This way it can overlap the other jobs better.
Signed-off-by: Jan Vesely
---
.travis.yml | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/.travis.yml b/.travis.yml
index 3043cd4cad7..b20a200c97d 100644
--- a/.travis.yml
+++ b/.travis.yml
@@ -17,6 +17,7 @@ env:
matrix
Python 3.4 was EOLed in March 2019
Signed-off-by: Jan Vesely
---
.travis.yml | 2 --
1 file changed, 2 deletions(-)
diff --git a/.travis.yml b/.travis.yml
index b20a200c97d..f0faa0c394f 100644
--- a/.travis.yml
+++ b/.travis.yml
@@ -20,8 +20,6 @@ matrix:
- env: BUILD=cmake
- python
On Fri, Jun 28, 2019 at 7:46 PM Dylan Baker wrote:
>
> Quoting Jan Vesely (2019-06-28 13:01:36)
> > From: Jan Vesely
> >
> > Signed-off-by: Jan Vesely
> > ---
> > framework/test/piglit_test.py | 3 ++-
> > 1 file changed, 2 insertions(+), 1 delet
On Fri, Jun 21, 2019 at 4:12 PM Matt Arsenault wrote:
>
>
>
> On Jun 20, 2019, at 2:38 PM, Jan Vesely wrote:
>
> sorry, I'm running against deadlines and traveling this week.
> Does the timeout patch work as expected in failure path?
>
>
> It seems to not work.
From: Jan Vesely
Signed-off-by: Jan Vesely
---
framework/test/piglit_test.py | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/framework/test/piglit_test.py b/framework/test/piglit_test.py
index b4e6eb444e0..0881f00a6d6 100644
--- a/framework/test/piglit_test.py
+++ b
On Wed, Jun 5, 2019 at 11:47 PM Jan Vesely wrote:
>
> From: Jan Vesely
>
> Signed-off-by: Jan Vesely
> ---
> framework/test/piglit_test.py | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/framework/test/piglit_test.py b/framework/test/piglit_test.py
>
On Fri, Jun 21, 2019 at 4:12 PM Matt Arsenault wrote:
>
>
>
> On Jun 20, 2019, at 2:38 PM, Jan Vesely wrote:
>
> sorry, I'm running against deadlines and traveling this week.
> Does the timeout patch work as expected in failure path?
>
>
> It seems to not work.
On Tue, Jun 18, 2019 at 12:36 PM Matt Arsenault wrote:
>
>
>
> > On Jun 5, 2019, at 10:05 PM, Jan Vesely wrote:
> >
> > On Wed, 2019-06-05 at 17:48 -0400, Matt Arsenault wrote:
> >>> On Jun 3, 2019, at 12:46 PM, Jan Vesely wrote:
> >>>
&
From: Jan Vesely
Signed-off-by: Jan Vesely
---
framework/test/piglit_test.py | 1 +
1 file changed, 1 insertion(+)
diff --git a/framework/test/piglit_test.py b/framework/test/piglit_test.py
index 166c2f792ca..b4e6eb444e0 100644
--- a/framework/test/piglit_test.py
+++ b/framework/test
On Wed, 2019-06-05 at 17:48 -0400, Matt Arsenault wrote:
> > On Jun 3, 2019, at 12:46 PM, Jan Vesely wrote:
> >
> > Does rocm do anything special other than using compute rings?
> > What was the HW you tested?
> I don’t think so. This is on gfx900
>
>
> >
On Mon, 2019-06-03 at 10:31 -0400, Matt Arsenault wrote:
> > On Jun 2, 2019, at 2:36 PM, Jan Vesely wrote:
> >
> > he other problem is that even with killed process hung GPU usually
> > makes the machine unable to suspend or reboot on its own, which kills
> >
On Sun, 2019-06-02 at 09:49 -0400, Matt Arsenault wrote:
> > On Jun 1, 2019, at 10:57 PM, Jan Vesely wrote:
> >
> > On Thu, 2019-05-30 at 08:40 -0400, Matt Arsenault wrote:
> > > Ping
> > >
> > > > On May 23, 2019, at 7:59 PM, arse...@gmail.
