* Remove unused code in clCopyImage
Some of the declarations that are still used now moved to testBase.h
as these are reused throughout the image tests.
Signed-off-by: Radek Szymanski <radek.szymanski@arm.com>
* Add missing file
Signed-off-by: Radek Szymanski <radek.szymanski@arm.com>
* Add memory_scope_all_devices testing
This duplicats memory_scope_all_svm_devices testing, but it seems pretty quick
so I don't think it hurts.
Fixes#990
* Address clang-format failures
* Address a further clang-format failure
* Remove binaries for spirv_new (#21)
Avoid storing binaries in version control, as it is generally not well
suited to this. Instead, in a followup commit we will add a script to
regenerate the .spv files from their sources.
Signed-off-by: Stuart Brady <stuart.brady@arm.com>
* Add Python script to assemble spirv_new binaries (#21)
This also changes the ext_cl_khr_spirv_no_integer_wrap_decoration assembly
source so that the SPIR-V version is listed as 1.0, to ensure that the
output of spirv-dis exactly matches the assembly.
Signed-off-by: Stuart Brady <stuart.brady@arm.com>
The following tests make calls to `clCompileProgram` or
`clBuildProgram` and therefore require a compiler in the runtime:
* `unload_repeated`
* `unload_compile_unload_link`
* `unload_build_unload_create_kernel`
* `unload_link_different↩`
* `unload_build_threaded`
* `unload_build_info`
* `unload_program_binaries`
Skip these tests if no compiler is present in the runtime.
* Reformat common help text
Signed-off-by: Stuart Brady <stuart.brady@arm.com>
* Reformat test harness code
This goes part of the way to fixing issue #625.
Signed-off-by: Stuart Brady <stuart.brady@arm.com>
* Permit half overflow within allowable ULP
Modify the algorithm for calculating half precision ULP error so
that it duplicates the behaviour of the single precision ULP algorithm,
in regards to allowing overflow within the defined ULP error.
In the case where the test value is infinity, but the reference is
finite, pretend the test value is 63336.0 and calculate the ULP error
against that.
Encountered this while testing half precision `hypot()` in PR !529,
for inputs `hypot(-48864.0, 43648.0)` which has reference
`65519.755799`. With RTE rounding this only just rounds to `65504` as half,
and returning INF is currently infinite ULP error. Using the leniency
introduced by this change however the error is `~0.5` within the `2` ULP
bounds defined by the spec.
* Run clang-format over changes
Code now conforms to style guidelines and allows `check-format.sh` to pass.
* Skip compiler unload/reload tests if compiler is not available
Note that tests that use the compiler helper functions but won't work without a
compiler are in the `subtests_to_skip_with_offline_compiler` list. I didn't do
that here because they tests directly use `clBuildProgram` etc directly,
because they're specifically testing it's behaviour in edge-cases.
* Change type of status variable
- Remove the build_<platform> scripts that were simply calling cmake in favor of
using cmake directly
- Move flag CRT_SECURE_NO_WARNING into a section specifically for visual studio
- Change vendor file selection to just use the file if present
- Add a variable for determining whether to link against pthread
- Delete all lines in CMakeVendor.txt so each implementation can define their own
Change-Id: Ibbd83521ce4d42d09dcbd0b16efa9fbe6cbf785d
When running sub-tests explicitly on the command line and that sub-test
fails, the test harness returns `0` from `main()`. This is problematic
in automated testing environments as failed tests can be hidden.
This patch iterates over all the test results in the `resultTestList`
after all sub-tests have completed, if any of the tests failed a
non-zero value is returned from `parseAndCallCommandLineTests()` which
is then propagated up the call stack to `main()`.
The host compiler will not always calculate reference values
the same, depending on optimization level. It generates
instructions that do not respond to CPU rounding mode in
the same way. Use QCOM rounding mode emulation to correctly
calculate reference values on aarch64.
Fix validate_float/half_write_results so that when nan/inf is
encountered on a channel, the rest of the channel values are still
considered for correctness.
