple clever ways to optimize
sign(a)*sign(b), but most of those ways would fail one or more of the
tests with ±Inf.
Signed-off-by: Ian Romanick
From: Ian Romanick
Signed-off-by: Ian Romanick
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=107772
Cc: Timothy Arceri
Cc: Eero Tamminen
Cc: Mark Janes
---
.../if-statement-in-macro-in-false-ifdef.vert | 18 ++
1 file changed, 18 insertions(+)
create mode
From: Ian Romanick
As of commit 781a78914c7, this test fails with:
arb_vertex_array-delete-object-0: src/mesa/main/hash.c:240:
_mesa_HashRemove_unlocked: Assertion `key' failed.
Aborted (core dumped)
Marek already has a patch on the Mesa list for this issue.
Signed-off-by: Ian Romanick
Cc
Reviewed-by: Ian Romanick
On 08/28/2018 06:33 PM, Timothy Arceri wrote:
> From Section 3.3. (Preprocessor) of the GLSL 4.60 spec:
>
> "The following operators are also available:
> defined
> ##
> ...
>
> Any directive not listed above wi
Reviewed-by: Ian Romanick
On 08/24/2018 03:57 AM, Alejandro Piñeiro wrote:
> Needed by the tessellation evaluation shader.
> ---
> .../tes-patch-input-array-vec2-index-invalid-rd.shader_test | 1
> +
> 1 file changed, 1 insertion(+)
>
> diff
I'm not even sure what you're trying to fix. As far as I can tell, that
test is making a perfectly valid sequence of calls. If a particular
driver fails on that, it is a bug in the driver.
On 08/25/2018 12:29 PM, Ernestas Kulik wrote:
> Actually, scratch that, I must have been high when I wrote
From: Ian Romanick
The negative tests in this group require some inference from various
specs. A spec bug has been submitted against the
GL_ARB_fragment_shader_interlock spec.
Currently none of the negative tests produce the expected result on
Mesa.
Signed-off-by: Ian Romanick
From: Ian Romanick
Currently none of the negative tests produce the expected result on
Mesa. There are other cases called out by the spec as things you cannot
do, but it's not clear what the expected result is in those cases
(compile- or link-time error, undefined rendering, GPU hang, house
On 08/21/2018 05:05 PM, Sagar Ghuge wrote:
> Signed-off-by: Sagar Ghuge
> ---
> tests/opengl.py | 8 +
> tests/spec/CMakeLists.txt | 1 +
> .../CMakeLists.gl.txt | 15 ++
> .../amd_depth_clamp_separate/CMakeLists.txt
write those
later, if you like.
> @@ -0,0 +1,9 @@
> +// [config]
> +// expect_result: fail
> +// glsl_version: 1.10
> +// [end config]
> +
If you add
void func(int x)
{
return;
}
so that there's only one compilation error,
Patches 4 and 5 are
Reviewed-by: Ian Romanick
On 06/27/2018 10:32 AM, Dylan Baker wrote:
> Fixes: bcaa0cd75fd7b1d35817792b4802cce5eb47754d
>("Test that VertexAttribP*() must use one of the appropriate types.")
> CC: Ian Romanick
> ---
>
> tests/opengl.p
On 06/27/2018 03:44 PM, Timothy Arceri wrote:
>
>
> On 28/06/18 08:28, Timothy Arceri wrote:
>> On 27/06/18 21:58, Tapani Pälli wrote:
>>> On 06/27/2018 02:43 PM, Timothy Arceri wrote:
---
V2: Fix some spelling typos and the commit description
On 06/27/2018 04:58 AM, Tapani Pälli wrote:
>
>
> On 06/27/2018 02:43 PM, Timothy Arceri wrote:
>> ---
>>
>> V2: Fix some spelling typos and the commit description
>>
>> ...on-vec4-mixed-arithmetic-input.shader_test | 37 +++
>> 1 file changed, 37 insertions(+)
>> create
From: Ian Romanick
The saturate version hits an assertion failure in the Mesa i965 driver:
src/intel/compiler/brw_fs.cpp:2377: bool fs_visitor::opt_algebraic(): Assertion
`!"unimplemented: saturate mixed types"' failed.