; > +
> > +// A CSR VGPR needs to be spilled/restored in the prolog/epilog, but
> > +// all lanes need to be made active to avoid clobbering lanes that did
> > +// not enter the call.
> > +kernel void conditional_call(global int* ret)
> > +{
> > +__asm volatile("v_mov_b32 v32, 0xabcd1234" : : : "v
From: Jan Vesely
This prevents race condition when a failed kernel launch would no longer
be in the command queue by the time we call clFinish.
Signed-off-by: Jan Vesely
---
tests/util/piglit-util-cl.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tests/util/piglit-util
From: Jan Vesely
Signed-off-by: Jan Vesely
---
tests/cl/api/create-program-with-binary.c| 3 ++-
tests/cl/custom/buffer-flags.c | 2 +-
tests/cl/custom/flush-after-enqueue-kernel.c | 3 ++-
tests/cl/custom/r600-create-release-buffer-bug.c | 9 ++---
tests/cl
On Thu, 2018-10-18 at 08:30 +, Haehnle, Nicolai wrote:
> On 17.10.18 19:57, Jan Vesely wrote:
> > This test breaks build for compilers that don't default to C++11+ for
> > cpp sources.
> >
> > https://travis-ci.org/jvesely/piglit/jobs/442796343
>
> Sorry a
This test breaks build for compilers that don't default to C++11+ for
cpp sources.
https://travis-ci.org/jvesely/piglit/jobs/442796343
Jan
On Thu, Sep 13, 2018 at 2:11 PM Nicolai Hähnle wrote:
> From: Nicolai Hähnle
>
> Test extremes of GS amplification under tessellation.
> ---
>
On Mon, 2018-09-10 at 20:36 -0700, Matt Arsenault wrote:
> v2: Use uintptr_t
>
> v3: Formatting
>
> v4: More uintptr_t
pushed, thanks.
Jan
> ---
> tests/cl/program/execute/realign-stack.cl | 93 +++
> 1 file changed, 93 insertions(+)
> create mode 100644
On Wed, Sep 5, 2018 at 10:16 AM Matt Arsenault wrote:
> ping
>
merged last week as a9752f23ef3531c6c44fc60cfff9811862fd594a.
Jan
>
> > On Aug 22, 2018, at 15:41, Matt Arsenault wrote:
> >
> > Also fixes apparently missing coverage for special
> > input arguments not passed in registers.
> >
On Wed, Sep 5, 2018 at 10:16 AM Matt Arsenault wrote:
> ping
>
> This one was merged last week.
Jan
> > On Aug 22, 2018, at 15:41, Matt Arsenault wrote:
> >
> > ---
> > .../cl/program/execute/calls-large-struct.cl | 36 +++
> > 1 file changed, 36 insertions(+)
> >
> > diff
On Wed, 2018-09-05 at 19:15 +0500, Matt Arsenault wrote:
> > On Aug 22, 2018, at 10:57, Jan Vesely wrote:
> >
> > On Tue, 2018-08-21 at 21:00 +0300, Matt Arsenault wrote:
> > > ping
> >
> > sorry. I won't have access to my machines until next week (possibl
l int* out) {
> > +volatile int misalign_stack = 0;
> > +*out = test_stack_object_alignment16();
> > +}
> > +
> > +kernel void kernel_call_stack_realign32_func(global int* out) {
> > +volatile int misalign_stack = 0;
> > +*out = test_stack_object_ali
On Mon, 2018-08-13 at 20:06 -0400, Jan Vesely wrote:
> On Mon, 2018-08-13 at 23:29 +0300, Matt Arsenault wrote:
> > > On May 8, 2018, at 18:45, Jan Vesely wrote:
> > >
> > > On Tue, 2018-05-08 at 13:28 +0300, Matt Arsenault wrote:
> > > > >
On Mon, 2018-08-13 at 23:29 +0300, Matt Arsenault wrote:
> > On May 8, 2018, at 18:45, Jan Vesely wrote:
> >
> > On Tue, 2018-05-08 at 13:28 +0300, Matt Arsenault wrote:
> > > > On Apr 8, 2018, at 19:56, Jan Vesely wrote:
> > > >
> > > > O
on POCL.
Reviewed-by: Jan Vesely
I'd help I you cc'ed me on patches you want me to merge directly,
rather than sending a ping later. sifting through the ML requires
extra time.