Signed-off-by: John Kesapides <john.kesapides@arm.com>
Signed-off-by: James Morrissey <james.morrissey@arm.com>
Co-authored-by: John Kesapides <john.kesapides@arm.com>
In embedded profile with cles_khr_int64 support,
the test did not consider the size of long. Thus the
number of argument computation was not coherent with
the returned CL_DEVICE_MAX_PARAMETER_SIZE leaded to test failure.
We fix this by taking the size of long when cles_khr_int64
is available in the device extension.
* Fail feature macro compare if compiler has more features than runtime
Because a C++11 `std::equal` only iterates over the first container, and
matches with items in the second, if the second container contains more items
the check can still pass even though they're not identical. Just use `==`
instead.
Fixes#979
* Move an expression to its point of use
A previously created Image was not being released leading to a
memory leak.
Change-Id: I9715b5c4193a8e89c6c182b443959009ce1f1129
Signed-off-by: Chetankumar Mistry <chetan.mistry@arm.com>
- Remove the requirement to detail values for device queries and list
extensions in the submission template. They are printed by computeinfo.
- Require that tests skipped in accordance with the conformance
process be listed in the submission template.
- Remove support for OPENCL_1_0_DEVICE. This variable doesn't control
anything anymore and the updated conformance process no longer refers
to it.
- Require that implementations running via a loader document the loader
that was used.
Signed-off-by: Kevin Petit <kevin.petit@arm.com>
* add test_compiler test for OpenCL C versions
* add checks for required OpenCL C versions
* bugfix: only OpenCL C 2.0 is required for OpenCL 2.1 and 2.2
* pass reference to feature struct vs. pointer
* address review comments regarding C++ identifiers and 3d image writes
The Volatile, Coherent, NonWritable and NonReadable decorations can
only be applied to images if they are storage images, which require
the Shader capability.
Note that in SPIR-V 1.4 and later, the NonWritable decoration can
also be applied to objects in the Private and Function storage classes
(although Private also requires the Shader capability), but this was
not covered by the existing test for the NonWritable decoration.
Signed-off-by: Stuart Brady <stuart.brady@arm.com>
* initial version, tests vec2 and vec4
* added all types and vector sizes
* fix formatting
* add checks for long and ulong support
* test floats unconditionally, not tied to long support (oops)
* fix for vec3 kernel arguments
* remove generic address space dependency
* This removes API specific csv files, as these are no longer required.
This also updates opencl_conformance_tests_quick.csv to include >= 2.0 tests.
No changes were required for opencl_conformance_tests_full.csv.
Addresses #863
* Pass the correct arguments to test_spirv_new.
* Test feature macro OpenCL 3.0
* Test feature macro OpenCL 3.0 - refactor to template function
* Use test_error and format fix issues
* Fix returned value type for CL_DEVICE_MAX_GLOBAL_VARIABLE_SIZE
* Code format fix
* Feature macro test - refactor code to have one test case
* add consistency checking comparing supported features to CL_DEVICE_OPENCL_C_FEATURES
* fix minor issues
* Feature macro test - fix test version
* Feature macro test - add -cl-std=CL3.0 build option
* Feature macro test - compilation errors
* Feature macro test - compilation errors OSX
* Feature macro test - fix adding build options
* Feature macro test - share function OutputBuildLogs
* Feature macro test - back removed cases
* Feature macro test - review fixes
* Feature macro test - split check image formats
* Feature macro test - fixed version
* Use Version type for cl_khr_spir extension version checks
The current method of using `std::find(versions, 1.2f)` is unreliable due to
performing a floating point comparison. I.e. on some compiles it will
spuriously report that the required version is missing and skip the tests.
* Address clang-format failure.
* Correct typo.
* The `atomic_flag` test assumes support for the `atomic_scope_device`
scope in the global scope test case. Since `atomic_scope_device` is
optional on an OpenCL-3.0 driver, this test should check for support and
skip otherwise.