Signed-off-by: Ian Romanick
---
.../built-in-functions/f
On 06/26/2018 05:15 PM, Timothy Arceri wrote:
> On 27/06/18 07:44, Ian Romanick wrote:
>> From: Ian Romanick
>>
>> The fsign(-abs(x)) tests tickle a bug in the Mesa i965 driver that was
>> found by inspection.
>>
>> Signed-off-by: Ian Romanick
From: Ian Romanick
While working on the fix, I thought of a couple other corner cases.
The original fix did not fix these tests... probably just squash into
a single commit.
---
.../execution/fs-sign-sat-neg-abs.shader_test | 22 +++
.../execution/vs-sign-sat-neg
From: Ian Romanick
The fsign(-abs(x)) tests tickle a bug in the Mesa i965 driver that was
found by inspection.
Signed-off-by: Ian Romanick
---
.../execution/fs-sign-neg-abs.shader_test | 28
.../glsl-1.10/execution/fs-sign-neg.shader_test| 26
Assuming this passes on closed-source drivers that support the feature,
Reviewed-by: Ian Romanick
On 06/23/2018 12:06 AM, Timothy Arceri wrote:
> From the ARB_draw_indirect spec:
>
> "Initially zero is bound to DRAW_INDIRECT_BUFFER. In the
> compatibility profil
a[];
>> } b;
>> should be an array of size 1 and not an unsized array, and so no linker
>> error should be generated.
>>
>> Ref: https://bugs.freedesktop.org/show_bug.cgi?id=106915
>>
>> v2:
>> - replaces the comment at the top with the relev
I think a bunch of those were rebase errors. :( I'm pretty sure patch 4
fixes a rebase error of mine.
Series is
Reviewed-by: Ian Romanick
On 06/25/2018 04:24 PM, Dylan Baker wrote:
> This is a few patches to fix a number of binaries that were removed, renamed,
> or
> neve
From: Ian Romanick
Signed-off-by: Ian Romanick
---
.../execution/shared-atomicMax-float.shader_test | 139
.../execution/shared-atomicMin-float.shader_test | 143 +
.../execution/ssbo-atomicMax-float.shader_test | 141
From: Ian Romanick
If the extension isn't enabled in the shader, the functions should not
be availale. Yes, I botched this in the initial implementation.
Signed-off-by: Ian Romanick
---
.../compiler/shared-atomicExchange-float-fail.comp | 20
.../compiler/shared
From: Ian Romanick
The behavior of NaN with atomicMin and atomicMax is nontrivial. This
test should hit all the possible combinations of non-NaN, signaling NaN,
and quite NaN. The the tests or the extension spec for more details.
The data for these tests was generated using Python code
From: Ian Romanick
Signed-off-by: Ian Romanick
---
.../ssbo-atomicCompSwap-float-NaN.shader_test | 64 ++
1 file changed, 64 insertions(+)
create mode 100644
tests/spec/intel_shader_atomic_float_minmax/execution/ssbo-atomicCompSwap-float-NaN.shader_test
diff --git
From: Ian Romanick
The idea is that implementations that use the integer compare-and-swap
hardware will not see that 0.0 == -0.0.
Signed-off-by: Ian Romanick
---
...-atomicCompSwap-float-negative-zero.shader_test | 63 ++
1 file changed, 63 insertions(+)
create mode
From: Ian Romanick
This test is similar to the NV_shader_atomic_float test for atomicAdd.
Instead of using atomicAdd, it uses a loop with atomicCompSwap to
open-code it.
Signed-off-by: Ian Romanick
---
.../shared-atomicCompSwap-float.shader_test| 66 +
.../ssbo
From: Ian Romanick
Signed-off-by: Ian Romanick
---
.../shared-atomicExchange-float.shader_test| 77 +
.../ssbo-atomicExchange-float.shader_test | 78 ++
2 files changed, 155 insertions(+)
create mode 100644
tests/spec
From: Ian Romanick
This is nearly identical to
tes-patch-input-array-vec2-index-rd.shader_test, but this test always
reads both array. One of the reads will use an invalid index, and this
should not crash.