Jan
> ---
> .../cl/program/execute/calls-large-struct.cl | 156 ++
> tests/cl/prog
On Mon, 2018-08-13 at 23:42 +0300, Matt Arsenault wrote:
> ping
Reviewed-by: Jan Vesely
>
> > On Oct 27, 2017, at 13:03, Matt Arsenault wrote:
> >
> > Doesn't really matter.
so why change it?
still passes on clover carrizo/iceland
Jan
> > ---
> > tests/
On Tue, 2018-05-08 at 13:28 +0300, Matt Arsenault wrote:
> > On Apr 8, 2018, at 19:56, Jan Vesely <jan.ves...@rutgers.edu> wrote:
> >
> > On Fri, 2018-04-06 at 00:49 -0400, Matt Arsenault wrote:
> > > ping
> >
> > I'll need to setup the ro
"v24","v25","v26","v27","v28","v29","v30","v31",
> > +
> > +"s0","s1","s2","s3","s4","s5","s6","s7",
> > +
On Fri, 2018-04-06 at 00:28 -0400, Matt Arsenault wrote:
> > On Apr 4, 2018, at 15:52, Jan Vesely <jan.ves...@rutgers.edu> wrote:
> >
> > redundant newline
>
> Not sure what you mean by this. Do you mean the newline to put the
> single array element on its own li
On Tue, 2018-04-03 at 18:03 -0400, Matt Arsenault wrote:
> ping
>
> > On Mar 29, 2018, at 10:35, Matt Arsenault wrote:
> >
> > ---
> > tests/cl/program/execute/realign-stack.cl | 96
> > +++
> > 1 file changed, 96 insertions(+)
> > create mode
uct_Char_IntArray st =
> > struct_smallregs_Char_IntArray_func(input, id);
> >
> > int sum = 0;
> > for (int i = 0; i < 4; ++i)
> > @@ -134,7 +134,7 @@ kernel void call_sret_Char_IntArray_func(global int*
> > output, global int* input)
> > }
> >
> > NOINLINE
> >
On Fri, 2018-01-19 at 14:43 -0500, Matt Arsenault wrote:
> > On Jan 18, 2018, at 15:02, Jan Vesely <jan.ves...@rutgers.edu> wrote:
> >
> > Why is this necessary? can't you just pass the offset argument as a
> > kernel input?
> >
> > Jan
>
> It n
ubtests it wants to run.
> > +if result.subtests and result.result ==
> > max(six.itervalues(result.subtests)):
> > for res in six.itervalues(result.subtests):
> > res = str(res)
> >
On Tue, 2017-11-28 at 14:20 -0800, Matt Arsenault wrote:
> Explanation in test comment.
> ---
> .../program/execute/amdgcn-mubuf-negative-vaddr.cl | 62
> ++
> 1 file changed, 62 insertions(+)
> create mode 100644 tests/cl/program/execute/amdgcn-mubuf-negative-vaddr.cl
>
>
On Tue, 2017-11-14 at 00:58 -0800, Matt Arsenault wrote:
> ping
Reviewed-by: Jan Vesely <jan.ves...@rutgers.edu>
I've fixed few whitespace errors locally, and I will push this by the
end of the week.
Jan
>
> > On Oct 27, 2017, at 03:02, Matt Arsenault <ars
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/gen_cl_math_builtins.py | 6 --
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/generated_tests/gen_cl_math_builtins.py
b/generated_tests/gen_cl_math_builtins.py
index bd1a859b7..149322ca6
account for fp32 rounding.
sinpi(n) should return 0 for any integer n
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/gen_cl_math_builtins.py | 9 -
1 file changed, 4 insertions(+), 5 deletions(-)
diff --git a/generated_tests/gen_cl_math_builtins
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/gen_cl_math_builtins.py | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/generated_tests/gen_cl_math_builtins.py
b/generated_tests/gen_cl_math_builtins.py
index 5bbc72a0a..e6abbf50a
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/gen_cl_math_builtins.py | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/generated_tests/gen_cl_math_builtins.py
b/generated_tests/gen_cl_math_builtins.py
index 726ac9652..5bbc72a0a
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/gen_cl_math_builtins.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/generated_tests/gen_cl_math_builtins.py
b/generated_tests/gen_cl_math_builtins.py
index 149322ca6..a4cae5314
using libm cosf(3.0f * M_PI / 2.0f) == 0x1.99bc5cp-27
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/gen_cl_math_builtins.py | 8 ++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/generated_tests/gen_cl_math_builtins.py
b/generated
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/gen_cl_math_builtins.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/generated_tests/gen_cl_math_builtins.py
b/generated_tests/gen_cl_math_builtins.py
index e6abbf50a..bd1a859b7
From: Jan Vesely <jan.ves...@rutgers.edu>
Add type3 vectors and bump required CLC to 1.1.
