[SYCL][Doc] Graph fusion extension proposal (#8678)

Experimental SYCL extension proposal for kernel fusion on top of the
[SYCL graphs
API](https://github.com/reble/llvm/blob/sycl-graph-update/sycl/doc/extensions/proposed/sycl_ext_oneapi_graph.asciidoc).

Constructing the sequence of kernels to fuse is completely left to the
graphs proposal, which provides two APIs to this end. One recording API
similar to the fusion mode for queues in the initial kernel fusion
proposal, and an explicit graph construction APIs. Both APIs are
supported for kernel fusion.

This proposal mainly introduces a number of properties to trigger fusion
of the graph and internalization of dataflow in the fused kernel.

This proposal continues some of the ideas of the [experimental SYCL
extension for kernel
fusion](https://github.com/intel/llvm/blob/sycl/sycl/doc/extensions/experimental/sycl_ext_codeplay_kernel_fusion.asciidoc).
In contrast to the original kernel fusion proposal, this proposal now
also allows internalization of USM pointers.

---------

Signed-off-by: Lukas Sommer <[email protected]>

Author: sommerlukas

sycl/doc/extensions/experimental/sycl_ext_codeplay_kernel_fusion.asciidoc

@@ -48,6 +48,18 @@ or may be in a preliminary state.  The specification itself may also change in
 incompatible ways before it is finalized.  *Shipping software products should
 not rely on APIs defined in this specification.*
 
+[NOTE]
+====
+There is a link:../proposed/sycl_ext_oneapi_graph_fusion.asciidoc[follow-up
+proposal] for fusion based on the 
+https://github.com/intel/llvm/blob/sycl/sycl/doc/extensions/experimental/sycl_ext_oneapi_graph.asciidoc[SYCL graph API].
+That proposal continues some of the ideas presented in this proposal, but uses
+the more versatile SYCL graphs API to define the sequence of kernels to
+execute.
+
+Once accepted and implemented, the new proposal will supersede this proposal.
+====
+
 [NOTE]
 ==== 
 This is an experimental extension for the SYCL specification.

sycl/doc/extensions/experimental/sycl_ext_oneapi_graph.asciidoc

@@ -1207,7 +1207,7 @@ The `sycl::ext::codeplay::experimental::property::queue::enable_fusion` property
 defined by the extension is ignored by queue recording.
 
 To enable kernel fusion in a `command_graph` see the
-https://github.com/sommerlukas/llvm/blob/proposal/graph-fusion/sycl/doc/extensions/proposed/sycl_ext_oneapi_graph_fusion.asciidoc[sycl_ext_oneapi_graph_fusion extension proposal]
+link:../proposed/sycl_ext_oneapi_graph_fusion.asciidoc[sycl_ext_oneapi_graph_fusion extension proposal]
 which is layered ontop of `sycl_ext_oneapi_graph`.
 
 ==== sycl_ext_oneapi_kernel_properties