This reproduces a bug that I found in the vec4 backend of the i965.
This test currently
On 06/14/2018 12:27 PM, Eleni Maria Stea wrote:
> Modified the unsized_array_member.shader_test to expect successful
> linking instead of linker error.
>
> According to the GLSLang Specification (version 4.60 Section 4.2 Scoping):
>
> "An array implicitly sized in one shader can be explicitly
From: Ian Romanick
This is identical to
tests/spec/arb_shader_storage_buffer_object/execution/ssbo-atomicCompSwap-int.shader_test,
but this test uses gl_HelperInvocation instead of gl_SampleMaskIn.
Signed-off-by: Ian Romanick
---
.../execution/ssbo-atomicCompSwap-int.shader_test | 78
From: Ian Romanick
v2: Use gl_SampleMaskIn to detect helper invocations. Suggested by
issue #22 of the GL_ARB_shader_image_load_store spec.
Signed-off-by: Ian Romanick
---
.../execution/ssbo-atomicCompSwap-int.shader_test | 84 ++
1 file changed, 84 insertions(+)
create
On 06/11/2018 11:22 PM, Jordan Justen wrote:
> From: Matt Turner
>
> Using ortho makes shrinking the window size much more difficult, which
> makes running tests in simulation take much longer.
>
> Ref: https://lists.freedesktop.org/archives/piglit/2013-November/008298.html
>
uld want is a function the child could call to
unregister all previously registered atexit functions.
Boo.
Reviewed-by: Ian Romanick
> It actually results in crashing with the Mesa radeonsi driver.
>
> Signed-off-by: Michel Dänzer
> ---
> .../spec/glx_ext_import_context/impor
On 06/05/2018 03:41 PM, Vinson Lee wrote:
> piglit-util-gl.c:1941:1: warning: ‘read_texture_via_fbo’ defined but not used
> [-Wunused-function]
> read_texture_via_fbo(int target, int level, int x, int y, int layer, int w,
> ^~~~
>
> Fixes: 77a4b73436c0 ("util: provide way to
On 06/05/2018 04:58 AM, Illia Iorin wrote:
> This test checks that if we allocate more memory for the buffer
> than possible, we get GL_OUT_OF_MEMORY error.
> V2:
> - check GL_ARB_texture_buffer_object extension
> - change storage format to GL_RGBA32I
> - check gl compat version 3.1 because
On 06/04/2018 06:45 AM, Illia Iorin wrote:
> This test checks that if we allocate more memory for the buffer
> than possible, we get GL_OUT_OF_MEMORY error.
>
> Signed-off-by: Illia Iorin
> ---
> tests/opengl.py | 1 +
> .../CMakeLists.gl.txt
Sorry for the lag... this hit the list while I was on vacation, and I
just forgot to check back.
Reviewed-by: Ian Romanick
On 05/11/2018 01:38 PM, Laura Ekstrand wrote:
> ---
> tests/opengl.py| 1 +
> tests/spec/gl-1.0/CMakeLists.gl.txt| 1 +
> tests
From: Ian Romanick
Signed-off-by: Ian Romanick
---
.../execution/ssbo-atomicCompSwap-int.shader_test | 75 ++
1 file changed, 75 insertions(+)
create mode 100644
tests/spec/arb_shader_storage_buffer_object/execution/ssbo-atomicCompSwap-int.shader_test
diff --git
From: Ian Romanick
v2: Use %i for the "data" parameter so that hex values can be specified.
Signed-off-by: Ian Romanick
---
tests/shaders/shader_runner.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/tests/shaders/shader_runner.c b/tests/shaders/shader_runner.c
index
From: Ian Romanick
Verify all versions of imageAtomicAdd and imageAtomicExchnage exist.