Simplify test to use only 1-2 vstore calls.
Zero the output buffer (fixes undefined values in results of some vstore
tests).
v2: Add canary after the written values
v3: Fixup canary
Signed-off-b
In python:
>>> math.erf(11.1)
1.0
Signed-off-by: Jan Vesely <jano.ves...@gmail.com>
---
These are straightforward fixes. I plan to push them I few days unless there
are objections.
generated_tests/gen_cl_math_builtins.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion
python:
>>> math.erfc(1.960/math.sqrt(2.0)).hex()
'0x1.9990c58859300p-5'
float rtne rounds up to 0x1.9990c6p-5
>>> math.erfc(2.576/math.sqrt(2.0)).hex()
'0x1.4784addb5643fp-7'
float rtne rounds up to 0x1.4784aep-7
Signed-off-by: Jan Vesely <jano.ves...@gmail.com&
On Mon, 2017-10-02 at 10:32 -0700, Matt Arsenault wrote:
> ping
Reviewed-by: Jan Vesely <jan.ves...@rutgers.edu>
Jan
>
> > On Sep 19, 2017, at 19:51, Matt Arsenault <arse...@gmail.com> wrote:
> >
> > ---
> > tests/cl/program/execute/store-hi16-generi
On Tue, 2017-09-19 at 22:02 -0700, Matt Arsenault wrote:
> v2: Fix some formatting
Reviewed-by: Jan Vesely <jan.ves...@rutgers.edu>
sorry for the delay, feel free to cc me on cl piglit patches. I don't
think anyone else is interested in piglit cl.
Are these targeting specific gcn ins
On Thu, 2017-10-05 at 11:51 -0400, Jan Vesely wrote:
> Add type3 vectors and bump required CLC to 1.1.
> Simplify test to use only 1-2 vstore calls.
> Zero the output buffer (fixes undefined values in results of some vstore
> tests).
>
> v2: Add canary after the written valu
On Thu, 2017-10-05 at 12:17 -0700, Matt Arsenault wrote:
> > On Oct 5, 2017, at 08:09, Jan Vesely <jan.ves...@rutgers.edu> wrote:
> >
> > On Mon, 2017-10-02 at 10:32 -0700, Matt Arsenault wrote:
> > > ping
> > >
> > > > On Sep 19, 20
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/gen_cl_vload_tests.py | 108 --
1 file changed, 64 insertions(+), 44 deletions(-)
diff --git a/generated_tests/gen_cl_vload_tests.py
b/generated_tests/gen_cl_vload_tests.py
index 644
Add type3 vectors and bump required CLC to 1.1.
Simplify test to use only 1-2 vstore calls.
Zero the output buffer (fixes undefined values in results of some vstore
tests).
v2: Add canary after the written values
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated
On Mon, 2017-10-02 at 10:32 -0700, Matt Arsenault wrote:
> ping
>
> > On Sep 19, 2017, at 19:25, Matt Arsenault wrote:
> >
> > These aren't great since they need more test
> > values, and generating half results is kind
> > of a pain from any other tool. The perfect
> >
This fails sin/cos/tan tests almost everywhere.
