This function can be used to query local work size that would be used in
case of querying ND range kernel with local_work_size set to NULL.
Change-Id: I86ac7c97cffb6c3e11b673a28285739edfabb4a6
Signed-off-by: Filip Hazubski <filip.hazubski@intel.com>
Add possibility to set Kernel execution type via clSetKernelExecInfo.
Change-Id: I2de0bf4e067921440e34e14c6975f640d3f47162
Signed-off-by: Filip Hazubski <filip.hazubski@intel.com>
-use clSetKernelExecInfo with param
CL_KERNEL_EXEC_INFO_THREAD_ARBITRATION_POLICY_INTEL
to change default value of ThreadArbitrationPolicy
Change-Id: I15d0de0840ed14687c16ae04890b662bc157de76
Signed-off-by: Katarzyna Cencelewska <katarzyna.cencelewska@intel.com>
This API allows querying for execution related information.
CL_EXECUTION_INFO_MAX_WORKGROUP_COUNT_INTEL parameter allows to query for
maximal work group count that is possible to run concurrently on the device.
Related-To: NEO-2712
Change-Id: I4e6b4c80aeb06ff966fb543c0a7f05ed54416dab
Signed-off-by: Filip Hazubski <filip.hazubski@intel.com>
Add usage of the function where applicable
Change-Id: I9c41ad37009d1f088fa1917d30525194a10d3061
Signed-off-by: Filip Hazubski <filip.hazubski@intel.com>
- When resource is uncached for surface state and not used in stateless manner
then it doesn't need to flush cache
- Minor cleanup
Change-Id: I4cfe5a6fe3e666200407d9acdd89e6f64b2b3eed
Signed-off-by: Mrozek, Michal <michal.mrozek@intel.com>
- Change kernel to properly detect true stateless resources
- do not turn of stateless l3 if arg is used in pure stateful manner
- refactor variable names to better reflect what they do
- improve mock kernel with internal to have setKernelArg capabilties
Change-Id: I2cdde04f2144d9b86dc1486126632db0fd7cad49
Signed-off-by: Mrozek, Michal <michal.mrozek@intel.com>
- With this flag resource will not be cached in L3 for stateful accesses.
Change-Id: Icf9a393ab92d55c2cdf30444420ea40da0d5630c
Signed-off-by: Mrozek, Michal <michal.mrozek@intel.com>
- If resources is constant arg it means it will only be read
- Therefore even for read_write buffer we may turn on caching.
Change-Id: Id2c34d4993111bf5c6523a1946464c397db95686
Signed-off-by: Mrozek, Michal <michal.mrozek@intel.com>
- Add basic allocation support for shared allocations
- Add kernel support for shared allocations.
Related-To: NEO-3148
Change-Id: Ie0523acc3a444eef6a5aeb6a56a041280df6a02e
Signed-off-by: Mrozek, Michal <michal.mrozek@intel.com>
- Add kernel support for host um allocations
- During make resident call choose only appropriate resources for residency
- change resource types to binary bit friendly values
- enhance memory manager to only make resident compatible types
Related-To: NEO-3148
Change-Id: Ic711a4425a0d8db151a335e0357440312dc09b7e
Signed-off-by: Mrozek, Michal <michal.mrozek@intel.com>
- Add vector of UM allocations to kernel that will be accessed indirectly
- Wire in support to clSetKernelExecInfo
Change-Id: I5858f2a93ab957597f3131022daee3741c3453ab
Related-To: NEO-3148
Signed-off-by: Mrozek, Michal <michal.mrozek@intel.com>
- Change function names to indicate they work on SVM allocations.
- Remove one function used only in tests.
Change-Id: I9b18d9fee3d4f2a46a7f458ca73d39b3863ce6d3
Signed-off-by: Mrozek, Michal <michal.mrozek@intel.com>