"Positive" compile testing for all the other functions will be handled
by the execution tests. I'm much too lazy to write execution tests for
all the variants of imageAtomicAdd and imageAtom
From: Ian Romanick
Signed-off-by: Ian Romanick
---
.../shared-atomicExchange-int.shader_test | 75 ++
1 file changed, 75 insertions(+)
create mode 100644
tests/spec/arb_compute_shader/execution/shared-atomicExchange-int.shader_test
diff --git
a/tests/spec
From: Ian Romanick
Signed-off-by: Ian Romanick
---
.../shared-atomicCompSwap-int.shader_test | 64 ++
1 file changed, 64 insertions(+)
create mode 100644
tests/spec/arb_compute_shader/execution/shared-atomicCompSwap-int.shader_test
diff --git
a/tests/spec
From: Ian Romanick
Signed-off-by: Ian Romanick
---
.../shared-atomicExchange-float.shader_test| 77 +
.../ssbo-atomicExchange-float.shader_test | 78 ++
2 files changed, 155 insertions(+)
create mode 100644
tests/spec
From: Ian Romanick
Signed-off-by: Ian Romanick
---
.../execution/shared-atomicAdd-float.shader_test | 59 ++
.../execution/ssbo-atomicAdd-float.shader_test | 70 ++
2 files changed, 129 insertions(+)
create mode 100644
tests/spec
From: Ian Romanick
Signed-off-by: Ian Romanick
---
.../execution/ssbo-atomicAdd-int.shader_test | 68 ++
1 file changed, 68 insertions(+)
create mode 100644
tests/spec/arb_shader_storage_buffer_object/execution/ssbo-atomicAdd-int.shader_test
diff --git
a/tests
From: Ian Romanick
Signed-off-by: Ian Romanick
---
.../execution/shared-atomicAdd-int.shader_test | 57 ++
1 file changed, 57 insertions(+)
create mode 100644
tests/spec/arb_compute_shader/execution/shared-atomicAdd-int.shader_test
diff --git
a/tests/spec
This series has two parts.
The second part adds tests for NV_shader_atomic_float. I worked on an
implementation of this extension for i965 until I realized the hardware
couldn't actually do it. I had the tests, so it seems a shame to let
them rot in a branch somewhere. Patches 10, 11, and 13
From: Ian Romanick
If the extension isn't enabled in the shader, the functions should not
be availale. Yes, I botched the SSBO and shared variable parts in the
initial implementation.
Signed-off-by: Ian Romanick
---
.../imageAtomicAdd-float-image2D-fail.frag | 20
From: Ian Romanick
Signed-off-by: Ian Romanick
---
.../execution/ssbo-atomicMax-int.shader_test | 139
.../execution/ssbo-atomicMax-uint.shader_test | 139
.../execution/ssbo-atomicMin-int.shader_test | 140
From: Ian Romanick
Signed-off-by: Ian Romanick
---
.../execution/shared-atomicMax-int.shader_test | 137
.../execution/shared-atomicMax-uint.shader_test| 137
.../execution/shared-atomicMin-int.shader_test | 141
From: Ian Romanick
Signed-off-by: Ian Romanick
---
.../execution/ssbo-atomicExchange-int.shader_test | 76 ++
1 file changed, 76 insertions(+)
create mode 100644
tests/spec/arb_shader_storage_buffer_object/execution/ssbo-atomicExchange-int.shader_test
diff --git
On 05/03/2018 02:59 PM, Ilia Mirkin wrote:
> On Thu, May 3, 2018 at 5:58 PM, Ian Romanick <i...@freedesktop.org> wrote:
>> On 04/27/2018 04:56 PM, Ilia Mirkin wrote:
>>> On Fri, Apr 27, 2018 at 9:07 AM, Ian Romanick <i...@freedesktop.org> wrote:
>>>>
On 04/27/2018 04:56 PM, Ilia Mirkin wrote:
> On Fri, Apr 27, 2018 at 9:07 AM, Ian Romanick <i...@freedesktop.org> wrote:
>> On 04/27/2018 12:01 AM, Ilia Mirkin wrote:
>>> NVIDIA hardware has "short" immediates, and some instructions have both
>>>
From: Ian Romanick <ian.d.roman...@intel.com>
A heavily modified version of this test passes on a Geforce3 (NV20)
running NVIDIA's 71.86.15 drivers. These are the last drivers that
support NV10 through NV2x, and they are shipped in 2011.