Signed-off-by: Jan Vesely <jano.ves...@gmail.com>
---
tests/util/piglit-util-cl.c | 32 +---
1 file changed, 21 insertions(+), 11 deletions(-)
diff --git a/tests/util/piglit-util-cl.c b/tests/util/piglit-util-cl.c
Basic tests. All new tests, but sincos pass on beignet(SKL) (sin/cos
tests fail on beignet)
Signed-off-by: Jan Vesely <jano.ves...@gmail.com>
---
generated_tests/gen_cl_math_builtins.py | 283 +++-
1 file changed, 280 insertions(+), 3 deletions(-)
diff
Signed-off-by: Jan Vesely <jano.ves...@gmail.com>
---
generated_tests/genclbuiltins.py | 16
1 file changed, 12 insertions(+), 4 deletions(-)
diff --git a/generated_tests/genclbuiltins.py b/generated_tests/genclbuiltins.py
index 338914cb9..25f2f7a9a 100644
--- a/generated
From: Aaron Watry <awa...@gmail.com>
Tested with libclc on a Radeon HD 7850.
Hex output values were generated via remainderf(float, float).
No implications as to the quality of the inputs is intended.
Signed-off-by: Aaron Watry <awa...@gmail.com>
Reviewed-by: Jan Vesely <jan.ves
cos(pi * 2k) = 1
Signed-off-by: Jan Vesely <jano.ves...@gmail.com>
---
generated_tests/gen_cl_math_builtins.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/generated_tests/gen_cl_math_builtins.py
b/generated_tests/gen_cl_math_builtins.py
index d1e0e3c24..4df4918e6
Passes on intel beignet, intel CPU, CUDA OpenCL, clover(carrizo)
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
.../atomic/atomic_int32_add-global-return.cl | 63 +
.../builtin/atomic/atomic_int32_add-global.cl | 60
.../builtin/
Hi Aaron,
looks like you never pushed this one. Did you have a reason to drop it?
Jan
On Fri, 2017-01-20 at 12:37 -0500, Jan Vesely wrote:
> On Tue, 2017-01-17 at 21:18 -0600, Aaron Watry wrote:
> > Tested with libclc on a Radeon HD 7850.
> >
> > Hex output valu
clc allows max long type
Fixes program@execute@calls on CUDA OpenCL
Fixes: e408ce1f2bff23121670a8206258c80bb3d9befd (cl: Add tests for function
calls)
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
tests/cl/program/execute/calls.cl | 2 +-
1 file changed, 1 insertion(+), 1 de
Add type3 vectors and bump required CLC to 1.1.
Simplify test to use only 1-2 vstore calls.
Zero the output buffer (fixes undefined values in results of some vstore
tests).
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/gen_cl_vstore_tests.py
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/gen_cl_vload_tests.py | 108 --
1 file changed, 64 insertions(+), 44 deletions(-)
diff --git a/generated_tests/gen_cl_vload_tests.py
b/generated_tests/gen_cl_vload_tests.py
index 644
f() is deprecated unprototyped declaration use f(void) instead.
funnction call requires a prototype.
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
tests/cl/api/link-program.c | 12 ++--
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/tests/cl/api/link-progr
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
tests/cl/api/link-program.c | 26 +-
1 file changed, 13 insertions(+), 13 deletions(-)
diff --git a/tests/cl/api/link-program.c b/tests/cl/api/link-program.c
index 75896d64c..48563a10c 100644
--- a/tests/cl/ap
v2: Fix xor local test
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
All test pass on clover carrizo/iceland (need libclc and mesa patches).
I haven't found any other implementation exposing int64 atomics
.../atomic/atomic_int64_add-global-return.cl
er test names, use device_regex (effectively restricting to ROCm)
thanks.
Reviewed-by: Jan Vesely <jan.ves...@rutgers.edu>
I want to run this on my carrizo machine before pushing, should be
~wednesday.
Jan
> ---
> tests/cl/program/execute/call-clobbers-amdgcn.cl | 68 +++
> tests/c
config->device_regex =
> add_dynamic_str_copy(value);
> } else if(regex_match(key,
> "^require_platform_extensions$")) {
> config->require_platform_extensions =
>
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
.../atomic/atomic_int64_add-global-return.cl | 63 +
.../builtin/atomic/atomic_int64_add-global.cl | 60
.../builtin/atomic/atomic_int64_add-local.cl | 71 +++
.../
On Sun, 2017-09-17 at 17:02 -0700, Matt Arsenault wrote:
> > On Sep 16, 2017, at 20:15, Jan Vesely <jan.ves...@rutgers.edu> wrote:
> >
> > afaik, noinline is not defined in CLC, so it should be ifdefed on
> > __clang__
> >
>
> It’s not, but my rea
+
> +// Test a basic tail call
> +__attribute__((noinline))
> +int tailcall_i32_func_i32_i32(int x, int y)
> +{
> +x += 5;
> +y += 10;
> +return i32_func_i32_i32(x, y);
> +}
> +
> +// Test a basic tail call with more arguments in the callee than
> +// caller.
