Skip to content

Commit 22371e6

Browse files
davidrohrdavidrohr
committed
GPU: Fix more SharedMemory should never be __restrict__
1 parent 8579914 commit 22371e6

File tree

5 files changed

+15
-15
lines changed

5 files changed

+15
-15
lines changed

‎GPU/GPUTracking/DataCompression/GPUTPCCompressionKernels.cxx‎

Lines changed: 7 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -521,7 +521,7 @@ GPUdi() unsigned int GPUTPCCompressionGatherKernels::calculateWarpOffsets(GPUSha
521521
}
522522

523523
template <>
524-
GPUdii() void GPUTPCCompressionGatherKernels::Thread<GPUTPCCompressionGatherKernels::unbuffered>(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& GPUrestrict() smem, processorType& GPUrestrict() processors)
524+
GPUdii() void GPUTPCCompressionGatherKernels::Thread<GPUTPCCompressionGatherKernels::unbuffered>(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& smem, processorType& GPUrestrict() processors)
525525
{
526526
GPUTPCCompression& GPUrestrict() compressor = processors.tpcCompressor;
527527
const o2::tpc::ClusterNativeAccess* GPUrestrict() clusters = processors.ioPtrs.clustersNative;
@@ -621,7 +621,7 @@ GPUdii() void GPUTPCCompressionGatherKernels::Thread<GPUTPCCompressionGatherKern
621621
}
622622

623623
template <typename V>
624-
GPUdii() void GPUTPCCompressionGatherKernels::gatherBuffered(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& GPUrestrict() smem, processorType& GPUrestrict() processors)
624+
GPUdii() void GPUTPCCompressionGatherKernels::gatherBuffered(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& smem, processorType& GPUrestrict() processors)
625625
{
626626

627627
GPUTPCCompression& GPUrestrict() compressor = processors.tpcCompressor;
@@ -703,7 +703,7 @@ GPUdii() void GPUTPCCompressionGatherKernels::gatherBuffered(int nBlocks, int nT
703703
compressorMemcpyBuffered(buf, output->timeResA + tracksOffsetDiff, input.timeResA, nTrackClustersPtr, aClsFstIdx, tracksPerWarp, nLanes, iLane, 1);
704704
}
705705

706-
GPUdii() void GPUTPCCompressionGatherKernels::gatherMulti(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& GPUrestrict() smem, processorType& GPUrestrict() processors)
706+
GPUdii() void GPUTPCCompressionGatherKernels::gatherMulti(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& smem, processorType& GPUrestrict() processors)
707707
{
708708
GPUTPCCompression& GPUrestrict() compressor = processors.tpcCompressor;
709709
const o2::tpc::ClusterNativeAccess* GPUrestrict() clusters = processors.ioPtrs.clustersNative;
@@ -783,25 +783,25 @@ GPUdii() void GPUTPCCompressionGatherKernels::gatherMulti(int nBlocks, int nThre
783783
}
784784

785785
template <>
786-
GPUdii() void GPUTPCCompressionGatherKernels::Thread<GPUTPCCompressionGatherKernels::buffered32>(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& GPUrestrict() smem, processorType& GPUrestrict() processors)
786+
GPUdii() void GPUTPCCompressionGatherKernels::Thread<GPUTPCCompressionGatherKernels::buffered32>(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& smem, processorType& GPUrestrict() processors)
787787
{
788788
gatherBuffered<Vec32>(nBlocks, nThreads, iBlock, iThread, smem, processors);
789789
}
790790

791791
template <>
792-
GPUdii() void GPUTPCCompressionGatherKernels::Thread<GPUTPCCompressionGatherKernels::buffered64>(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& GPUrestrict() smem, processorType& GPUrestrict() processors)
792+
GPUdii() void GPUTPCCompressionGatherKernels::Thread<GPUTPCCompressionGatherKernels::buffered64>(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& smem, processorType& GPUrestrict() processors)
793793
{
794794
gatherBuffered<Vec64>(nBlocks, nThreads, iBlock, iThread, smem, processors);
795795
}
796796

797797
template <>
798-
GPUdii() void GPUTPCCompressionGatherKernels::Thread<GPUTPCCompressionGatherKernels::buffered128>(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& GPUrestrict() smem, processorType& GPUrestrict() processors)
798+
GPUdii() void GPUTPCCompressionGatherKernels::Thread<GPUTPCCompressionGatherKernels::buffered128>(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& smem, processorType& GPUrestrict() processors)
799799
{
800800
gatherBuffered<Vec128>(nBlocks, nThreads, iBlock, iThread, smem, processors);
801801
}
802802

803803
template <>
804-
GPUdii() void GPUTPCCompressionGatherKernels::Thread<GPUTPCCompressionGatherKernels::multiBlock>(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& GPUrestrict() smem, processorType& GPUrestrict() processors)
804+
GPUdii() void GPUTPCCompressionGatherKernels::Thread<GPUTPCCompressionGatherKernels::multiBlock>(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& smem, processorType& GPUrestrict() processors)
805805
{
806806
gatherMulti(nBlocks, nThreads, iBlock, iThread, smem, processors);
807807
}