v2: Use piglit_draw_rect_z instead of the VBO comp
On 05/02/2018 11:06 AM, Eric Anholt wrote:
> Ian Romanick <i...@freedesktop.org> writes:
>
>> From: Ian Romanick <ian.d.roman...@intel.com>
>>
>> The issuses section of GL_NV_fog_distance says:
>>
>> How does this ext
From: Ian Romanick <ian.d.roman...@intel.com>
v2: Explicitly use GL_NO_ERROR instead of expected_error in the
'piglit_is_extension_supported("GL_NV_fog_distance") &&
!piglit_is_extension_supported("GL_SGIS_fog_function")' block. Based on
conversation with
From: Ian Romanick <ian.d.roman...@intel.com>
And vice versa.
v2: Elimintate the static arrays used to populate operators[]. This
simplifies the code a lot. Suggested by Eric.
Signed-off-by: Ian Romanick <ian.d.roman...@intel.com>
---
tests/spec/gl-1.0/
On 05/02/2018 11:02 AM, Eric Anholt wrote:
> Ian Romanick <i...@freedesktop.org> writes:
>> diff --git a/tests/spec/nv_fog_distance/simple-draw.c
>> b/tests/spec/nv_fog_distance/simple-draw.c
>> new file mode 100644
>> index 0..e7be6bf5e
>
On 05/02/2018 10:54 AM, Eric Anholt wrote:
> Ian Romanick <i...@freedesktop.org> writes:
>
>> From: Ian Romanick <ian.d.roman...@intel.com>
>>
>> Signed-off-by: Ian Romanick <ian.d.roman...@intel.com>
>> ---
>> tests/all.py
On 05/02/2018 10:47 AM, Eric Anholt wrote:
> Ian Romanick <i...@freedesktop.org> writes:
>
>> From: Ian Romanick <ian.d.roman...@intel.com>
>>
>> And vice versa.
>>
>> Signed-off-by: Ian Romanick <ian.d.roman...@i
On 05/02/2018 10:47 AM, Eric Anholt wrote:
> Ian Romanick <i...@freedesktop.org> writes:
>
>> From: Ian Romanick <ian.d.roman...@intel.com>
>>
>> There were two bugs here. First, GL_EXT_blend_color and
>> GL_EXT_blend_func_separate were only tested if t
On 04/27/2018 12:01 AM, Ilia Mirkin wrote:
> NVIDIA hardware has "short" immediates, and some instructions have both
> long and short immediate variants. Nouveau was picking the wrong one
> under certain conditions. This adds tests which are likely to cause
> these kinds of issues to surface in a
Reviewed-by: Ian Romanick <ian.d.roman...@intel.com>
On 04/26/2018 11:57 PM, Ilia Mirkin wrote:
> Nouveau does not merge the multiple requests together into one, which
> leads to an interesting pattern on nv50 (since dst reg == src reg) where
> we accidentally overwrite things we
On 04/21/2018 06:34 PM, Ilia Mirkin wrote:
> Nouveau does not merge the multiple requests together into one, which
> leads to an interesting pattern on nv50 (since dst reg == src reg) where
> we accidentally overwrite things we aren't supposed to.
>
> Signed-off-by: Ilia Mirkin
On 04/23/2018 12:16 PM, Laura Ekstrand wrote:
> Only tested on Broadwell gt2.
> ---
> tests/all.py | 1 +
> tests/spec/gl-1.0/CMakeLists.gl.txt| 1 +
> tests/spec/gl-1.0/bitmap-heart-dance.c | 215
> +
> 3 files changed, 217
From: Ian Romanick <ian.d.roman...@intel.com>
Signed-off-by: Ian Romanick <ian.d.roman...@intel.com>
---
tests/all.py | 5 +
tests/spec/CMakeLists.txt| 1 +
tests/spec/nv_fog_distance/CMakeLists.gl.txt | 11 ++
tests/spec/nv_
1 - 100 of 921 matches
Mail list logo