> +__attribute__((noinline))
> +int tailcall_i32_func_i32_i32_extra_arg(int x, int y)
> +{
> +int z = x + y + 1;
> +x += 5;
> +y += 10;
> +return i32_func_i32_i32_i32(x, y, z);
> +}
> +
> +// Test a basic tail call with fewere arguments in the callee than
> +// caller.
> +__attribute__((noinline))
> +int tailcall_i32_func_i32_i32_i32_fewer_args(int x, int y, int z)
> +{
> +x += 5;
> +y += 10;
> +return i32_func_i32_i32(x, y + z);
> +}
> +
> +kernel void kernel_call_tailcall(global int* output,
> + global int* input0,
> + global int* input1)
> +{
> +int id = get_global_id(0);
> +output[id] = tailcall_i32_func_i32_i32(input0[id], input1[id]);
> +}
> +
> +kernel void kernel_call_tailcall_extra_arg(global int* output,
> + global int* input0,
> + global int* input1)
> +{
> +int id = get_global_id(0);
> +output[id] = tailcall_i32_func_i32_i32_extra_arg(input0[id], input1[id]);
> +}
> +
> +kernel void kernel_call_tailcall_fewer_args(global int* output,
> +global int* input0,
> +global int* input1,
> +global int* input2)
> +{
> +int id = get_global_id(0);
> +output[id] = tailcall_i32_func_i32_i32_i32_fewer_args(input0[id],
> input1[id], input2[id]);
> +}
> +__attribute__((noinline))
> +int4 v4i32_func_v4i32_v4i32_v4i32_v4i32_v4i32_v4i32_v4i32_v4i32_i32(
> +int4 arg0, int4 arg1, int4 arg2, int4 arg3,
> +int4 arg4, int4 arg5, int4 arg6, int4 arg7,
> +int arg8)
> +{
> +// Try to make sure we can't clobber the incoming stack arguments
> +// with local stack objects.
> +volatile int4 args[8] = { arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7
> };
> +volatile int scalar_arg = arg8;
> +
> +int4 total = 0;
> +for (int i = 0; i < 8; ++i)
> +{
> +total += args[i];
> +}
> +
> +return total * scalar_arg;
> +}
> +
> +// Test a basic tail call
> +__attribute__((noinline))
> +int4 tailcall_v4i32_func_v4i32_v4i32_v4i32_v4i32_v4i32_v4i32_v4i32_v4i32_i32(
> +int4 arg0, int4 arg1, int4 arg2, int4 arg3,
> +int4 arg4, int4 arg5, int4 arg6, int4 arg7,
> +int arg8)
> +{
> +arg0 *= 2;
> +return v4i32_func_v4i32_v4i32_v4i32_v4i32_v4i32_v4i32_v4i32_v4i32_i32(
> +arg0, arg1, arg2, arg3, arg4,
> +arg5, arg6, arg7, arg8);
> +}
> +
> +kernel void kernel_call_tailcall_stack_passed_args(global int4* output,
> + global int4* input0,
> + global int4* input1,
> + global int4* input2,
> + global int4* input3,
> + global int4* input4,
> + global int4* input5,
> + global int4* input6,
> + global int4* input7,
> + global int* input8)
> +{
> +int id = get_global_id(0);
> +output[id] =
> tailcall_v4i32_func_v4i32_v4i32_v4i32_v4i32_v4i32_v4i32_v4i32_v4i32_i32(
> +input0[id],
> +input1[id],
> +input2[id],
> +input3[id],
> +input4[id],
> +input5[id],
> +input6[id],
> +input7[id],
> +input8[id]);
> +}
--
Jan Vesely <jan.ves...@rutgers.edu>
signature.asc
Description: This is a digitally signed message part
___
Piglit mailing list
Piglit@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/piglit
On Fri, 2017-09-08 at 11:32 -0700, Dylan Baker wrote:
> Quoting Jan Vesely (2017-09-01 12:30:52)
> [snip]
> > diff --git a/generated_tests/gen_cl_shuffle2_builtins.py
> > b/generated_tests/gen_cl_shuffle2_builtins.py
> > new file mode 100644
> > index
v2: "The size of each element in the mask must match the size of each element
in the result."
