[SYCL][CUDA] Adjust local memory size

npmiller · npmiller · commit 9a718f1982a6 · 2021-12-09T18:25:18.000Z
diff --git a/sycl/plugins/cuda/pi_cuda.hpp b/sycl/plugins/cuda/pi_cuda.hpp
@@ -650,7 +650,8 @@ struct _pi_kernel {
       size_t alignedLocalOffset =
           localOffset + alignment - (localOffset % alignment);
 
-      add_arg(index, sizeof(size_t), (const void *)&(alignedLocalOffset), size);
+      add_arg(index, sizeof(size_t), (const void *)&(alignedLocalOffset),
+              size + (alignedLocalOffset - localOffset));
     }
 
     void set_implicit_offset(size_t size, std::uint32_t *implicitOffset) {

Original file line number	Diff line number	Diff line change
`@@ -650,7 +650,8 @@ struct _pi_kernel {`
`650`	`650`	`size_t alignedLocalOffset =`
`651`	`651`	`localOffset + alignment - (localOffset % alignment);`
`652`	`652`
`653`		`- add_arg(index, sizeof(size_t), (const void *)&(alignedLocalOffset), size);`
	`653`	`+ add_arg(index, sizeof(size_t), (const void *)&(alignedLocalOffset),`
	`654`	`+ size + (alignedLocalOffset - localOffset));`
`654`	`655`	`}`
`655`	`656`
`656`	`657`	`void set_implicit_offset(size_t size, std::uint32_t *implicitOffset) {`