Merge pull request #2365 from IntelPython/add_sycl_queue_fill

Add `dpctl.SyclQueue.fill()` method

Author: vlad-perevezentsev

CHANGELOG.md

@@ -13,6 +13,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 * Added `create_kernel_bundle_from_sycl_source`, `is_sycl_source_compilation_available`, and `dpctl.SyclDevice.can_compile` for supporting the creation of `dpctl.SyclKernelBundle`s from SYCL source strings via DPC++ extension, as well as corresponding C-API functions to support it [gh-2206](https://github.com/IntelPython/dpctl/pull/2206)
 * Added `dpctl.SyclQueue.memset` and `dpctl.SyclQueue.memset_async` methods [gh-2361](https://github.com/IntelPython/dpctl/pull/2361)
 * Added `DPCTLQueue_MemsetWithEvents` C-API function to support `dpctl.SyclQueue.memset_async` [gh-2361](https://github.com/IntelPython/dpctl/pull/2361)
+* Added `dpctl.SyclQueue.fill` and `dpctl.SyclQueue.fill_async` methods [gh-2365](https://github.com/IntelPython/dpctl/pull/2365)
+* Added `DPCTLQueue_Fill8/16/32/64/128WithEvents` C-API functions to support `dpctl.SyclQueue.fill_async` [gh-2365](https://github.com/IntelPython/dpctl/pull/2365)
 
 ### Changed
 * Bump minimum NumPy version to 1.26 [gh-2192](https://github.com/IntelPython/dpctl/pull/2192)

dpctl/_backend.pxd

@@ -21,7 +21,7 @@
 types defined by dpctl's C API.
 """
 
-from libc.stdint cimport int64_t, uint8_t, uint32_t, uint64_t
+from libc.stdint cimport int64_t, uint8_t, uint16_t, uint32_t, uint64_t
 from libcpp cimport bool
 
 
@@ -674,6 +674,66 @@ cdef extern from "syclinterface/dpctl_sycl_queue_interface.h":
         size_t Count,
         const DPCTLSyclEventRef *depEvents,
         size_t depEventsCount)
+    cdef DPCTLSyclEventRef DPCTLQueue_Fill8(
+        const DPCTLSyclQueueRef Q,
+        void *Dest,
+        uint8_t Val,
+        size_t Count)
+    cdef DPCTLSyclEventRef DPCTLQueue_Fill8WithEvents(
+        const DPCTLSyclQueueRef Q,
+        void *Dest,
+        uint8_t Val,
+        size_t Count,
+        const DPCTLSyclEventRef *depEvents,
+        size_t depEventsCount)
+    cdef DPCTLSyclEventRef DPCTLQueue_Fill16(
+        const DPCTLSyclQueueRef Q,
+        void *Dest,
+        uint16_t Val,
+        size_t Count)
+    cdef DPCTLSyclEventRef DPCTLQueue_Fill16WithEvents(
+        const DPCTLSyclQueueRef Q,
+        void *Dest,
+        uint16_t Val,
+        size_t Count,
+        const DPCTLSyclEventRef *depEvents,
+        size_t depEventsCount)
+    cdef DPCTLSyclEventRef DPCTLQueue_Fill32(
+        const DPCTLSyclQueueRef Q,
+        void *Dest,
+        uint32_t Val,
+        size_t Count)
+    cdef DPCTLSyclEventRef DPCTLQueue_Fill32WithEvents(
+        const DPCTLSyclQueueRef Q,
+        void *Dest,
+        uint32_t Val,
+        size_t Count,
+        const DPCTLSyclEventRef *depEvents,
+        size_t depEventsCount)
+    cdef DPCTLSyclEventRef DPCTLQueue_Fill64(
+        const DPCTLSyclQueueRef Q,
+        void *Dest,
+        uint64_t Val,
+        size_t Count)
+    cdef DPCTLSyclEventRef DPCTLQueue_Fill64WithEvents(
+        const DPCTLSyclQueueRef Q,
+        void *Dest,
+        uint64_t Val,
+        size_t Count,
+        const DPCTLSyclEventRef *depEvents,
+        size_t depEventsCount)
+    cdef DPCTLSyclEventRef DPCTLQueue_Fill128(
+        const DPCTLSyclQueueRef Q,
+        void *Dest,
+        uint64_t *Val,
+        size_t Count)
+    cdef DPCTLSyclEventRef DPCTLQueue_Fill128WithEvents(
+        const DPCTLSyclQueueRef Q,
+        void *Dest,
+        uint64_t *Val,
+        size_t Count,
+        const DPCTLSyclEventRef *depEvents,
+        size_t depEventsCount)
     cdef DPCTLSyclEventRef DPCTLQueue_Memset(
         const DPCTLSyclQueueRef Q,
         void *Dest,

dpctl/_sycl_queue.pxd

@@ -107,6 +107,10 @@ cdef public api class SyclQueue (_SyclQueue) [
     cpdef SyclEvent copy_async(
         self, dest, src, size_t count, list dEvents=*, str dtype=*
     )
+    cpdef fill(self, dest, value, size_t count, str dtype=*)
+    cpdef SyclEvent fill_async(
+        self, dest, value, size_t count, list dEvents=*, str dtype=*
+    )
     cpdef memset(self, mem, int val, size_t count=*)
     cpdef SyclEvent memset_async(
         self, mem, int val, size_t count=*, list dEvents=*