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/CMakeLists.txt | 4 +
generated_tests/gen_cl_shuffle_builtins.py | 149 ++
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/CMakeLists.txt | 4 +
generated_tests/gen_cl_shuffle2_builtins.py | 154
2 files changed, 158 insertions(+)
create mode 100644 generated_tests/gen_cl_shuffle2_builtins.py
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/CMakeLists.txt | 4 +
generated_tests/gen_cl_shuffle_builtins.py | 129 +
tests/cl.py| 3 +
3 files changed, 136 insertions(+)
create mode
On Mon, 2017-08-28 at 15:15 -0700, Vinson Lee wrote:
> On Wed, Jul 19, 2017 at 3:51 PM, Jan Vesely <jan.ves...@rutgers.edu> wrote:
> > v2: Fix warning in cl-api-get-mem-object-info
> > v3: Filter permissible queue properties combinations
> > Use and test clCreat
On Wed, 2017-08-23 at 14:54 -0700, Dylan Baker wrote:
> Quoting Jan Vesely (2017-08-23 12:38:53)
> > Hi Dylan,
> >
> > do you mind taking a look at the python parts?
> >
> > thanks,
> > Jan
>
> Sure, I have a few comments, but mostly this looks
On Thu, 2017-08-24 at 12:54 -0400, Jan Vesely wrote:
> From: Tom Stellard <thomas.stell...@amd.com>
Hi Tom,
this patch was never pushed. Do you remember what hw failed in this
case? Do you mind if I push it?
thanks,
Jan
>
> ---
> generated_tests/gen_cl_math_builtins.py
From: Tom Stellard
---
generated_tests/gen_cl_math_builtins.py | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/generated_tests/gen_cl_math_builtins.py
b/generated_tests/gen_cl_math_builtins.py
index 29b4e1a3f..d1e0e3c24 100644
---
Hi Dylan,
do you mind taking a look at the python parts?
thanks,
Jan
On Wed, 2017-08-16 at 20:39 -0400, Jan Vesely wrote:
> v2: simplify
> mark local storage volatile
> Passes on beignet(IVB), and intel CPU
>
> Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
>
v2: mark temporary storage as volatile
vstore_half requires workaround using short as local/private type
Tests pass on beignet(IVB), and intel CPU
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/gen_cl_vstore_tests.py | 81 ++
v2: simplify
mark local storage volatile
Passes on beignet(IVB), and intel CPU
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
clover on carrizo passes as well, apart from vload_half tests, because
the function is missing in libclc
generated_tests/CMakeLists.txt
Extend to other types
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/CMakeLists.txt | 4 +
generated_tests/gen_cl_vload_tests.py | 230 +
tests/cl.py| 2 +
tests/cl/program/e
vstore_half requires workaround using short as local/private type
Tests pass on clover(Turks,Carrizo,Iceland), beignet(IVB), and intel CPU
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
generated_tests/gen_cl_vstore_tests.py | 81 ++
1 file chang
just a heads up. unless there are objections, I plan to push this in
few days
Jan
On Fri, 2017-08-04 at 16:38 -0400, Jan Vesely wrote:
> based on 9f65c0a26 "cl: Add tests for 24-bit div / rem optimization" by Matt
> Arsenault
>
> Signed-off-by: Jan Vesely &
based on 9f65c0a26 "cl: Add tests for 24-bit div / rem optimization" by Matt
Arsenault
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
tests/cl/program/execute/scalar-arithmetic-uint.cl | 1657
1 file changed, 1657 insertions(+)
diff --git a/tests/cl
Let me know if you want me to postpone, the patch is not on critical
path.
regards,
Jan
>
> --Aaron
>
> On Wed, Jul 19, 2017 at 5:51 PM, Jan Vesely <jan.ves...@rutgers.edu> wrote:
> > v2: Fix warning in cl-api-get-mem-object-info
> > v3:
I plan to push this early next week unless there are objections.
