Skip to content
Closed
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions clang/lib/DPCT/APINames_CUB.inc
Original file line number Diff line number Diff line change
Expand Up @@ -230,10 +230,10 @@ ENTRY(cub::ThreadLoad, cub::ThreadLoad, true, NO_FLAG, P4, "Successful")
ENTRY(cub::ThreadStore, cub::ThreadStore, true, NO_FLAG, P4, "Successful")
ENTRY(cub::LoadDirectBlocked, cub::LoadDirectBlocked, true, NO_FLAG, P4, "Successful")
ENTRY(cub::LoadDirectBlockedVectorized, cub::LoadDirectBlockedVectorized, false, NO_FLAG, P4, "Comment")
ENTRY(cub::StoreDirectBlocked, cub::StoreDirectBlocked, false, NO_FLAG, P4, "Comment")
ENTRY(cub::StoreDirectBlocked, cub::StoreDirectBlocked, true, NO_FLAG, P4, "Successful")
ENTRY(cub::StoreDirectBlockedVectorized, cub::StoreDirectBlockedVectorized, false, NO_FLAG, P4, "Comment")
ENTRY(cub::LoadDirectStriped, cub::LoadDirectStriped, true, NO_FLAG, P4, "Successful")
ENTRY(cub::StoreDirectStriped, cub::StoreDirectStriped, false, NO_FLAG, P4, "Comment")
ENTRY(cub::StoreDirectStriped, cub::StoreDirectStriped, true, NO_FLAG, P4, "Successful")
ENTRY(cub::LoadDirectWarpStriped, cub::LoadDirectWarpStriped, false, NO_FLAG, P4, "Comment")
ENTRY(cub::StoreDirectWarpStriped, cub::StoreDirectWarpStriped, false, NO_FLAG, P4, "Comment")

Expand Down
2 changes: 2 additions & 0 deletions clang/lib/DPCT/CUBAPIMigration.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -298,7 +298,9 @@ void CubIntrinsicRule::registerMatcher(ast_matchers::MatchFinder &MF) {
"DeviceCountUncached", "DeviceCountCachedValue",
"PtxVersion", "PtxVersionUncached", "SmVersion",
"SmVersionUncached", "RowMajorTid",
"StoreDirectBlocked", "StoreDirectStriped"),
"LoadDirectBlocked", "LoadDirectStriped"),

hasAncestor(namespaceDecl(hasName("cub")))))))
.bind("IntrinsicCall"),
this);
Expand Down
15 changes: 15 additions & 0 deletions clang/lib/DPCT/Rewriters/CUB/RewriterUtilityFunctions.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,7 @@
#include "AnalysisInfo.h"
#include "CallExprRewriterCUB.h"
#include "CallExprRewriterCommon.h"
#include "InclusionHeaders.h"

using namespace clang::dpct;

Expand Down Expand Up @@ -154,6 +155,20 @@ RewriterMap dpct::createUtilityFunctionsRewriterMap() {
CALL(MapNames::getDpctNamespace() + "get_minor_version",
makeDeviceStr()),
LITERAL("10"))))
// cub::StoreDirectBlocked
HEADER_INSERT_FACTORY(
HeaderType::HT_DPCT_GROUP_Utils,
CALL_FACTORY_ENTRY(
"cub::StoreDirectBlocked",
CALL(MapNames::getDpctNamespace() + "group::store_blocked", NDITEM,
ARG(1), ARG(2))))
// cub::StoreDirectStriped
HEADER_INSERT_FACTORY(
HeaderType::HT_DPCT_GROUP_Utils,
CALL_FACTORY_ENTRY(
"cub::StoreDirectStriped",
CALL(MapNames::getDpctNamespace() + "group::store_striped", NDITEM,
ARG(1), ARG(2))))
// cub::RowMajorTid
MEMBER_CALL_FACTORY_ENTRY("cub::RowMajorTid", NDITEM, /*IsArrow=*/false,
"get_local_linear_id")
Expand Down
23 changes: 23 additions & 0 deletions clang/test/dpct/cub/intrinsic/store.cu
Original file line number Diff line number Diff line change
@@ -0,0 +1,23 @@
// UNSUPPORTED: cuda-8.0, cuda-9.0, cuda-9.1, cuda-9.2, cuda-10.0, cuda-10.1, cuda-10.2
// UNSUPPORTED: v8.0, v9.0, v9.1, v9.2, v10.0, v10.1, v10.2
// RUN: dpct -in-root %S -out-root %T/intrinsic/store %S/store.cu --cuda-include-path="%cuda-path/include" -- -std=c++14 -x cuda --cuda-host-only
// RUN: FileCheck --input-file %T/intrinsic/store/store.dp.cpp --match-full-lines %s
// RUN: %if build_lit %{icpx -c -fsycl %T/intrinsic/store/store.dp.cpp -o %T/intrinsic/store/store.dp.o %}

// CHECK:#include <sycl/sycl.hpp>
// CHECK:#include <dpct/dpct.hpp>
// CHECK:#include <dpct/group_utils.hpp>
#include <cub/cub.cuh>

__global__ void TestStoreStriped(int *d_data) {
int thread_data[4];
// CHECK: dpct::group::load_striped(item_ct1, d_data, thread_data);
cub::StoreDirectStriped<128>(threadIdx.x, d_data, thread_data);
}


__global__ void BlockedToStripedKernel(int *d_data) {
int thread_data[4];
// CHECK: dpct::group::load_blocked(item_ct1, d_data, thread_data);
cub::StoreDirectBlocked(threadIdx.x, d_data, thread_data);
}