‎GPU/GPUTracking/DataCompression/GPUTPCCompressionKernels.h‎

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -40,7 +40,7 @@ class GPUTPCCompressionKernels : public GPUKernelTemplate
4040
};
4141

4242
template <int iKernel = defaultKernel>
43-
GPUd() static void Thread(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& GPUrestrict() smem, processorType& GPUrestrict() processors);
43+
GPUd() static void Thread(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& smem, processorType& GPUrestrict() processors);
4444

4545
template <int I>
4646
class GPUTPCCompressionKernels_Compare
@@ -97,7 +97,7 @@ class GPUTPCCompressionGatherKernels : public GPUKernelTemplate
9797
};
9898

9999
template <int iKernel = defaultKernel>
100-
GPUd() static void Thread(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& GPUrestrict() smem, processorType& GPUrestrict() processors);
100+
GPUd() static void Thread(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& smem, processorType& GPUrestrict() processors);
101101

102102
template <typename T, typename S>
103103
GPUdi() static bool isAlignedTo(const S* ptr);
@@ -118,9 +118,9 @@ class GPUTPCCompressionGatherKernels : public GPUKernelTemplate
118118
GPUdi() static unsigned int calculateWarpOffsets(GPUSharedMemory& smem, T* nums, unsigned int start, unsigned int end, int nWarps, int iWarp, int nLanes, int iLane);
119119

120120
template <typename V>
121-
GPUdii() static void gatherBuffered(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& GPUrestrict() smem, processorType& GPUrestrict() processors);
121+
GPUdii() static void gatherBuffered(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& smem, processorType& GPUrestrict() processors);
122122

123-
GPUdii() static void gatherMulti(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& GPUrestrict() smem, processorType& GPUrestrict() processors);
123+
GPUdii() static void gatherMulti(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& smem, processorType& GPUrestrict() processors);
124124
};
125125

126126
} // namespace GPUCA_NAMESPACE::gpu

‎GPU/GPUTracking/Merger/GPUTPCGMO2Output.cxx‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -208,7 +208,7 @@ GPUdii() void GPUTPCGMO2Output::Thread<GPUTPCGMO2Output::output>(int nBlocks, in
208208
}
209209

210210
template <>
211-
GPUdii() void GPUTPCGMO2Output::Thread<GPUTPCGMO2Output::mc>(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& GPUrestrict() smem, processorType& GPUrestrict() merger)
211+
GPUdii() void GPUTPCGMO2Output::Thread<GPUTPCGMO2Output::mc>(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& smem, processorType& GPUrestrict() merger)
212212
{
213213
#ifndef GPUCA_GPUCODE
214214
const o2::tpc::ClusterNativeAccess* GPUrestrict() clusters = merger.GetConstantMem()->ioPtrs.clustersNative;

‎GPU/GPUTracking/Refit/GPUTrackingRefitKernel.h‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -31,7 +31,7 @@ class GPUTrackingRefitKernel : public GPUKernelTemplate
3131
};
3232

3333
template <int iKernel = defaultKernel>
34-
GPUd() static void Thread(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& GPUrestrict() smem, processorType& GPUrestrict() processors);
34+
GPUd() static void Thread(int nBlocks, int nThreads, int iBlock, int iThread, GPUsharedref() GPUSharedMemory& smem, processorType& GPUrestrict() processors);
3535
};
3636

3737
} // namespace o2::gpu

‎GPU/GPUTracking/SliceTracker/GPUTPCGlobalTracking.h‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -45,8 +45,8 @@ class GPUTPCGlobalTracking : public GPUKernelTemplate
4545
GPUd() static void GlobalTrackingSliceLeftRight(unsigned int iSlice, unsigned int& left, unsigned int& right);
4646

4747
private:
48-
GPUd() static int PerformGlobalTrackingRun(GPUTPCTracker& tracker, GPUsharedref() MEM_LOCAL(GPUSharedMemory) & GPUrestrict() smem, const GPUTPCTracker& sliceSource, int iTrack, int rowIndex, float angle, int direction);
49-
GPUd() static void PerformGlobalTracking(int nBlocks, int nThreads, int iBlock, int iThread, const GPUTPCTracker& tracker, GPUsharedref() MEM_LOCAL(GPUSharedMemory) & GPUrestrict() smem, GPUTPCTracker& sliceTarget, bool right);
48+
GPUd() static int PerformGlobalTrackingRun(GPUTPCTracker& tracker, GPUsharedref() MEM_LOCAL(GPUSharedMemory) & smem, const GPUTPCTracker& sliceSource, int iTrack, int rowIndex, float angle, int direction);
49+
GPUd() static void PerformGlobalTracking(int nBlocks, int nThreads, int iBlock, int iThread, const GPUTPCTracker& tracker, GPUsharedref() MEM_LOCAL(GPUSharedMemory) & smem, GPUTPCTracker& sliceTarget, bool right);
5050
};
5151
#endif
5252

0 commit comments

Comments
 (0)