Jan
On Mon, 2017-07-17 at 17:47 -0400, Jan Vesely wrote:
> CL_RGBA + CL_FLOAT is in minimal supported set.
> Fixes sampler test on beignet (IVB).
>
> Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
> --
Based on attributes test. Passed on beignet (IVB) and clover(Turks) with posted
libclc patch.
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
tests/cl/program/execute/kernel_exec.cl | 39 +
1 file changed, 39 insertions(+)
create mode 100644 te
ping.
Tom, Zoltan,
you added the original test, does this look OK to you?
thanks,
Jan
On Mon, 2017-07-17 at 17:47 -0400, Jan Vesely wrote:
> CL_RGBA + CL_FLOAT is in minimal supported set.
> Fixes sampler test on beignet (IVB and SKL).
>
> Signed-off-by: Jan Vesely <jan.ves
On Tue, 2017-07-25 at 00:09 +, Vinson Lee wrote:
> Fix Coverity overflowed return value defect.
>
> CID: 1364163
> Signed-off-by: Vinson Lee <v...@freedesktop.org>
although it relies on signed overflow wraparound. if it cleans up
coverity output
Acked-by: Jan Vesely <j
v2: Fix warning in cl-api-get-mem-object-info
v3: Filter permissible queue properties combinations
Use and test clCreateCommandQueueWithProperties on OCL 2.0
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
I see few regressions with this patch on beignet(SKL) in generated
CL_RGBA + CL_FLOAT is in minimal supported set.
Fixes sampler test on beignet (IVB).
Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
---
tests/cl/program/execute/sampler.cl | 18 +-
1 file changed, 9 insertions(+), 9 deletions(-)
diff --git a/tests/cl/program/e
> + return version;
> > }
> >
> > bool
> > --
> > 2.11.0
> >
> > ___
> > Piglit mailing list
> > Piglit@lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/p
c6e1dc5247 switched to use gbm_bo_get_offset and gbm_bo_get_stride_for_plane.
These symbols were added in mesa 17.1. Since there is no alternative to using
them, bump the gbm requirement to 17.1.
Fixes: c6e1dc5247 "piglit_drm_dma_buf: fix GPU offsets and strides"
Signed-off-by:
On Mon, 2017-06-19 at 16:22 -0700, Eric Anholt wrote:
> Marek Olšák writes:
>
> > From: Marek Olšák
> >
> > Most of the original code is simply wrong.
> >
> > This patch makes sure that at least the returned GPU offsets and strides
> > are correct. This
On Tue, 2017-05-16 at 15:34 -0500, Daniel Díaz wrote:
> Hello!
>
>
> On 05/16/2017 03:06 PM, Jan Vesely wrote:
> > On Mon, 2017-05-15 at 20:27 -0500, Daniel Díaz wrote:
> > > diff --git a/CMakeLists.txt b/CMakeLists.txt index
> > > a4ff99e..cc26f
tatus
> make[2]: *** [bin/point-sprite] Error 1
>
> Instead of relying on the header file, actually try to link
> using that symbol to determine if PIGLIT_HAS_GBM_BO_MAP
> should be defined.
>
> Signed-off-by: Daniel Díaz <daniel.d...@linaro.org>
> ---
> v2: Us
tatus
> make[2]: *** [bin/point-sprite] Error 1
>
> Instead of relying on the header file, actually try to link
> using that symbol to determine if PIGLIT_HAS_GBM_BO_MAP
> should be defined.
>
> Signed-off-by: Daniel Díaz <daniel.d...@linaro.org>
> ---
> v2: Us
On Thu, 2017-05-11 at 14:17 -0500, Daniel Díaz wrote:
> gbm_bo_map() and _unmap() have been added recently to Mesa,
> and this update may not have reached all implementations of
> GBM, such as the one provided by Mali r6, where said
> definitions can be found in the header file but not in the
>
On Fri, 2017-04-14 at 11:32 -0400, Jan Vesely wrote:
> Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu>
> ---
> tests/cl/program/execute/program-scope-arrays.cl | 60
>
> 1 file changed, 60 insertions(+)
>
> diff --git a/tests/cl/pr
1 - 100 of 323 matches
Mail list logo